See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/385904045

Mathematical and Data Literacy

Research · November 2024

DOI: 10.13140/RG.2.2.20462.63049

CITATIONS READS

0 13

5 authors, including:

Cathy Smith
The Open University (UK)
14 PUBLICATIONS 91 CITATIONS

SEE PROFILE

All content following this page was uploaded by Cathy Smith on 18 November 2024.

The user has requested enhancement of the downloaded file.

Mathematical and
Data Literacy
Competencies and curriculum
implications at the intersection of
mathematics, data science,
statistics and computing
Cathy Smith, Vinay Kathotia, Robert Ward-Penny, Oli
Howson and Michel Wermelinger.
April 2023
§
Contents
Contents 2

1. Introduction 5

2. Summary 5

Part One 10

3. Competencies 10

4. Approach 12

4.1 Reference points 13

4.2 Examining the Intersections 16

4.3 Terminology 22

5. Competencies at the intersection 23

Six intersectional competencies 25

5.1 Positioning the competencies amongst other mathematics

and statistics/data science frameworks 26

5.2 Unpacking the mathematics toolkit 33

5.3 Unpacking the data skills toolkit 35

5.4 Unpacking the computing toolkit 36

6. Citizen categories and school leavers 37

6.1 Intersectional competencies for school leavers 40

6.2 Mathematical toolkit competencies for school leavers 42

April 2023 pg. 2

 6.3 Statistics/data science toolkit competencies for school
leavers 43

6.4 Computational toolkit competencies for school leavers 44

Part Two 46

7. Curriculum models 46

8. Case Studies 48

8.1 Ontario curriculum pathways 48

8.2 ‘Mathematics and Statistics’ in Aotearoa New Zealand 53

8.3 ‘Analysis’ and ‘Applications’ pathways in the International

Baccalaureate 58

8.4 ProDaBi – Project Data Science and Big Data in German

schools 63

8.5 Scotland’s post-15 National Progression Awards in Data

Science 68

8.6 MEI supplementary Data Science Courses 71

8.7 Geography and Core Mathematics Project 75

8.8 Cross-curricular work in Civic Statistics 78

8.9 Bootstrap integrating computing with other subjects 82

8.10 Urban Data School 86

9. Overview of Case Studies 89

9.1 The case for change 90

April 2023 pg. 3

 9.2 Promoting data skills through computing and across
subjects 91

9.3 Introducing strands within mathematics 95

9.4 Introducing pathways within mathematics 97

9.5 Pitfalls and possibilities of mathematical and data literacy

as a separate curriculum area 100

9.6 The challenge and potential of assessment 102

9.8 The important role of integrative, focused and ongoing

professional development 103

9.8 Feasibility overview of models 104

9.9 The value of a joined-up approach 105

10. References 106

Appendix A Expert interviews 112

Appendix B Royal Society’s indicative curriculum models 113

April 2023 pg. 4

 1. Introduction
This report contributes to the Mathematical Futures Programme aim of building a new
vision of mathematics education that anticipates and supports the role of mathematics for
individuals, economies and society, strengthening diversity and reducing inequity. Previous
work on this project had identified three areas with the most potential and need: (1) the
integration of data science into the mathematics curriculum, (2) opportunities for
enhancing programming and computational thinking in mathematics, and (3) a coherent
and cross phase approach to the use of digital technology in mathematics.

The Royal Society has commissioned this report to investigate the intersection of
mathematics, data science, statistics and computing, thereby addressing areas (1) and, to
some extent, (2) above. A combination of desk research and expert interviews was used to
synthesise sector-specific documents that set out visions for competencies needed by
future school leavers and reported curriculum initiatives relevant to those goals. This was
used to create a body of knowledge at the intersection of mathematics, data science,
statistics and computing, listing top-level competencies that young adults should have
when they leave school or college, and to consider the implications for various models of
curriculum change.

2. Summary
Part One of this report establishes a framework of competencies for activities at the
intersection of mathematics, statistics, data science and computing. It provisionally terms
this area of activity ‘mathematical and data literacy’. It was based on a survey of
curriculum documents and research papers, supplemented by expert interviews, within the
four disciplines of mathematics, statistics, data science and computing education.

Two principles have guided the identification of mathematical and data literacy
competencies:

 Competencies included in the intersection should, as far as possible, align with

consensus views in the literature, increasing the potential for adoption.
 The overall set of competencies should be minimal and coherent, providing a basis
for future curriculum development work.

April 2023 pg. 5

Recommendations for curriculum mapping

 For the purposes of future-facing school curricula, statistics should be

disaggregated from mathematics and, instead, data science and statistics can be
treated as cognate. Data science, with its contextual relevance and ethical aspects,
provides the opportunity for an engaging, data-driven education that includes all
the statistics needed by school leavers.

 In defining the intersections of mathematics, statistics/data science and computing,

a helpful focus is on mathematics competencies most closely associated with data-
driven questions, whether from pure or applied mathematics, rather than on finding
connections to all of mathematics.

 While fluent use of technology is clearly related to computing (which includes

programming), their changing relationship and different take-up in schools makes it
helpful to consider the two separately.

 Knowledge and decision-making within a context, and with relevant tools, are so
important to statistics/data science that they cannot be omitted from the
intersection. We conclude that the intersection of mathematics, statistics/data
science and computing is concerned with the study of variation and generalisability,
with precision and context and technology and ethics.

Intersectional Competencies

Figure 1 Model for intersectional competencies

The work on consensus areas produced a framework shown in Figure 1 that aligns
mathematics, statistics/data science and computing around the overarching activity of

April 2023 pg. 6

posing and answering questions. This is based on six high-level competencies common
across the reference literature. They are related but different, and each is necessary for the
resulting activity to be considered as characteristic of the intersection of mathematics,
statistics data science and computing. See 5.1 for details.

Within this framework, problem solving can draw on ‘toolkits’ in mathematics, computing
and statistics/ data science, not just a collection of techniques, but the ways of thinking and
reasoning, and the language, concepts and tools inherent to those disciplines.

 The mathematical toolkit involves: Using quantities and methods; Analysing

covariation; Reasoning mathematically; Using mathematical representations; Using
mathematical aids and tools.
 The statistics/data science toolkit involves: Data stewardship; Handling data; Data
representation; Statistical thinking and methods; Probabilistic reasoning; Using
computational aids and tools.
 The computing toolkit involves: Computational thinking (including developing
algorithms; Using algorithms; Programming; Representing and manipulating data;
Safe use of technology; Using a range of aids and tools including emerging
technologies.

For detailed competency descriptions, see 5.2 – 5.4.

Competencies for school leavers

A repeated call in workplaces is for ‘T-shaped’ knowledge - broad across many areas with
deep knowledge in one area. The six high-level mathematical and data literacy
competencies, and their toolkits, can be further elaborated as competencies for school
leavers. The competencies elaborating skills ‘for all’ and ‘for many’ provide the broad
knowledge and activities that allow students to engage fully and critically in society, and to
move into entry-level roles that involve decision-making based on quantitative data. Skills
‘for specialists’ allow students to develop depth at school. Together they could provide the
basis of a new course of study, or show how students studying e.g. computing, could use
this specialist knowledge to extend their mathematical and data literacy.

See Section 6 for competencies for all, for many and for specialists.

The feasibility of curriculum models for embedding mathematical and data literacy

Part Two reports an analysis of curriculum models for achieving the competency goals set
out in Section 6. Ten case studies were identified that are compatible with these aims (see
Section 8, which may be treated as an appendix). These were compared with nine

April 2023 pg. 7

curriculum models, ranging from ‘No change’, through a series of changes and additions to
the current curriculum and assessment at KS 3, KS 4 and KS 5, to ‘A new baccalaureate
system’.

Evidence from the case studies suggests:

 There are few precedents for teaching interdisciplinary content within computing
and small numbers studying at GCSE reduce its reach.

 The opposite holds: there are several successful trials of teaching computing
through other subjects, including some of developing teachers in other subjects to
teach mathematical and statistical skills. Gains are noted mainly in social science
and science subjects, with little evidence for arts, reducing the reach in a free choice
system.

 Curricula that have attempted to raise the profile of what was previously considered
a subset of mathematics (for example promoting statistics in New Zealand or
modelling in Ontario) do so both by creating separate strands that carry teaching
and assessment requirements and by providing trialled teaching materials for
teachers to adapt.

 Curricula such as the International Baccalaureate, New Zealand and Ontario that
provide multiple pathways for 16-18 study have good take up of pathways related to
mathematical and data literacy as well as those leading to STEM careers. All these
systems have a minimum requirement for studying mathematics at KS 5.

 Projects and curricula that have successfully embedded aspects of mathematical

and data literacy have all included substantive project work. This may be non-
assessed, assessed by teachers, also externally moderated or even externally
marked. Extra teacher workload associated with assessment is noted, but also
recognition of its value.

 Case studies speak to the potential for teaching 11-14 year olds to pose and answer
questions based on data relevant to them. They also emphasise the need for
teaching. Stand-alone courses that include a data skills project would need
curriculum time and teacher support.

 Integrative, focused and ongoing professional development is a common factor in

making effective change. Similarly; an approach involving multiple stakeholders and
organisations supports curriculum design and delivery.

April 2023 pg. 8

 Three features of current assessment structures recur as barriers for creating
mathematical and data literacy courses for all:

o the complicated relationship between English teachers, policy makers,

inspectors and multiple exam boards;

o the stifling of any inter-disciplinary teaching activity by high-stakes GCSE

assessment at 16;

o unconstrained subject choice at 14 and 16 creates subject silos and prevents

coordinated approaches to planning interdisciplinary study.

April 2023 pg. 9

Part One
3. Competencies
For this report we take competencies to be enactments of activities appropriate within the
emerging domain of mathematical and data literacy. Taken as a whole, a mathematics
competency framework responds to the question ‘What does it take for a learner to
become a doer or user of mathematics?’, rather than the question ‘What does it take to
become a knower?’ 1. Individual competencies add specificity to a general notion of
mathematical competence. Thus: “a mathematical competency is someone’s insightful
readiness to act appropriately in response to a specific sort of mathematical challenge in
given situations” 2. While there is widespread agreement that knowing and doing are
interdependent 3, a curricular focus on competencies emphasises a sociocultural
understanding that a main purpose of education is for students to play an active role in
society.

Interest in mathematics competencies developed in the early 2000s, stimulated by

curriculum reform in the US (National Research Council, 2001) and the growth of
comparative curriculum/performance studies such as TIMSS 1995 4. One reason given for
working with competencies is that they allow a focus on processes independently of

1
Mogens Niss et al., ‘Conceptualisation of the Role of Competencies, Knowing and
Knowledge in Mathematics Education Research’, in Proceedings of the 13th International
Congress on Mathematical Education, ed. Gabriele Kaiser, ICME-13 Monographs (Cham:
Springer International Publishing, 2017), 235–48, https://doi.org/10.1007/978-3-319-62597-
3_15.
2
Mogens Niss and Tomas Højgaard, ‘Mathematical Competencies Revisited’, Educational
Studies in Mathematics 102, no. 1 (1 September 2019): 14, https://doi.org/10.1007/s10649-019-
09903-9.
3
A. Schoenfeld, ‘Learning to Think Mathematically: Problem Solving, Metacognition and
Sense Making in Mathematics’, in Handbook of Research on Mathematics Teaching and
Learning, ed. D. Grouws (New York: Macmillan, 1992), 335–70; Bethany Rittle-Johnson,
Michael Schneider, and Jon R. Star, ‘Not a One-Way Street: Bidirectional Relations Between
Procedural and Conceptual Knowledge of Mathematics’, Educational Psychology Review 27,
no. 4 (December 2015): 587–97, https://doi.org/10.1007/s10648-015-9302-x.
4
Jeremy Kilpatrick, ‘Competency Frameworks in Mathematics Education’, in Encyclopedia of
Mathematics Education, ed. Stephen Lerman (Cham: Springer International Publishing, 2020),
110–13, https://doi.org/10.1007/978-3-030-15789-0_27.

April 2023 pg. 10

subject content areas. This facilitates tracing similar mathematical activities at different
levels of schooling, where content statements may appear very different, and thus supports
coherent curriculum progressions. It also facilitates comparisons across the curricula of
different educational systems. In curricula where content areas and competencies are
distinguished, their relationship may be represented via dimensions in an array (Germany,
PISA) or facets of a polygon (e.g. the well-known Singapore pentagon).

Between them, content statements and cognitive competencies still only describe part of
what is intended as educational outcomes. Some frameworks also include affective
statements such as resilience and confidence, and attitudinal statements such as
understanding that data shapes our world. This wider view responds to the international
movement around 21st century skills for lifelong learning 5.

Competencies are used with many purposes. They are used normatively to design the
aims, structure, organisation and assessment of curricula; descriptively by teachers and
researchers to characterize mathematics teaching and learning; diagnostically and
metacognitively in interactions with students 2. A useful distinction is offered between
internal and external use 6. Competencies for internal use support communication between
practitioners within and about a discipline. Mathematics competencies for external use
function as a communication with parents, government, society; they position and justify
the role of mathematics in education. Both types of communication are necessary but they
may require different emphases. As a relevant example, for pure mathematicians,
‘mathematical modelling’ may appear as just one of several competencies that
characterise mathematics but it is central to applied mathematics and usually
emphasised in external communications about why everyone should study mathematics.

Finally, competencies are written at different granularities, ranging from titular noun
phrases (e.g. China’s core competencies of intuitive imagination, data analysis 7) to
statements of actions with specific qualities, contexts and tools. The elaboration of high-

5
Malcolm Tight, ‘Twenty-First Century Skills: Meaning, Usage and Value’, European Journal
of Higher Education 11, no. 2 (3 April 2021): 160–74,
https://doi.org/10.1080/21568235.2020.1835517.
6
Jens Hojgaard Jensen and Uffe Thomas Jankvist, ‘Disciplinary Competence Descriptions
for External Use’, Nordic Studies in Mathematics Education 23, no. 2 (2018): 3–24.
7
Jing Cheng, Jiansheng Bao, and Dianzhou Zhang, ‘From “Two Basics”, to “Four Basics” to
“Core Mathematics Competencies” in Mainland China’, in Beyond Shanghai and PISA:
Cognitive and Non-Cognitive Competencies of Chinese Students in Mathematics, ed.
Binyan Xu, Yan Zhu, and Xiaoli Lu, Research in Mathematics Education (Cham: Springer
International Publishing, 2021), 1–13, https://doi.org/10.1007/978-3-030-68157-9_1.

April 2023 pg. 11

level competencies into, for example, ‘specific expectations’ with teaching examples, tips,
prompts, and sample tasks 8 is a necessary part of curriculum development and – crucially
- continues for years as teachers, task designers and quality controllers negotiate what is
considered educational achievement in teaching and assessment.

4. Approach
The purpose of Part One of this report is to survey existing vision literature in mathematics,
data science, statistics and computing (MDSC) in order to create a body of knowledge
listing the top-level intersectional data/ statistical/ problem-solving/ computing
competencies that young adults should have when they leave school or college to prepare
them for employment and life.

The evidence base for this synthesis was provided by a literature review and a series of
interviews with experts associated with curriculum initiatives in MDSC (see Appendix A),
each strand establishing credentials and high-level interpretations for the other.

Sources were purposively sampled based on a combination of:

 Recommendations from experts in MDSC education of papers considered to be

significant scholarship that aligns with consensus views.

 Sources derived from research databases, using the Academic Search Complete,
British Education Index, Education Abstracts, Education Research Complete, and ERIC
databases, for the period 01/01/2018 to 02/02/2023, applying the search terms:

i. "mathematics education" AND "computational thinking" AND curriculum (29

results, which led to 5 relevant papers)
ii. "mathematics education" AND (computing or "computer science") AND
(statistics or "data science”) (85 results, which led to 4 further relevant
papers)

 Following up references from key papers and authors.

The question that originally framed our reading was broad: “What does the existing vision
literature in MDSC have to say about competencies for citizens and the proposed school
curriculum models?” This developed into:

8
Ontario Ministry of Education, The Ontario Curriculum, Grade 9: Mathematics (2021), 2021,
https://www.dcp.edu.gov.on.ca/en/curriculum/secondary-mathematics/courses/mth1w.

April 2023 pg. 12

  What does the existing vision literature in MDSC indicate are areas of consensus
within and between the relevant disciplines?

 What are examples of stable, well-documented and high-performing curricula or

curriculum initiatives considered to address these intersections of MDSC?

 What are the implications for competencies for citizens and the proposed school
curriculum models? (cf. Part Two)

4.1 Reference points

Expert interviews and an initial review of the literature enabled us to identify examples of
stable, well-documented and high-performing curricula or curriculum initiatives. We used
these as reference points, comparing them within each discipline to establish what was
agreed to be both in scope and possible (a sense of central tendency and spread,
perhaps!). We compared them also between disciplines, now focusing on what was
common and what was compatible.

In Mathematics the three reference frameworks chosen are: the Danish KOM competency
framework (2003), the Ontario Curriculum (2007, revised 2020), the OECD PISA
mathematical literacy framework (2003, revised 2012, 2021). The Danish competencies of
mathematical learning (KOM) project was chosen because (together with the US NCTM
Standards) it is regarded as one of the first, most influential, competency frameworks. It is
stable, with subsequent educational initiatives (to reduce inequality, for example) focusing
on whole-school enactment and upskilling teachers 9. KOM’s attention is squarely on the
discipline of mathematics. In contrast, the 2005 Ontario curriculum exemplifies curricula
organised around mathematics as a tool for everyday life 10, corresponding to an aim of
Mathematical Futures. Ontario’s 2020-21 reforms have since incorporated programming
into elementary mathematics. Canadian students out-perform English students on PISA
tests, and Ontario is close to the Canadian average 11. In particular, they are strong on

9
OECD, Education Policy Outlook 2019: Working Together to Help Students Achieve Their
Potential, Education Policy Outlook (OECD, 2019), https://doi.org/10.1787/2b8ad56e-en.
10
Cathy Smith and Candia Morgan, ‘Curricular Orientations to Real-World Contexts in
Mathematics’, The Curriculum Journal 27, no. 1 (2 January 2016): 24–45,
https://doi.org/10.1080/09585176.2016.1139498.
11
Kathryn O’Grady et al., ‘Measuring up: Canadian Results of the OECD PISA 2018 Study’ (Council
of Ministers of Education, Canada, 2019).

April 2023 pg. 13

interpreting mathematics in context 12. Finally, the racy definitions of mathematical literacy
are internationally influential and mobilised for large-scale computer-based assessment.

In Statistics, reference frameworks are: the New Zealand Mathematics and Statistics school
curriculum 13 (2007, with updates in subsequent years, and being revised 2023), and the
GAISE (2007, updated 2020) framework 14. New Zealand is well regarded internationally for
its statistics education 15. It gives statistics equal footing alongside mathematics with its
three inter-related strands of Statistical investigation, Statistical literacy and Probability
being developed from the start of primary education through to Year 13. A high proportion
of students continue to study statistics beyond compulsory schooling: the New Zealand
take-up rate for advanced mathematics or statistics is three times that of England 16. The
Pre-K–12 Guidelines for Assessment and Instruction in Statistics Education II (GAISE II)
framework is an internationally developed framework that bridges statistics and data
science, that officially underpins the statistics education standards of the US National
Council of Teachers of Mathematics (NCTM) and is endorsed by the American Statistical
Association. GAISE II has recommendations and exemplar teaching and assessment
activities that support the development of statistical reasoning and statistical literacy
across three levels.

In curricular terms, data science is the newest arrival on the block (post-2010). Moreover,
data science has not been an academic innovation but is an instance where the education
sector is playing catch up with a rapidly expanding industry and responding to growing
demand/clamour for data science courses at tertiary level 17, and recognising its value for

12
Pierre Brochu et al., Measuring up, Canadian Results of the OECD PISA Study: The Performance
of Canada’s Youth in Mathematics, Reading and Science: 2012, First Results for Canadians Aged
15 (Toronto: CMEC, 2014), http://site.ebrary.com/lib/celtitles/docDetail.action?docID=10831057.
13
https://nzcurriculum.tki.org.nz/The-New-Zealand-Curriculum/Mathematics-and-statistics OR
https://new.censusatschool.org.nz/new-zealand-statistics-curriculum/
14
A. Bargagliotti, C. Franklin, P. Arnold, R. Gould, S. Johnson, L. Perez, and D. Spangler, Pre-K-12
Guidelines for assessment and instruction in statistics education (GAISE) report II, American
Statistical Association, Alexandria, VA, 2020. www.amstat.org/asa/files/pdfs/GAISE/GAISEIIPreK-
12_Full.pdf
15
Sharleen Forbes, ‘The Coming of Age of Statistics Education in New Zealand, and Its Influence
Internationally’, Journal of Statistics Education 22, no. 2 (July 2014): 2,
https://doi.org/10.1080/10691898.2014.11889699.
16
Jim Ridgway, ‘Implications of the Data Revolution for Statistics Education: The Data Revolution
and Statistics Education’, International Statistical Review 84, no. 3 (December 2016): 528–49,
https://doi.org/10.1111/insr.12110.
17
David Donoho, ‘50 Years of Data Science’, Journal of Computational and Graphical Statistics
26, no. 4 (2 October 2017): 745–66, https://doi.org/10.1080/10618600.2017.1384734.

April 2023 pg. 14

school education 16. School-based courses are less developed 18 and have greater
challenges. Given the industry-driven emergence of data science, our reference points for
data science started with the Standards of the UK’s Association for Data Science
Professionals 19, contrasting them with GAISE II and three research-informed data science
frameworks developed for use in schools.

The International Data Science in Schools Project (IDSSP) was set up in 2018 by statisticians
and computer scientists from across the world. Envisaged school-based trials were
disrupted by the Covid-19 pandemic but the IDSSP frameworks 20 (for curriculum and
teacher professional development) have been widely endorsed and influential.

The UCLA and Los Angeles Unified School District’s Introduction to Data Science (IDS)21 is a
data-driven course for high school that brings together statistical and computational
thinking, civic engagement and mobile technology to support ‘students to think critically
about and with data’ 22. First implemented in 2014, the course has engaged over 42,000
students in a total of 151 high schools across 74 districts in the USA.

Project Data Science and Big Data at School (ProDaBi) 23, developed by researchers at
Paderborn University, Germany has been developing and trialling courses since 2018. The
courses span all of secondary school (Grades 5 through 12 in Germany) and the project
includes development of teaching materials, professional development courses for
teachers and related evaluation and research.

Finally, for computing, the reference frameworks are the 2013 national curriculum in
England 24, the GCSE, A level and league-table approved computing qualifications in
England and Wales, the US Common Core Standards for Computer Science 25 (and the US
Advanced Placement Program (AP) Computer Science courses 26), and the Singapore

18
Neville Davies and Neil Sheldon, ‘Teaching Statistics and Data Science in England’s Schools’,
Teaching Statistics 43 (2 June 2021): S52–70, https://doi.org/10.1111/test.12276.
19
AfDSP, ‘Data Science Standards’ (Alliance for Data Science Professionals, 2022),
https://alliancefordatascienceprofessionals.co.uk//standards.
20
http://www.idssp.org/files/IDSSP_Frameworks_1.0.pdf
21
https://www.idsucla.org/
22
https://www.idsucla.org/other-mobilize-curricula
23
https://www.prodabi.de/en/
24
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_d
ata/file/239067/SECONDARY_national_curriculum_-_Computing.pdf
25
https://www.codelicious.com/k-12-computer-science-standards
26
https://apcentral.collegeboard.org/courses/ap-computer-science-a and
https://apcentral.collegeboard.org/courses/ap-computer-science-principles/course

April 2023 pg. 15

curriculum based around an O level and A level in Computing. As we considered digital
technology skills alongside computing, we also looked at Dr Erika Kispeter’s Digital Skills and
Inclusion Research Working Group Evidence Brief 27 which looked at ‘What digital skills do
adults need to succeed in the workplace now and in the next 10 years?’

4.2 Examining the Intersections

The Royal Society project brief has the title “Intersection of mathematics and computing,
statistics, and data science”. Given the different granularities at which one can list
competencies, and the variety of subject definitions in play, the intersection could range
from the entirety of these subject areas to the empty set. This is not simply a semantic issue
or a territorial dispute but is crucial for how we delineate the subject matter for this report.
Our survey of the literature revealed three areas on which opinions can clearly differ but on
which we needed to take a position. In this section we examine each in turn:

1. The relationship between statistics and data science.

2. The curriculum balance of pure and applied mathematics.

3. The role of computing and programming in mathematical and data literacy.

4.2.1 The relationship between statistics and data science

Analysing and making sense of data, and consequently data literacy, are central elements
of both data science and statistics. There are contrasting views as to whether data science
is the same as statistics.

Tukey (1962) 29 and Owen (2015) 28, would say that the distinction is fundamental, cultural
and much needed. Owen draws attention to the extra importance in data science of
software engineering, with teams bringing together diverse skills working within live data, at
scale and in real time. While the rise of computational power, the opening up of vast fields
of data and proliferation of data types are at the heart of the emergence of data science,
that does not encompass the whole story. Tukey argued that data analysis should be
viewed as a new science rather than an application of mathematical statistics: data
science “must seek for scope and usefulness rather than security” and “must use

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_d
27

ata/file/807831/What_digital_skills_do_adults_need_to_succeed_in_the_workplace_now_a
nd_in_the_next_10_years_.pdf
28
https://medium.com/@srowen/what-50-years-of-data-science-leaves-out-2366c9b61d3d

April 2023 pg. 16

mathematical argument and mathematical results as bases for judgment rather than as
bases for proof or stamps of validity” 29.

From this perspective, mathematics and statistics are implicated in data science
(providing the representations and tools for data manipulation and analysis) but equally
important are domain-specific knowledge, computing and the use of technology, and
aspects of the data such as its nature, size, provenance, accuracy, reliability, ethical use,
and the needs of the industry/clients posing the problems.

Such arguments for the distinctiveness of data science tend to treat statistics as part of
mathematics. This perspective is common in school: in many curricula, including the
English one, probability and statistics are strands of mathematical content knowledge.
Similarly, the OECD’s PISA 2021 framework states that mathematical reasoning includes
“both mathematical (deductive) and statistical (inductive) type reasoning”. 30

Nevertheless, this narrower view of school statistics as a subset of mathematics does

not represent how it has been defined by later statisticians, or in Higher Education, as “the
only discipline whose primary activity is finding meaning in data”. 31 This wider view does
include those valued elements of data provenance, accuracy, usage, etc that a purely
mathematical lens might abstract away. Moreover, the omission of these in the enacted
school curriculum is identified as a barrier to good teaching of statistics. 32 Thus, for the
purposes of this future-facing paper on school curricula, we have preferred to treat data
science and statistics as cognate. At school level, what the disciplines have in common far
outweighs the differences that exist in approach and content. Those divergences which are
attended to by school-level study are often in the finer detail met only by specialists; for
example, hypothesis testing based on a probability appears in the 2017 A level Mathematics

29
John W. Tukey, ‘The Future of Data Analysis’, The Annals of Mathematical Statistics 33, no. 1
(1962): 6.
30
OECD, ‘PISA 2021 Mathematics Framework (Second Draft)’, 2018, 3,
https://www.oecd.org/pisa/sitedocument/PISA-2021-mathematics-framework.pdf.
31
Robert Gould, ‘Toward Data-Scientific Thinking’, Teaching Statistics 43, no. S1 (July 2021),
https://doi.org/10.1111/test.12267.
32
ACME and the Royal Statistical Society (RSS), ‘Embedding Statistics at A Level: A Report on
Statistical Requirements and Assessment across A Level Courses in Biology, Business, Chemistry,
Geography, Psychology and Sociology’, 2015, http://www.acme-
uk.org/media/32719/embeddingstatistics-at-a-level.pd; Neville Davies and Neil Sheldon,
‘Teaching Statistics and Data Science in England’s Schools’, Teaching Statistics 43, no. S1 (2021):
S52–70, https://doi.org/10.1111/test.12276.

April 2023 pg. 17

curriculum, whereas a data science approach might focus instead on using data to
determine prediction intervals.

This approach of considering statistics/data science together is supported by the literature

about innovations in statistics and data science education, which provide similar and
certainly compatible messages 33. Data science aligns with statistics in its central concern
with gathering, processing and analysing data to generate insights and predictions in
complex situations. In the reference curricula for data science this translates to a central
role for the statistical investigation cycle, and the use of statistical representations and
tools for analysis, communication and evaluation. Much of this may also be automated via
technology, resulting in a greater overlap between data science and computing than was
the case with previous school statistics but entirely compatible with its actual use beyond
school. The consensus amongst school-based statistics/data science educators is that
data science, with its contextual relevance and ethical aspects, provides the opportunity
for an engaging, data-driven education that includes all the statistics needed by school
leavers.

4.2.2 The curriculum balance of pure and applied mathematics

The importance of extra-mathematical contexts in statistics/data science raises questions

about how such contexts feature in school mathematics curricula, and whether there are
opportunities for data-driven mathematics outside what is normally deemed to be
‘statistics’ or ‘data handling’ content. A curriculum that aims to promote the applications of
statistics via data science should pay equal attention to applications of mathematics.

The current mathematics specification of the English national curriculum is largely context-
free. Subject content is listed with the overarching statement that, through the
mathematics content, pupils should be taught to develop fluency, reason mathematically
and solve problems. The value of applied mathematics is recognised, since some problems
must be set in financial contexts and others may involve “model[ling] situations
mathematically 34. However there is no explicit statement of the relationship between pure
and applied mathematics or of the range of situations in which mathematics should/could

33
As a sign of merging messages, in 2021, the American Statistical Association’s Journal of
Statistics Education evolved to become the Journal of Statistics and Data Science Education.
34
Department for Education, Mathematics Programmes of Study: Key Stage 4. National
Curriculum in England, 2014,
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_da
ta/file/331882/KS4_maths_PoS_FINAL_170714.pdf.

April 2023 pg. 18

be applied. The lack of an explicit statement has implications for what is assessed.
Comparison with the reference curricula shows that other choices can be made.

Out of the three reference curricula for mathematics, the high-level Danish KoM framework
chooses to distinguish problems arising within mathematics itself and those which answer
questions relating to an external, often real-world domain, naming these ‘mathematical
problem-handling’ and ‘mathematical modelling’ (see Figure 1). The contrast and balance
between them has been retained and clarified over time 35 suggesting that the explicit
distinction is deemed valuable. Problem-handling includes devising and implementing
strategies for problems that are already formulated mathematically but are non-routine to
the problem solver. Modelling has at its heart mathematical actions that take account of
the purposes, data, facts, features and properties of an extra-mathematical domain. It also
gains extra emphasis as a way for students to understand that an essential feature of
mathematics as a discipline is how it is applied within other fields of practice.

Figure 2 A visual representation of the eight KOM mathematical competencies 35

In contrast, the Ontario curriculum “balances” fundamental concepts and skills in

mathematics with opportunities to apply them and requires students at all grades “to make
connections between their lived experiences, mathematical concepts, other subject areas,
and situations outside of school.” (The Ontario Curriculum, Grades 1-8: Mathematics, 2020,
p65). Grade 11 course titles such as Mathematics for Work and Everyday Life, or Functions
and Applications are a good representation of this more integrated approach. Questions

35
Niss and Højgaard, ‘Mathematical Competencies Revisited’.

April 2023 pg. 19

and data pertaining to real-world situations are pervasive as a purpose and a motivation
for learning mathematics and, largely, also for developing new mathematics. In this
curriculum, data largely serves the purpose of illustrating applications of mathematics. A
less realistic example is using Pythagoras’s theorem to calculate ramp heights (where
practitioners would use a table, an app, or CAD software); a more realistic one is using
exponential graphs to predict growth.

Similarly, the OECD’s PISA framework has moved towards integrating pure and applied
mathematics. The Formulate-Employ-Interpret modelling cycle has been central to its
definition of mathematical literacy since 2003. In 2022, however, it added mathematical
reasoning within that cycle and argued that solving problems within mathematics can be
considered as applying mathematics in a scientific domain, in this case that of
mathematics itself. As in Ontario, the questions use real-world contexts but are not
necessarily driven by data.

The similarity of approach between the Ontario curriculum and PISA may account for the
Ontarian students’ relative success in the international assessment. In our opinion it
achieves simplicity by ignoring some distinctions that are relevant to this study of the
intersection of MDSC. Specifically, it does not provide a means of examining the different
roles played by numerical examples or data in mathematics, for example distinguishing
between the nature and provenance of numbers generated by a geometric sequence and
those obtained by measurement or sampling. Our resolution in this report is to focus on
identifying mathematics competencies most closely associated with data-driven
questions, whether from pure or applied mathematics. We have also thought it most
helpful to consider a smaller set of close associations, i.e. those competencies that must be
included in a data-driven strand or version of the mathematics curriculum. While
mathematics for its own sake can be inspiring and important, it is already widely enacted
and laid out in the English curriculum (e.g. as symbolic fluency, circle theorems).

4.2.3 The role of computing and programming in mathematical and data literacy

The role of methods - computations/calculations/algorithms - in how we transform

quantitative inputs into outputs (and similarly in how we process data in data science) is
integral to mathematical and statistical problem solving. Commenting on the role of
computation in determining what we teach, Cobb (2007) observes that “historically, we
have always tended to underestimate the extent to which what we are able to do shapes

April 2023 pg. 20

what we think we ought to do.” 36 This, coupled with modern computing having evolved as a
branch of applied mathematics, makes it difficult to disaggregate/delineate the
intersection of computing with mathematics.

Moreover, this enactment of computation/algorithmics can play out at different levels. In

mathematics, students can be taught to follow algorithms, as in the widespread ‘I do, we
do, you do’ teaching routine and/or be tasked to choose and adapt them. In computing,
students could write and implement their own programs, work with user-friendly formats
that permit some adaptation (e.g. using Python via a menu-based Jupyter Notebook) 37, or
the computation/algorithm can be packaged and treated as a ‘black box’, whose outputs
are more relevant than understanding its workings. This variation (in how we compute) can
be represented as a spectrum of computing, with programming at one end and use of pre-
packaged technology at the other end.

Policy and pedagogic issues add further tensions to the spectrum discussed above. In 2013,
the English national curriculum shifted from ‘ICT’ to 'computing’ (including using two or
more programming languages). The curriculum is specified in detail for Key Stage (KS) 3,
which may in some schools be only 2 years, and is not assessed. Some pupils will be taught
either in a cross-curricular fashion or within a potentially optional subject up until the end of
KS 4. A subset of these may choose to study computing up until the end of KS 5. The
changes from ICT to Computing are still becoming established and provision in schools
can be highly variable.

Our current readings and consultations have produced no consensus as to how much
programming is needed for school leavers, where the balance should lie between text-
based and visual/block-based programming, and whether programming should be
developed within computing or in context. Our resolution is that, for schools, we will
consider the use of technology and computing (which includes programming) as
separate but related.

36
George W Cobb, ‘The Introductory Statistics Course: A Ptolemaic Curriculum?’, Technology
Innovations in Statistics Education 1, no. 1 (12 October 2007): 10,
https://doi.org/10.5070/T511000028.
37
R. Biehler and Y. Fleischer, Introducing students to machine learning with decision trees using
CODAP and Jupyter Notebooks, Teaching Statistics 43 (2021), S133–S142.
https://doi.org/10.1111/test.12279

April 2023 pg. 21

4.3 Terminology
We end this section with some notes on terminology in this report.

Schools and colleges: in 2021, 38% of 16-17 year olds studied full-time in colleges, 38% in
state-funded schools and 6% in independent schools (with 16% not in education) 38. It
would be appropriate to refer to the competencies needed by school and college leavers.
Nevertheless, for brevity, we use ‘school' in its wider sense to refer to both.

Mathematical and data literacy: for the reasons given above, our use of ‘mathematics’
does not include ‘statistics’. We have adopted the terminology ‘mathematical and data
literacy’ as a pragmatic label with antecedents in mathematics, statistics and computing.
In our proposed framework, we will consider separately the problem solving toolkits
provided by mathematics, statistics/data science and computing.

Computing: Computing and Computer Science are often used interchangeably in higher
education and industry. Within the UK – and specifically English and Welsh - systems,
Computer Science is generally used to refer to the problem solving, computational thinking,
programming-based part of the subject area. 39 In this report, computing is used as an
umbrella term to include Computer Science, Information Technology (IT) and Digital
Literacy. 40

Disciplines and domains: we have referred to mathematics, statistics, data science and
computing as ‘disciplines’, given that they have their own academic and professional
standards. We note that professional data scientists argue that their work is truly inter-

38
Department for Education, ‘Participation in Education, Training and Employment Age 16 to 18,
Calendar Year 2021’, 2021, https://explore-education-statistics.service.gov.uk/find-
statistics/participation-in-education-and-training-and-employment.
39
https://www.computingatschool.org.uk/resource-library/2014/september/computing-in-the-
national-curriculum-a-guide-for-secondary-teachers
40
After the reboot: computing education in UK schools, 2017, Royal Society.
https://royalsociety.org/topics-policy/projects/computing-education/

April 2023 pg. 22

disciplinary. We follow them by referring to knowledge of the context where analytics will be
applied as ‘domain knowledge’ 41.

Variation, covariation and variability: Statistics is sometimes described as the study of

variation 42. ‘Variation’ is also used widely within mathematics education: to name a
learning theory (variation theory 43), to describe a phenomenon underpinning functions
(covariant relationships 44) or as a big idea related to abstraction and generalisability
(invariance and change 45). Despite these many meanings variation is a useful umbrella
term.

In statistics, ‘variability’ which occurs due to chance, to errors or to sampling, depending on

context 46. The mathematical version of variation is ‘co-variation’, that is studying the
possible relationships between variables that change together. Given a set of numbers
alone, without context or purpose, either approach can be useful and appropriate.

Extending this, we argue that the intersection of mathematics, statistics/data science

and computing is concerned with the study of variation and generalisability, with
precision and context and technology and ethics.

5. Competencies at the intersection

We consider mathematical and data literacy to be defined by common ways of working
within mathematics, statistics/data science and computing, where we attend not only to

41
Institute of Analytics, ‘IoA Analyst Competency Framework’, 2022, https://ioaglobal.org/wp-
content/uploads/2022/06/The_IoA_Analyst_Competency_Framework.pdf; AfDSP, ‘Data Science
Standards’.
42
e.g. ‘Understanding variation as the heart of statistics.’ OECD, ‘PISA 2021 Mathematics
Framework (Second Draft)’, p15.
43
Angelika Kullberg, Ulla Runesson Kempe, and Ference Marton, ‘What Is Made Possible to Learn
When Using the Variation Theory of Learning in Teaching Mathematics?’, ZDM 49, no. 4 (1 August
2017): 559–69, https://doi.org/10.1007/s11858-017-0858-4.
44
Anne Watson, Key Ideas in Teaching Mathematics: Research-Based Guidance for Ages 9-19
(Oxford, United Kingdom: Oxford University Press, 2013).
45
Ilya Sinitsky and Bat-Sheva Ilany, Change and Invariance: A Textbook on Algebraic Insight into
Numbers and Shapes, 2016,
https://search.ebscohost.com/login.aspx?direct=true&scope=site&db=nlebk&db=nlabk&AN=14
29092.
46
ACME and the Royal Statistical Society (RSS), ‘Embedding Statistics at A Level: A Report on
Statistical Requirements and Assessment across A Level Courses in Biology, Business, Chemistry,
Geography, Psychology and Sociology’.

April 2023 pg. 23

the interplay at the intersections but also to what are necessary complementary aspects
for future school leavers. The competencies that follow are organised with this in mind,
including to allow consideration of different curricular structures. Our Intersectional
competency framework is a simplification, although not so simple as to obscure real
differences of perspective. We have sought to integrate where sensible but also to show
where differences and possibilities exist. Thus we have distinguished between
mathematical, statistical and computational approaches, or ‘toolkits’ for solving problems.
It is by no means the only possible way of slicing this curricular space, but we suggest it is
one that should be intelligible and agreeable to the main stakeholders. Deciding on a
curriculum framework and then developing it – a process of years – will involve re-
organisation as well as detailing expectations at the intersections of content,
competencies, dispositions and experiences.

We propose an alignment of mathematics, statistics/data science and computing

around the overarching activity of posing and answering questions. This activity allows
for flexibility in what one asks questions about, the types of information you draw on, the
tools and methods you use and the oversight you bring; choices of which contribute to
leaning the enquiry closer either to what has been called mathematics, or data science or
computing. Nevertheless, the activity remains recognisably the same in each. The six
high-level competencies shown in Figure 3 are common across the literature reviewed.
They are related but different, and each is necessary for the resulting activity to be
considered as characteristic of the intersection of mathematics, statistics data science
and computing.

Figure 3 Mathematical, computational and data literacies as posing and answering questions

April 2023 pg. 24

We continue this section by briefly introducing the six intersectional competencies
encapsulated by this framework for mathematical, computational and data literacy. In 5.1
we show how this framework is positioned in relation to the reference frameworks in
mathematics and statistics/data science, which serves to explain the nature of the
intersectional competencies further. In 5.2 - 5.4 we unpack the competencies in
mathematics, computing and statistics/data science needed to solve data-driven
questions. In 6 we then elaborate how the competencies could be achieved at different
levels to support school-leavers’ choice of work and study focus.

Six intersectional competencies

Iterative and holistic enquiry: conduct investigations using the full

mathematical/statistical/computational enquiry cycle in situations enabling access to
multivariate data and technological tools.

Posing questions: pose and understand questions framed in (extra-mathematical)

domain knowledge and an ethical space that anticipate the properties of variation,
complexity or uncertainty inherent in data and the analytic methods available.

Solving data-driven problems: devise and implement strategies that may be non-routine
to the problem solver, drawing on a repertoire or ‘toolkit’ of appropriate concepts, analytic
methods, reasoning, representations and aids. There are differences between what is
foregrounded in a toolkit that forms a basis for data and statistical skills, one that forms a
basis for mathematics, and one for computing, which we elaborate in sections 5.2, 5.3 and
5.4 below.

Interpreting and communicating: report and justify findings to others using language,
displays and measures appropriate to the domain and the audience.

Evaluating and critiquing: compare different measures, methods, interpretations and

modes of communication; discuss their own and others’ choices and the implications for
ethics and for confidence in the results. Reflect on and make judgements about the
appropriateness of their own and others’ approaches and conclusions.

Using technology: as technologies enter classrooms, it will often be appropriate to use

technology-supported methods as part of problem-solving, interpreting, communicating
and critiquing. Students should have opportunities to engage in guided enquiry with access
to the internet and computer-based representational tools.

April 2023 pg. 25

5.1 Positioning the competencies amongst other
mathematics and statistics/data science frameworks

Why posing and answering questions?

The framework that we propose is based overall on posing and answering questions. This is
not currently foregrounded in the English mathematics curriculum but is a high-level
priority in all three mathematics and data skills/statistics reference curricula. ‘Posing and
answering questions’ is one of the two overarching competences offered by the Danish KoM
framework, with the other being ‘the ability to deal with mathematical language and tools’.

Since 2003, PISA’s approach to mathematical literacy has similarly started with a cycle
based on formulating models or problems in a context, employing mathematical methods,
interpreting and evaluating the results. The 2022 PISA framework retains this cycle and adds
a new emphasis that these processes are underpinned by mathematical reasoning,
described as “The ability to reason logically and present arguments in honest and
convincing ways” 47.

The Ontario curriculum is organised differently and focuses on principles for effective
learning. Throughout, it balances understanding and application of concepts and skills. For
example, the following premise appears in the Grade 1-8 curriculum:

“It is based on the belief that all students learn mathematics most effectively when
they develop a solid understanding of the fundamental concepts and skills in
mathematics and are given opportunities to apply these concepts and skills as they
solve increasingly complex tasks and investigate mathematical ideas, applications,
and situations in everyday contexts.” 48

By grades 11 and 12, the curriculum “embeds the learning of mathematics in the solving of
problems based on real-life situations” 49. Answering questions derived from extra-
mathematical domains is thus central to this curricular approach. Posing such questions is

47
https://pisa2022-maths.oecd.org/#Mathematical-Reasoning
48
Ontario Ministry of Education, The Ontario Curriculum, Grades 1-8: Mathematics (2020), 2020,
65, https://www.dcp.edu.gov.on.ca/en/curriculum/elementary-mathematics.
49
Ontario Ministry of Education, The Ontario Curriculum, Grades 11 and 12: Mathematics (2007),
2007, 4, https://www.edu.gov.on.ca/eng/curriculum/secondary/math1112currb.pdf.

April 2023 pg. 26

described as an ability that students develop over time (ibid., p.6) and one that “develops
their math agency” (Grades 1-8, p.75).

The computing literature suggests that students should show creativity in creating and
combining artefacts to solve a problem with goals specified by themselves and others 50.
Developing a computer program involves iterative cycles of posing requirements,
designing, coding, testing, debugging and feeding back to modify/extend requirements.
Computer science curricula such as the one in England may not centre-stage this explicitly
in terms of ‘posing and answering data-driven questions’ but even in the England
curriculum it is implicit in the very first expectation of the Key Stage 3 content: “design, use
and evaluate computational abstractions that model the state and behaviour of real-world
problems and physical systems” 51. The first of six practices in the US Advanced Placement
Program (AP) Computer Science Principles course, is Computational Solution Design:
Design and evaluate computational solutions for a purpose. It also has Creative
Development as its first foundational ‘Big idea’, which involves “a formal iterative design
process or a less rigid process of experimentation. While using either approach, developers
will encounter phases of investigating and reflecting, designing, prototyping, and testing.” 52

All the reference statistics/data science curricula include posing and answering questions
as a framing concept. There are a number of representations for the statistical or data
enquiry cycle. We share two here:

Figure 4 GAISSE II data cycle 14 (left) and IDSSP data cycle 20 (right)

50
Shuchi Grover and Roy Pea, ‘Computational Thinking in K–12: A Review of the State of the Field’,
Educational Researcher 42, no. 1 (1 January 2013): 38–43,
https://doi.org/10.3102/0013189X12463051.
51
DfE, Computing programmes of study: key stages 3 and 4, 2013.
52
AP Computer Science Principles Course and Exam Description, 2020, p. 17.
https://apcentral.collegeboard.org/courses/ap-computer-science-principles/course

April 2023 pg. 27

Moreover, it is from this perspective that including a competency Conducting iterative and
holistic enquiry becomes necessary. Not only is each step of the cycle a competency but
being able to plan for and enact the cycle is an essential and overarching competency.
Iterations are not simply repeats of the full cycle but there may be multiple overlapping
sub-cycles, as in the GAISE II diagram.

As a contrast, we note that Grillenberger and Romeike, working solely on the intersection of
computing and data science, developed a more minimal competency framework that
focuses solely on analysis and evaluation. However they comment that they are unusual in
specifically excluding the aspect ‘problem – ask questions from data' which they regard as
“being out-of-scope of the model, but without neglecting its importance” 53. Their high-level
processes, such as cleansing and optimizing, fit rather into our statistics/data science or
computational toolkits. This is not surprising since they are integrating only two disciplines.

It has been argued within mathematics education (e.g. PISA 2021 54) that it is often not
necessary to engage in every part of the modelling cycle, especially in the context of an
assessment. For statistics/data science, however, the anticipation and refinement of
questions, data representations, analytic methods and convincing arguments for end users
are unavoidable 55. Recent mathematics education research also stresses the role of
anticipation in modelling, so that “modellers must be aware of the specific role of sub-
competencies and how they interrelate in parallel and in sequence. […] This means that
modellers must be able to both enact modelling subcompetencies and synthesise these
into a global competency.” 56

The centrality of domain knowledge and ethics

Two aspects that substantially distinguish data science from traditional mathematics are
the integral roles of (a) domain-specific knowledge and (b) the ethics, provenance and
stewardship of data. Accounts of the modelling cycle (e.g. in PISA 2021) can make a virtue of

53
Andreas Grillenberger and Ralf Romeike. 2018. Developing a Theoretically Founded Data
Literacy Competency Model. In Proceedings of the 13th Workshop in Primary and Secondary
Computing Education (WiPSCE ’18), October 4–6, 2018, Potsdam, Germany. ACM, 7.
https://doi.org/10.1145/3265757.3265766
54
OECD, ‘PISA 2021 Mathematics Framework (Second Draft)’.
55
Anna Bargagliotti et al., ‘Pre-K-12 Guidelines for Assessment and Instruction in Statistics
Education (GAISE) Report II’ (American Statistical Association and National Council of Teachers
of Mathematics., 2020), https://hdsr.mitpress.mit.edu/pub/cqncbp3l.
56
Vince Geiger et al., ‘Developing a Task Design and Implementation Framework for Fostering
Mathematical Modelling Competencies’, Educational Studies in Mathematics 109, no. 2 (February
2022): 317, https://doi.org/10.1007/s10649-021-10039-y.

April 2023 pg. 28

abstraction, referring to context only at the beginning or end of a process, and with no
acknowledgement that domain knowledge will be required. Similarly, in GCSE
mathematics, the ‘Interpret’ part of a data handling question will typically have 1 mark
available, and students are routinely advised to ignore it.

In statistics/ data science, data come with a context and the frameworks take an opposing
position: “In mathematics, context obscures structure. In data analysis, context provides
meaning” 57. The IDSSP project adds the arguments that school-level data science is not
only concerned with teaching analytic techniques but teaching how data is used in society,
and that this is motivating to young people: “At all times, the focus will be on questions,
problems and data that are meaningful to their lives and attendant social and ethical
issues that arise in acquiring and working with data” 58.

Situating problems in a non-mathematical context, and engaging with the potential

messiness of that context, is therefore essential for a curriculum that addresses data
science. Indeed, there is considerable evidence from mathematics education that teachers
too readily assume that students can make connections between domain and
mathematical knowledge. Teaching contextualised problem solving, for example in
apprenticeships or Core Maths, requires time spent discussing domain-specific issues,
variables and representations 59. An antecedent lies in the Royal Society’s
recommendations for embedding mathematics in T-levels through “the use of
mathematical models that provide mathematical insight into real (and complex) situations
and problems that are simplified by making a number of assumptions.” 60

57
Cobb & Moore, 1997, p. 803 cited in Bargagliotti et al., ‘Introducing GAISE II’, 7.
58
International Data Science in Schools Project (IDSSP) Curriculum Team, ‘Curriculum
Frameworks for Introductory Data Science’, 2019, 11,
https://www.idssp.org/pages/framework.html. www.idssp.org/files/IDSSP_Frameworks_1.0.pdf
59
Geoff Wake, ‘Making Sense of and with Mathematics: The Interface between Academic
Mathematics and Mathematics in Practice’, Educational Studies in Mathematics 86, no. 2 (June
2014): 271–90, https://doi.org/10.1007/s10649-014-9540-8; Jennie Golding, Cathy Smith, and Mick
J Blaylock, ‘Learning to Teach Contextualized Problem-Solving in a Non-Calculus Mathematics
Pathway’, Teaching Mathematics and Its Applications: An International Journal of the IMA 37, no.
2 (4 June 2018): 69–83, https://doi.org/10.1093/teamat/hry009.
60
Royal Society’s Advisory Committee on Mathematics Education’s (ACME) Post-16 Contact
Group, ‘Mathematics for the T Level Qualifications: A Rationale for General Mathematical
Competences (GMCs)’, 2019, https://royalsociety.org/~/media/policy/topics/education-
skills/Maths/Mathematics%20for%20the%20T%20Level%20Qualifications%20-
%20a%20rationale%20for%20GMCs.pdf?la=en-GB.

April 2023 pg. 29

The ethical space of the domain is a necessary part of the domain knowledge. Indeed,
ethics is the one compulsory area required by Data Science professionals, including
knowledge of ethical issues relating to the domain and to the data. Undergraduate
frameworks agree that: “Academic institutions should ensure that ethics is woven into the
data science curriculum from the beginning and throughout.” 61 It is also an area of
consensus in the literature on digital literacy: “The digital world also expects its citizens to
use the technology and to respond to the ethical issues of using technology
appropriately.”62 Interviewees strongly recommended extending this to school curricula.
One noted some children can and do learn to code by studying alone, but they do not learn
ethics or the need for it.

This evidence informs our second competency Posing questions framed in a domain and
an ethical space. It is the latter part of that competency that extends what has previously
been attempted in mathematics. It is perhaps worth noting again that these competencies
anticipate each other: their order is not fixed. Enquiry may not always start with question
formulation. This is particularly relevant for data science in industry, where available data
may be the starting point for a cycle. In education, students may use given data as a way
of starting a discussion of domain issues, and then pose their own questions.

Two other competencies, Interpreting and Communicating and Evaluating and critiquing,
appear universally, more or less in this form, across the literature about data-driven
mathematics, statistics/data science and computing 63. As with problem posing, there is an
added dimension to the traditional mathematics approach because data science requires
analysts to report and justify findings with a sense of audience and using language,
displays and measures appropriate to the domain 64, and not only to mathematics. Critique
extends to comparing algorithms, interpretations, modes of communication and discussing
the implications for ethics and for confidence in the results.

61
National Academies of Sciences, Engineering, and Medicine, Data Science for Undergraduates:
Opportunities and Options (Washington, D.C.: National Academies Press, 2018), 3,
https://doi.org/10.17226/25104.
62
Meize Guo and Anne Ottenbreit-Leftwich, ‘Exploring the K-12 Computer Science Curriculum
Standards in the U.S’, in Proceedings of the 15th Workshop on Primary and Secondary Computing
Education (WiPSCE ’20: Workshop in Primary and Secondary Computing Education, Virtual Event
Germany: ACM, 2020), 1, https://doi.org/10.1145/3421590.3421594.
63
The 2007 England curriculum, the International Baccalaureate (IB) MYP and Diploma
Programme, and the US Common Core are examples of curricula that treat these as key
competencies. This is also true for the statistics and data science curricula and frameworks we
have referenced (New Zealand, California, GAISE II, IDSSP, IDS, ProDaBi).
64
https://alliancefordatascienceprofessionals.co.uk/documents/AfDSP_Standards_June22.pdf

April 2023 pg. 30

Solving data-driven problems with mathematical, statistics/data science and
computing toolkits

A central competency in this framework is Solving data-driven problems. At a high level,

the constituent competencies of problem-solving (decomposition, pattern recognition,
abstraction, pattern generalisation) are common to mathematical or statistical thinking.
With the addition of algorithm design, they also constitute computational thinking. 65
However, staying with the surface similarity of these competencies would mask how they
play out quite differently in practice. Here they are represented as distinctive aspects of
problem solving; any could be foregrounded depending on the requirements of the
problem and the skills of the solver. A conscious decision was made to keep the problem-
solving (and analysis) toolkits of mathematics, computing and statistics/ data science
separate. By toolkit, we don’t just mean a collection of techniques, but the ways of thinking
and reasoning, and the language, concepts and tools inherent to those disciplines. We
exemplify these in Sections 6, 7 and 8, but at this stage we would like to draw attention to
the following contrasts.

Prior to the arrival of modern computing, mathematical methods were our main tool for
grappling with complexity, by ‘finding the simple in the complex’ 66. Computing on the other
hand enables us to hold, process and infer from large and complex data sets. This sets up a
fundamental difference in mathematical and statistical problem solving. Mathematics
flattens the data, abstracting away the context to hone in on structural meaning. In
statistics/data science, context and data lead. Mathematical reasoning seeks to justify
and prove, whereas statistical reasoning supports inferences and makes predictions.

An illustration of this is curve-fitting. In mathematics, curves are fitted to data to explain

and classify the co-variation of target variables. The purpose is to find and use a functional
relationship. Students may ignore other variation (seen as ‘error’) or seek to minimise it by
optimising the fitting based on relevant criteria (e.g., least squares). In statistics and data
science, variation is the subject of study – not to be minimised but to be anticipated,
accounted for, analysed and interpreted.

65
https://www.computingatschool.org.uk/resource-library/2014/september/computing-in-the-
national-curriculum-a-guide-for-secondary-teachers
66
Jacob Schwartz in his essay on Computer Science in the book Discrete Thoughts: Essays on
Mathematics, Science and Philosophy https://doi.org/10.1007/978-0-8176-4775-9

April 2023 pg. 31

Using technology
Our three reference mathematics frameworks were chosen for stability, and it is not
therefore surprising that they are not built around using digital technologies. In the KOM
framework, digital tools appear within the second overall competence of “Handling
mathematical constructs, language and tools”. The Ontario curriculum has recently
introduced ‘coding’ as a strand of algebra for grades 1-8, involving writing, executing,
reading and altering programmes of increasing sophistication and efficiency. 48 PISA is
developing computer-based assessment. The global direction of travel in mathematics is
for wider inclusion of technology in both teaching and assessment, an issue that is being
studied in a complementary report for the Royal Society. Moreover, education researchers
increasingly argue that technologies do not only amplify mathematical or statistical
thinking but change it 67. Technologies used in school-level data science curricula include
mobile phones, data cards, spreadsheets, menu-driven user-friendly software
environments and programming in Python or R. All agree that engaging with technology is
not only a tool but a purpose of teaching data science.

There is less consensus around programming. In the Los Angeles based Introduction to
Data Science (IDS) project, which sees computational aspects as an essential component
of data science, programming (e.g. via RStudio) is used to merge computational and
statistical thinking 31 . Programming is a way of executing and communicating reproducible
strategies, and code serves as a notation system and as a modelling tool. Other experts
believe there is a middle-ground of building on unplugged approaches (at younger ages)
and then using user-friendly interfaces such as Jupyter Notebooks that support tool-use
without programming 68. Others, such as the New Zealand curriculum, consider that it is
more important for students to use easily accessible visualisation technologies as a basis
for inferential reasoning.

Given the diversity of viewpoints, and resource in schools, we have decided to frame the
intersectional competence Using technology at a high-level, reflecting the consensus that
students should have opportunities to engage in guided enquiry with access to the internet

67
Gould, ‘Toward data-scientific thinking’; E. Geraniou and M. Misfeldt, ‘The Mathematical
Competencies Framework and Digital Technologies’, in U. T. Jankvist, E. Geraniou (eds.),
Mathematical Competencies in the Digital Era, (Cham: Springer International Publishing, 2022),
pp. 39–60.
68
Rolf Biehler and Yannik Fleischer, ‘Introducing Students to Machine Learning with Decision
Trees Using CODAP and Jupyter Notebooks’, Teaching Statistics 43, no. S1 (July 2021),
https://doi.org/10.1111/test.12279.

April 2023 pg. 32

and computer-based representational tools. This competency informs all stages of posing
and answering problems.

5.2 Unpacking the mathematics toolkit

In this section we do not attempt to identify or characterise the whole of mathematics;
there are plenty of curricula and frameworks that provide ways of doing this. Instead we
have compared the three reference frameworks with the intersectional competencies
outlined above to identify their connections. What follows are six high-level areas of
distinctively mathematical activity that are most clearly involved when solving data-driven
problems with technology in the (late) secondary curriculum.

Solving data-driven problems within mathematics involves:

Using quantities and methods: estimating and measuring, choosing appropriate number
operations (e.g. division) and representations (e.g. rates, percentages), understanding the
effect of choices and errors.

Analysing covariation: finding and describing patterns and relationships between

variables. Proportional reasoning. Choose and apply appropriate graphical, spatial and
symbolic methods to determine/investigate values taken by one variable given knowledge
about related variable(s), including if appropriate the rate of change.

Reasoning mathematically: producing and/or critiquing extended arguments that use

mathematical methods and inferences to justify and/or prove conclusions. Use logical
comparators within methods, arguments and discussion.

Using representations: choosing and, importantly, comparing, translating and combining

different mathematical representations (e.g. 2-way tables, schematics, graphs, symbolic
forms) generated with technology where possible.

Using aids and tools: choosing and using appropriate manipulatives, visual aids and
technologies in mathematical work, understanding their affordances, limitations and
connections.

April 2023 pg. 33

These were checked for compatibility with the ten General Mathematical Competencies
(GMCs) identified for vocational T-levels69. The first two, Using quantities and methods and
Analysing covariation, are derived most directly from the focus in the Ontario curriculum on
understanding how numerical data and mathematical models are combined to explain
variation in the world. The special nature of Reasoning mathematically is highlighted in
both PISA 2022 and the KoM framework (and is not explicit in T-level GMCs). The final two
competencies describe aims for how students will use and think with representations and
tools which are inherently mathematical. One surprise in this listing is that we have not
explicitly included spatial thinking. That is solely because of the focus on school-leavers
and data-driven approaches. Spatial thinking is important for its own sake. It may also be a
context from which to derive measurements and data. More significantly, the awareness
given by spatial thinking (what the Chinese curriculum calls ‘intuitive imagination’ 70)
pervades these competencies for school leavers. It underpins notions that are central to
computational thinking such as finding structure, decomposition. It also aids understanding
of visual representations such as 2-way tables and graphs.

69
Measuring with precision; Estimating, calculating and error spotting; Working with proportion,
Using rules and formulae; Processing data; Understanding data and risk; Interpreting and
representing with mathematical diagrams; Communicating using mathematics; Costing a
project; Optimising work processes.
70
Yanjun Hong and Uffe Thomas Jankvist, ‘Comparing Two Mathematical Competencies
Frameworks from China and Denmark’, Asian Journal for Mathematics Education 1, no. 4 (2022):
494–514, https://doi.org/10.1177/27527263221144193.

April 2023 pg. 34

5.3 Unpacking the data skills toolkit
Solving problems within statistics/data science involves:

Data stewardship: attend to privacy, security, and ethical aspects such as findability,
accessibility, interoperability, and reusability and how these interact. While ethical
considerations are fundamental to posing and answering questions within the domain,
these are the ethical aspects that relate specifically to holding and processing data.

Handling data: gathering, organising, cleaning, displaying data, purposefully and with
respect to its use within a legal and ethical framework.

Data representation: choosing and, importantly, comparing, connecting and combining

different data representations, generated with technology where possible.

Statistical thinking and methods: anticipating, recognising and accounting for variety and
variability (including error or uncertainty) in data and how they shape our analyses and
predictions; engaging in exploratory data analysis: finding, describing and analysing
patterns, relationships, trends and variation using data visualisations and shape
representations; using appropriate measures of central tendency, spread and variation;
choosing and applying statistical models to support making inferences and predictions,
and testing hypotheses.

Probabilistic reasoning: as appropriate, choosing and applying a probabilistic model (and

often proportional reasoning), using randomisation and simulations.

Using aids and tools: as relevant, using a range of computational tools, including
calculators, spreadsheets, mathematical and statistical software, data analysis packages,
visualisation tools, machine learning algorithms and programming.

April 2023 pg. 35

5.4 Unpacking the computing toolkit
Solving data-driven problems with and within computing involves:

Computational thinking (including developing algorithms): for a given problem, carrying

out abstraction, decomposing it into simpler sub-problems, recognising and generalising
patterns, and developing relevant algorithms (without using a programming language).

Using algorithms: being able to follow and implement a range of common algorithms, for
example for searching and sorting data, and recognising how they could apply to a given
problem.

Programming: creating programs to solve problems using text-based or visual/block-

based programming environments.

Representing and manipulating data: choosing relevant representations and using these
to harness, transform and create data.

Safe use of technology: have a good understanding of how to use technology safely, an
awareness of threats and the protection options that are available.

Using aids and tools: using a range of hardware, software, sensors and networks, design
and computational tools, including calculators, spreadsheets, databases, component-
based software packages, and new and emerging technologies.

April 2023 pg. 36

 6. Citizen categories and school leavers
The Royal Society proposed an outline of five citizen categories, and invited reconsideration
of these in relation to specifying the needs of school leavers. Curricula with differentiated
pathways usually describe these pathways in terms of immediate goals for work or further
study. Original categories 1 and 2 describe mathematical needs necessary for citizenship,
without including any further workplace requirements. Categories 4 and 5 are graduate
STEM professions while category 3 covered a wide range in between.

Original 1 Basic 2 Mathematical 3 Traditionally non- 4 Jobs/professions 5 Mathematical

RS functional literacy to quantitative where mathematical sciences, e.g.
categories numeracy understand jobs/professions, but competences are a core mathematicians,
for and critique requiring mathematical component, e.g. mathematics
personal numerical expertise: e.g., lawyers, analysts, engineers, teachers.
decision claims journalists, civil servants, financial professionals,
making. politicians, health scientists, social
practitioners, scientists.
administrators, teaching
practitioners (not
mathematics), technical
and vocational professions
– including technicians.

Table 1 Original Royal Society citizen categories

The original framing of category 1 appears to arise from a combination of three aspects:
mathematics in this category is applied to everyday contexts where individuals typically
make decisions; it serves to construct a solution and the data involved appears well-
determined. Activities in category 2 involve interactions with, and for, other people, perhaps
in the workplace or society; they are framed as responding to claims that may involve
uncertainty.

The constructive/responsive aspect to this distinction is one that is valued in mathematics

and data sciences competency frameworks, for example the KOM framework describes
constructive/receptive facets of each competency. This duality seems worth keeping.
However, there are reasons to question the other two aspects. Few personal decisions of
any significance are based on neat and certain data. They are instead likely to involve
finding and interpreting data sources that need to be understood and critiqued. In addition,
a 2018 UK review of adults’ digital skills concludes that, in a society with widespread digital
access to information, there are no clear distinctions between basic skills and workplace

April 2023 pg. 37

needs and that basic digital and information skills act as a gateway to employment71. Our
suggestion is that categories 1 and 2 be combined as the mathematical and data literacy
that contributes to active citizenship and employment for all school leavers.

In contrast, category 3 seems wide. What Kispeter calls ‘higher-level digital skills’, and
others have called ‘techno-mathematical literacies’ (TMLs), are typically learnt within the
specialised setting of the workplace, incorporating the vital aspect of domain knowledge.
TMLs do not necessarily require new mathematical knowledge but, instead, knowledge of
the relevant data and models accessible through technology that support working
practices. Examples might be teachers’ use of pupil attainment packages to predict
performance, or electricians’ entering variables on online platforms that guide the process
of issuing safety certificates. Kispeter’s review concludes that such workplaces require ‘T-
shaped’ knowledge - broad across many areas with deep knowledge in one area. It seems
appropriate then to create competencies for many students, who should use data skills in
a range of contexts with some involving multiple variables and responsibility to others.

In such settings a distinction has been made between practitioners and advanced
practitioners. Practitioners are those who make decisions using models and quantitative
data that are provided for them (including when those go wrong 72), whereas advanced
practitioners have extra responsibility (and skills) for choosing and critiquing those models
and choices of data. This distinction appears not only in the digital skills literature, but also
in employment-based data skill literature and in the statistics literature about knowledge of
policy makers. However, these all argue that there is transition between the two levels –
practitioners can develop to use advanced knowledge; experts will need to be embedded
in practice. 73 For this reason, we have not indicated extra intersectional competencies for
this category beyond those specified ‘for many’. We also note findings from ACME’s

71
Erika Kispeter, ‘What Digital Skills Do Adults Need to Succeed in the Workplace Now and in the
next 10 Years?’, Report (London: Department of Culture, Media and Sport, June 2018),
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_da
ta/file/807831/What_digital_skills_do_adults_need_to_succeed_in_the_workplace_now_an
d_in_the_next_10_years_.pdf.
72
Nathalie J. van der Wal, Arthur Bakker, and Paul Drijvers, ‘Which Techno-Mathematical
Literacies Are Essential for Future Engineers?’, International Journal of Science and Mathematics
Education 15, no. 1 (1 May 2017): 87–104, https://doi.org/10.1007/s10763-017-9810-x; Celia Hoyles,
ed., Improving Mathematics at Work: The Need for Techno-Mathematical Literacies, 1st ed (New
York, NY: Routledge, 2010).
73
https://ctale.org/2020/10/06/what-skills-and-tools-do-policy-makers-use-to-analyse-
data/; https://wol.iza.org/articles/big-data-in-economics/long

April 2023 pg. 38

Mathematical Needs report 74 that interviewees felt the need to have studied mathematics
to a level higher than that habitually used at work. The 16-19 curriculum could include
suitable courses that develop these same competencies over a wider range of contexts
and/or with wider mathematical or statistical toolkits.

Finally, we blend categories 4 and 5 that need competencies for specialists. Category 5
differed in nature to the others since it requires in-depth knowledge about mathematics as
a discipline with valued ways of reasoning (such as deductive proof) and sociocultural
practices. From a Bernsteinian perspective, mathematicians, sociologists of mathematics,
curriculum writers and mathematics teachers are the archetypal insiders and maintainers
of boundaries in this practice. For graduates the 4/5 distinction matters as mathematical
research contributes strongly to the UK’s economic and academic progress. 75 However, at
school-leaver level, in terms of mathematical and data literacy, the needs of future
mathematicians are similar to the needs of future physicists/engineers etc in RS category 4.

The competencies that we have developed are thus proposed at three levels: for all, for
many and for specialists. They correspond to categories as shown in Table 1.

Mathematical and data Data skills for many (plus wider contexts For specialists (and mathematics
literacy for all: or tools) specialists)

Mathematical and data Workplace roles that Workplace roles Workplace roles Those who need to
literacy to understand involve decision- that involve where data science understand
and respond critically to making based on responsibility for competencies mathematics as a
numerical information quantitative data and choosing or (developed in parallel discipline, e.g.
and claims and to use models that are now critiquing with e.g. computing, mathematicians,
these for decision- accessible through quantitative data engineering, mathematics
making e.g. to buy technology e.g. and models: e.g., mathematics, teachers and
insurance, run a hair classroom teacher, lawyers, science) allow the educators.
salon. health practitioners, journalists, civil creation of analytic
electricians, some servants, models e.g. analysts,
journalists, civil politicians, lead engineers, financial
servants, politicians, teachers. professionals,
artists scientists, social
scientists.

RS 1 RS 2 RS 3 RS 4 RS 5

Table 2 Reworked categories for school leavers: competencies for all, many and specialists

74
ACME Mathematical Needs report (2011)
75
Philip Bond, ‘The Era Of Mathematics: An Independent Review of Knowledge Exchange in the
Mathematical Sciences.’ (UKRI, 2018), https://www.ukri.org/publications/the-era-of-
mathematics/.

April 2023 pg. 39

 In terms of the current English curriculum structure, the ‘For all’ level would correspond to
what must be taught in KS 3 and 4, thus Foundation level mathematics GCSE and KS 4
computing (although our competencies start from the different goal of promoting
mathematical and data literacy). At the ‘For specialists’ level, current students choose
amongst A levels in Mathematics, Further Mathematics, Computer Science (A level
Statistics is negligeable, and there is no Data Science A level). There is no single set of
qualifications corresponding to the ‘For many’ level; currently some of the cohort study
Higher Level GCSE, some study Core Maths, some study mathematics, statistics and
computing as part of T levels and vocational qualifications.

6.1 Intersectional competencies for school leavers

The following statements give more detailed competency statements for all, most and
specialist school leavers in the six intersectional competencies. They were synthesised
from the vision papers, report trials and curricula identified as reference points in 4.1,
combining their reported goals with our knowledge of young people’s educational progress
at 18.

All students can: Most students can: Specialists can:

Conducting Create and conduct investigations using ALSO use the full enquiry cycle in ALSO use the enquiry cycle with
iterative and the full situations requiring detailed attention to sophisticated tools and
mathematical/statistical/computational knowledge from another school formal reasoning appropriate to
holistic enquiry
enquiry cycle in domains related to subject. mathematical modelling/ statistical
health, personal finances, civics, reasoning/ computing.
including using data that has been
gathered from familiar digital
interactions (e.g. Youtube views),
attending to the context and variability
in the data.

Posing Pose and relate to questions framed in ALSO formulate, relate to and ALSO formulate questions using
questions domain knowledge; identify ethical connect a sequence of specialist language, tools and reasoning
implications of different ways of questions: the mathematical of mathematics/ statistics/ computing.
formulating questions; reformulate a and domain questions used to For example, in statistics, posing
question to allow use of a given frame an investigation, those inferential investigative questions
mathematical/statistical/computational used to collect data, and those regarding causality and prediction for
construct or tool. used to guide analysis and surveys, observational studies and
interpretation. experiments.

Solving Explore and use patterns of association ALSO formally model ALSO create analytic methods to
problems (with between variables, attending to relationships among multiple investigate associations; estimate and
quantitative, algebraic, visual and variables and account for quantify errors and develop
mathematics,
spatial features; plan approaches using variation from proposed models confidence/reliability measures for
computing and
appropriate models, tools and or between models; adapt solutions and predictions.
data toolkits) representations; propose solutions and approaches to take account of
make predictions. data complexity and
computational efficiency.

April 2023 pg. 40

Interpreting and Use analytic outcomes as evidence to ALSO interpret comparisons or ALSO use evidence from analyses to
communicating answer questions; identify implications combinations of analytic interpret, classify and communicate
in the originating domain (e.g. to inform outcomes (e.g. interpreting errors, bias or uncertainty and discuss
decision making). Interpret information using measures of central their implications; produce well-
in the media involving the use of data. tendency and spread, or value documented programs that are
Use language, digital artefacts and and direction of change) and accessible and re-usable;
visualisations to communicate answers provide comprehensive communicate results and formal
and how the answers were arrived at, answers; draw inferences; mathematical/statistical reasoning
including a statement of uncertainty compare interpretations; make through reports and presentations using
and plausibility when needed. sense of others’ code or multiple applications and a variety of
programs and annotate and formats.
present programs so that others
can follow; create and reflect on
different ways of
communicating answers and
conclusions for relevant
audiences.
Evaluating and Compare effects of different conditions; ALSO compare the effect of ALSO distinguish between different levels
critiquing reflect on and make judgements about systematically varying methods of rigour in evidence and arguments;
the appropriateness of their own and and conditions (including of evaluate the efficiency and applicability
others’ conclusions; consider possible implementing different common of algorithms and programs; justify or
sources of confusion, error and bias (e.g. algorithms) within the given prove conjectures and conclusions
choice of training data) and take these domain; comment on the using formal
into account when evaluating their own validity of generalisations based mathematical/statistical/computational
and others’ reasoning and on inputs and choices; evaluate reasoning and methods (e.g. for training
communication; appreciate there are the impact of outliers, missing and testing machine learning
multiple stakeholders and viewpoints; values, and the errors in algorithms).
discuss implications for ethics and for modelled solutions and
confidence in the results. predictions; compare the
outcomes of machine learning
algorithms for a classification or
prediction problem.

Use technology ALSO: use varied sources of data ALSO: use specialised languages and
Use data derived from online platforms.
(e.g. sensors, large datasets); current technologies to create
Experience working with data collected
use current technologies (e.g. representations and algorithms that
from personal devices. Use technologies
spreadsheet, graph plotter, support computation, modelling,
to implement routines and create text,
Orange) to implement and prediction, data analysis and display.
graphs, charts and visualisations.
adapt existing routines.
Table 3 Intersectional competencies for school leavers: for all, many and specialists

April 2023 pg. 41

 6.2 Mathematical toolkit competencies for school leavers
These competency statements expand on the mathematical toolkit needed for solving
data-driven problems. They are synthesised from goals set in the vision papers, report
trials and mathematics curricula identified as reference points in 4.1 and our knowledge of
mathematics pedagogy. They are not comprehensive in that they do not outline all the
mathematics needed by school leavers. Instead they outline a subset that would be
needed for a coherent course at the intersection of MSDC.

All students can: Most students can: Mathematics specialists can:

Specify and relate to familiar ALSO construct formulae for multi- ALSO: use and manipulate symbolic
Using
quantities (time, money) with a stage operations; use exponential expressions for multi-stage operations
quantities and contextual meaning, appreciate role of expressions (e.g. for compound with fluency.
units, choose appropriate number interest); use representations of
methods
operations (e.g. division) and number to show appropriate
representations (e.g. rates, accuracy. Distinguish between
percentages), use number sense and absolute and relative change (e.g.
fluent estimation to assess percentage vs percentage points).
reasonableness of claims; compare Consider the effect of systematically
effects of choices and errors. changing a parameter.

Find patterns in data or in familiar ALSO Identify and describe quadratic ALSO Identify and describe periodic
Analysing
contexts, and thus Identify and and exponential relationships in relationships. Use methods of
covariation describe proportional, linear and non- familiar contexts. Use a graph plotter differential calculus. Investigate values
linear relationships. Read graphs and to construct graphs from equations or taken by one variable given
tables to determine values taken by data. Choose and apply appropriate knowledge about related variable(s)
one variable given knowledge about graphical, spatial and symbolic and the rate of change. Interpret area
related variable(s). Interpret gradients methods to investigate values taken under a speed-time graph.
of linear functions qualitatively (0, +, -) by one variable given knowledge
as indicating rate of change. about related variable(s).

Explain how to arrive at an answer, ALSO; use general properties in ALSO: Produce and/or critique
Reasoning
working forwards or working informal deductive arguments; use extended arguments that use
mathematically backwards. Identify and use logical rules purposefully to create algebraic mathematical methods and
language involving comparators, equivalences, or to construct and inferences to justify and/or prove
conditions and negations e.g. ‘If, Then, adapt algorithms and sequences of conclusions.
Else’, AND, OR, NOT. Determine whether instructions.
statements are always, sometimes or
never true.

Use and relate to tables, 2-way tables, ALSO: choose, compare, combine and ALSO: use function notation. Compare
Using
graphs and other schematics using translate between mathematical and combine representations of
representations length to represent quantity, representations, including algebraic related variables (e.g. use the graph of
generated with technology where expressions and schematics using a function to sketch the graph of its
possible. area, volume or colour to represent derivative).
quantity.

Use hand-held and online calculators; ALSO: use visual aids and technology ALSO: Understand the affordances,
Using aids and
use tools to organise data and thinking that support graph plotting, scale limitations and connections of
tools (e.g. writing frames, grids, algebra tiles, drawing, sorting, searching and manipulatives, visual aids and
spreadsheets). repeated calculations. technologies.

Table 4 Mathematical toolkit competencies for school leavers: for all, many and specialists

April 2023 pg. 42

 6.3 Statistics/data science toolkit competencies for school
leavers
These competency statements expand on the statistics and data science toolkit needed for
solving data-driven problems. They were synthesised from the curricular levels in the GAISE
II statistics and data science standards and in the New Zealand statistics curriculum, with
some input from other reference data science sources mentioned in 4.1.

All students can: Some students can: Specialists can:

recognise the purposes for which data ALSO: ensure data that may contain ALSO: enact practices that enhance
Data
are collected and processed, and sensitive information are handled reproducibility and ensure ethical use of
stewardship attend to privacy, security, and ethical appropriately; appreciate how to data; appreciate the ethical
concerns in a responsible manner. manage data so that it is accessible, consequences of their experiments and
usable, re-usable and well- analyses; design or adapt data
documented. processing systems to address the
privacy and security of data.

Handling data gather, organise, clean, store, display ALSO: use appropriate data handling ALSO: select data analysis techniques
and interrogate data purposefully methods depending on the type of data appropriate for the type of data they
(primary and secondary data); or setting (e.g., a survey or experiment); have; connect data from different
distinguish between types of variables devise appropriate data collection sources, including for large data sets.
(e.g., categorical or quantitative) and plans; have some experience of working
understand what characteristic they with large data sets and know how to
represent; engage with data arising address missing data.
from familiar sources (e.g. mobile
phones) and from multiple sources.

choose and, importantly, compare, ALSO: work with multi-dimensional ALSO: design and code visualisations
Data
connect and combine different data interactive visualisations and graphics. for large data sets (e.g. hexagonal bin
representation representations, generated with plots).
appropriate technology; use
appropriate representations, (e.g.
proportions for categorical data, two-
way tables, tree diagrams and
confusion matrices); represent the
variability in data using appropriate
visualisations.

anticipate, recognise and account for ALSO: identify appropriate ways to ALSO: summarise their data using
Statistical
variety and variability in data (including summarise quantitative or categorical graphical displays and numerical
thinking and via error or chance) and appreciate data using tables, displays, and summaries; understand the constraints
how they shape analyses and
methods summary statistics, including using on generalising and on predictions
predictions; engage in exploratory data standard deviation as a measure of based on attributes of the data;
analysis: find, describe and analyse variability; summarise and describe understand the issues of bias and
patterns, relationships and trends; look relationships among multiple variables; confounding variables in observational
for association between two variables; recognise the uncertainty caused by studies; use multivariate thinking to
use appropriate measures of central sample to sample variability; use understand how variables impact one
tendency, spread and variation; measures of correlation; choose and another; understand the power and
describe key features of distributions apply appropriate statistical models limitations of conclusions based on
such as mean, median, range, and methods to support making samples and experiments; be able to
symmetry; use scatter plots to explore inferences and predictions (e.g. least- quantify the related uncertainty using
association between variables; squares regression line); compare margins of error; describe associations
compare conditional proportions across results for different conditions in an between two categorical variables
categorical variables; understand that experiment. using measures such as difference in
sample data can be used to answer proportions and relative risk; implement
questions about a population. bootstrapping where relevant.

April 2023 pg. 43

 use natural and expected frequencies to ALSO: use two-way tables, tree ALSO: understand the role of
Probabilistic
estimate and communicate diagrams and Venn diagrams for randomness and the effect of sample
reasoning probabilities and risks; interpret calculating probabilities including size on the variability of estimates;
contextual statements to determine conditional probabilities; use understand how and why bootstrapping
appropriate denominators for simulations to investigate associations works; use simulations to determine
probabilistic scenarios; choose and between two variables and to compare approximate sampling distributions;
apply a probabilistic model (and often groups; understand the role of random calculate and interpret expected values
proportional reasoning) where relevant, selection and random assignment in and standard deviations of discrete
in particular for risk and statistical experiments and its implications for random variables; use distributions
literacy problems. generalisability. such as the Poisson, binomial, and
normal.
use a range of relevant computational ALSO: use appropriate technology to ALSO: use technology to decompose
Using aids and
tools, including calculators, implement machine learning prediction and filter data sets and transform
tools spreadsheets, mathematical and or classification models. variables, including smoothing for time
statistical software, data analysis series data; use technology and/or
packages, visualisation tools, and program to compare and evaluate
machine learning algorithms. machine learning algorithms and
consider their impact.

Table 5 Statistics / Data science toolkit competencies for school leavers: for all, many and specialists

6.4 Computational toolkit competencies for school leavers

These competency statements expand on the computational toolkit needed for solving
data-driven problems. The competency statements below were synthesised from the
national curricula of England and Wales (subject content documents and related
guidance), individual examination board specifications as well as input from the Singapore
and US K-12 curricula, which can be mapped to the competencies below. Only
competencies which have some level of intersection with mathematics and data science
have been included.

All students can: Some students can: Specialists can:

look at a problem and carry out basic ALSO: compare a problem to others ALSO: look at a complex problem and
Computational
abstraction – removing unneeded faced previously; decompose a carry out a range of abstraction
thinking detail; decompose a problem by problem by breaking it down into sub- techniques appropriate to the problem
breaking it down into sub-problems problems using formal techniques such at hand; decompose a problem using
that are easier to solve; write a set of as top down design and with the formal techniques, with the concepts of
instructions (an algorithm) to solve the concept of a subroutine in mind so that subroutines and modules in mind and
particular problem they are faced with. the sub-problems are easier to solve; with a view to re-composition; when
write an algorithm to solve a particular writing algorithms consider developing
problem in general terms; identify some an efficient solution in terms of run-
areas of repetition or patterns within time and data storage efficiency;
the problem and identify the potential identify patterns within the problem
to reuse parts of the algorithm. and make efficient and effective use of
repetition, subroutines and pre-written
solutions.

follow a range of common algorithms ALSO: follow, discuss and implement a ALSO: follow, discuss and implement a
Using
for searching and sorting data and see range of common algorithms for range of algorithms for traversing
algorithms their applicability to real-life situations; searching and sorting data and choose graphs and domain specific problems
apply comparators when searching between them based on their efficiency and identify their use in modelling real-
and sorting data. and applicability to a given problem. life situations.

April 2023 pg. 44

 create simple programs to solve ALSO: create programs to solve well- ALSO: create programs to solve a range
Programming
carefully curated problems; have articulated problems within a given of novel problems within a new domain;
experience of at least two domain; use at least three languages, use an imperative language that offers
programming languages, one of which including a text-based imperative exposure to object-orientated
should be text based; use input, output, language and a text-based declarative programming techniques and have
manipulate variables and use simple language such as SQL; use 1-D and 2-D also had some exposure to other
iteration and selection. lists or arrays, nested selection and a programming paradigms such as
range of iterative techniques and have functional programming (such as
had some exposure to subroutines. Haskell) or logical programming (such
as Prolog); use pre-developed code
and subroutines using varying
parameters and return methodologies,
and are comfortable with modular
programming and the use of classes,
objects, records and similar structures.

convert between different numerical ALSO: perform basic binary ALSO: perform binary manipulations
Representing
bases and have an understanding that manipulations (add, shift); manipulate (add, shift, subtract, multiply, mask);
and all data can be represented in binary individual and list variables to achieve manipulate multi-dimensional list
regardless of what it is; choose suitable a desired effect; perform simple tasks variables to achieve a desired effect;
manipulating
variables for a given situation or upon data held within a relational perform relatively complex tasks upon
data problem and manipulate individual database; work with data from multiple data held within a relational database;
variables to achieve a desired effect; sources and a range of data-types. work with more specialised and
work with data generated by unfamiliar data-types, including live
themselves and their peers. and automated data collection, e.g.
remote or embedded sensors and log
data.
use standard technologies safely, for ALSO: recognise different types of ALSO: act based on a thorough
Safe use of
example choosing safe passwords, use malware and other threats; know the understanding of different types of
technology of anti-virus software, and an protection options that are available malware and other threats and the
awareness of relevant legal structures. and have an awareness of the legal, protection options that are available;
social, ethical and moral guidance discuss and debate the legal, social,
under which they should act. ethical and moral guidance under
which they should act as well as the
relevant professional guidelines.

use a range of hardware, software, and ALSO: create and combine artefacts ALSO: work with novel and emerging
Using aids and
networks, design and computational from two or more applications; use technologies, including the opportunity
tools tools, including calculators, databases and component-based to develop and create robotics, Internet
spreadsheets, software packages, and software packages; work on projects of Things, Artificial Intelligence, Virtual
programming; engage with novel and that may include some aspects of Reality, or other emerging technology
emerging technologies such as physical computing, data collection or artefacts.
robotics, Internet of Things, Artificial logging with novel and emerging
Intelligence and Virtual Reality as users. technologies.

Table 6 Computational toolkit competencies for school leavers: for all, many and specialists

April 2023 pg. 45

Part Two
The second part of this report is a feasibility study on a set of curriculum models that could
be the basis of reform in the English national system of teaching and assessment,
delivering on the educational aims proposed in Part One. In short, these are that all
students have opportunities to develop- and are assessed on - mathematical and data
literacy; some also have opportunities – and are assessed on – using mathematical and
data skills in context; finally, others are prepared to become STEM and data science
specialists. We have used the literature form Part One and the curriculum models proposed
to identify case studies that are relevant to achieving these aims. Section 7 outlines the
curriculum models; Section 8 contains the 10 case studies; Section 9 provides an overview
of these in the light of feasibility criteria and may be read without Section 8.

7. Curriculum models
The set of curriculum models considered is informed by the indicative list provided by the
Royal Society (Appendix B). Table 7 shows it fleshed out and organised by where change is
envisaged. After the ‘No change’ model 1, the top of the table (models 2- 4c) situates
change primarily in the Key Stage (KS) 4 curriculum, affecting GCSEs or equivalent
qualifications, but with the current system of specialised A levels. This would have
implications for KS 1, KS 2 and KS 3. The lower part of the table (models 5-8) envisages
major change all the way to KS 5. This could have implications for the whole future
landscape of Level 2 and 3 qualifications 76. Amongst these, the new T levels (vocational
assessments) already have some integration of mathematics within domain knowledge;
BTECs (alternative academic assessments) are more varied. The academic pathway of
GCSE resit, Core Maths and A levels could be most affected structurally.

76
Department for Education, ‘Guide to the Post-16 Qualifications Landscape at Level 3 and below
for 2025 and Beyond’ (Department for Education, 2023).

April 2023 pg. 46

 Curriculum models RS # Most relevant examples

1 No change. Current structure in England 1

2 Current Structure for KS 3 and KS 4 with 2b ProDaBi project 77 (KS 3 onwards,

content added to Computing Germany)
Change within KS 3 & 4 but

3 Current Structure for KS 3 and KS 4 with 3 Urban Data School project (KS 2/3,
content added across subjects (GCSE) England)78

4a Current Structure for KS 3 and KS 4 with 2a

content added to a Mathematics course

4b Additional content at KS 3 and 4 creates 4 New Zealand curriculum has parallel

not KS 5

two parallel or sequential Mathematics (GCSE) mathematics and statistics pillars.

GCSE courses MEI proposal has Essential Maths
GCSE followed by Maths 79

5 Current KS 5 structure with the addition of 5 MEI data science courses, IB project.
Change in KS 5 drives change in KS

credit-bearing project-based work e.g. Scotland NPAs

EPQ.

6 Current KS 5 Structure with content added 3 (post- Nuffield and Score projects adding
across subjects 16) quantitative content to A levels 80.
Geography in Core maths.

7 Current KS 5 Structure with additional 4 (post- IB has choice of course length and
content creating choice of mathematics 16) analysis/applications. Ontario is
pathways (examined as A/AS level or Core pathway-based..
Maths)

8 A new baccalaureate system at KS 5 and 6 IB (highly-constrained); NZ and

below (requiring study in core subjects) Ontario curricula (less constrained).
3&4

Table 7 Curriculum models

77
https://www.prodabi.de/en/
78
Annika Wolff, Michel Wermelinger, and Marian Petre, ‘Exploring Design Principles for Data
Literacy Activities to Support Children’s Inquiries from Complex Data’, International Journal of
Human-Computer Studies 129 (September 2019): 41–54.
79
MEI, ‘Curriculum and Assessment in Mathematics in England at KS4 and KS5 - Position Paper’
(Mathematics in Education and Industry, 2022).
80
SCORE, ‘Mathematics within A Level Science 2010 Examinations’ (SCORE (Science Community
Representing Education), 2012),

April 2023 pg. 47

 8. Case Studies
The case studies start with larger, well-established national curricula and end with smaller-
scale innovative projects trialled in England. An overview and discussion are presented in
section 9. The cases included are:

8.1 Ontario curriculum pathways

8.2 Mathematics and Statistics’ in Aotearoa New Zealand

8.3 ‘Analysis’ and ‘Applications’ pathways in the International Baccalaureate

8.4 ProDaBi – Project Data Science and Big Data in German schools

8.5 Scotland’s post-15 National Progression Awards in Data Science

8.6 MEI supplementary post-16 courses in Data Science

8.7 Royal Geographical Society and Core Maths

8.8 Cross-curricular perspectives in Civic Statistics

8.9 Bootstrap (US) integrating computing with other subjects

8.10 Urban Data School

8.1 Ontario curriculum pathways

Ontario offers an example of a curriculum with multiple pathways for older students (our
Model 7), and one where real-world applications are notably pervasive 81. The 2005
mathematics curriculum for elementary grades 1-8 is undergoing a rolling transition from
2020 82. Ontario also “de-streamed” grade 9 in 2021, following other states in a drive to
reduce entrenched inequities worsened by early pathway choices. The grade 10-12
curriculum remains in effect 83.

Curriculum change

The 2005 Ontario curriculum (influenced by US Standards) heavily emphasises

mathematical processes (problem solving; reasoning and proving; reflecting; connecting;
communicating; representing; selecting tools and strategies) in the familiar content areas
of number, algebra, measurement, geometry and statistics. It also stresses use of

81
Smith and Morgan, ‘Curricular Orientations to Real-World Contexts in Mathematics’.
82
Ontario Ministry of Education, The Ontario Curriculum, Grades 1-8: Mathematics (2020), 1–8.
83
‘Course Descriptions and Prerequisites’, accessed 23 March 2023, https://www.dcp.edu.gov.on.ca/en/course-
descriptions-and-prerequisites/mathematics.

April 2023 pg. 48

mathematics: ”Mathematical knowledge becomes meaningful and powerful in application.
This curriculum embeds the learning of mathematics in the solving of problems based on
real-life situations.” 84

Recent changes to the grade 1-9 curriculum have retained the seven processes and much
of the content 85. New strands have been added for financial literacy, responding to a
concern about basic skills 86, and for social-emotional learning (SEL) skills while using
mathematical processes. Measures and geometry have been combined in to one strand
of Spatial Sense, with attention to how this underpins mathematical reasoning in other
areas 87. Coding and mathematical modelling have been added into the algebra strand.
While applications of mathematics were already present, the new curriculum means that
students must be taught to model “messy problems”. 88 This identification of strand content
matters as they define areas that must be assessed.

All assessment is by teachers: it has a common structure in all subjects but can be tailored
to local circumstances. 89 Teachers report on student learning at 4 levels in in four main
areas: knowledge and understanding, thinking, communication, and application. For
secondary students, 70% of the assessment occurs during the course and 30% at the end.
Provincial assessments at grade 3, 6 and 9 inform the Ministry about performance trends.

Implementing the curriculum

Curriculum development in Ontario is seen as having decades of change and

consolidation, with the recent change responding to the availability of technology. 90
Teacher professional development has the highest priority in good mathematics teaching
and embedding change. Ontario’s consistent approach to improving mathematics
performance is by increasing subject-specific CPD, funding release time for teachers to

84
Ontario Ministry of Education, The Ontario Curriculum, Grades 11 and 12: Mathematics (2007), 4.
85
Christine Suurtamm, ‘Elementary Mathematics Curriculum, Teaching, And Learning: The Past Ten Years’, Ontario
Mathematics Gazette 60, no. 3 (2022): 19–24.
86
Egan J. Chernoff, ‘The Canadian Math Wars’, Canadian Journal of Science, Mathematics and Technology Education
19, no. 1 (1 March 2019): 73–76, https://doi.org/10.1007/s42330-018-0037-9.
87
Research to Inform Elementary Mathematics Curriculum Revision, 2020,
https://www.youtube.com/watch?v=AmZtIpAHZs4.
88
Suurtamm, ‘Elementary Mathematics Curriculum, Teaching, And Learning: The Past Ten Years’.
89
Jill Lazarus, ‘Secondary Curriculum Over the Last Decade’, Gazette - Ontario Association for Mathematics 60, no. 3
(2022): 52–54.
90
Lazarus.

April 2023 pg. 49

work collaboratively, in particular to low-performing schools. 91 The Ministry works within an
education ecosystem, partnering with practitioners and researchers to develop
professional learning and classroom-ready resources before implementing a new
curriculum. 92 Assessment resources are an important part of this change.

Overlapping Pathways from 15

As in other Canadian states, mathematics is effectively compulsory until age 17 (grade 11).
Students need at least 3 mathematics credits to graduate high school, with one from grade
11 or 12.

Ten mathematics courses are specified for grades 10 – 12, in three overlapping pathways,
most with online teaching versions to allow individual study. Aims are described in terms of
preparing for school leavers’ next destinations: the workplace (referred to as W below),
college or university. Within college preparation, a distinction is made between technology-
related programs (CT) and others (C). Within university preparation, the distinction is
between degree majors needing mainly statistical data management such as social
sciences (UDM); those needing knowledge of functions such as business, health science
(UF) and those also needing calculus such as STEM or economics (US).

Each course is described in terms of strands, corresponding broadly to topic areas, plus
overall and specific expectations that outline activities and provide examples. While there is
overlap between course strands, the only overarching framework is the processes. We
have therefore synthesised the course expectations to highlight five competencies that
show how Ontario brings data literacy and real-world applications into mathematics for
the range of school leavers.

Making financial decisions: all students solve problems using compound and simple
interest, more or less formally linked to sequences. Substantial parts of the W and C courses
(only) involve sourcing and dealing with data representations (e.g. payroll/tax statements,
airline websites) to make budgeting decisions as citizens and consumers.

Investigating and handling data; by 15 all students have solved problems involving 1- and
2-variable data gathered collectively from experiments (e.g. with sensors) or online
sources. W courses focus on constructing and critiquing arguments with categorical data.

91
‘Ontario Dedicating $60 Million for Renewed Math Strategy’, news.ontario.ca, accessed 23 March 2023,
https://news.ontario.ca/en/release/36364/ontario-dedicating-60-million-for-renewed-math-strategy; Haizhen Mou
and Michael M. Atkinson, ‘Want to Improve Math Scores? An Empirical Assessment of Recent Policy Interventions in
Canada’, Canadian Public Policy / Analyse de Politiques 46, no. 1 (2020): 107–24.
92
Suurtamm, ‘Elementary Mathematics Curriculum, Teaching, And Learning: The Past Ten Years’.

April 2023 pg. 50

C and CT courses extend this to dealing with 2-variable data in context; plus measures of
correlation, with equations of lines of best fit for CT; relative measures e.g. indices for C
courses. University courses vary considerably. UF and UC courses fit functions to data
without theorizing related statistical ideas. The UDM course includes making informed
decisions about sampling, questioning data validity, drawing conclusions from presented
data, all brought together in a substantive, assessed culminating investigation.

Dealing with uncertainty: all students have previously calculated and represented
theoretical probabilities. W courses revisit representations that guide personal decisions
(e.g. health risks); C course includes interpreting media use of statistics, questioning data
provenance and exploring the effects of changing assumptions. UDM course is the only
one to include combinatorics and probability distributions for discrete and continuous
variables. CT, UF and US courses only implicitly deal with uncertainty, by replacing data with
models.

Modelling with quantitative data: all students have extensive experience of solving
problems involving linear relationships or graphs of non-linear relationships. W courses
focus on situations that involve proportional and non-proportional reasoning including the
effect of errors. Other students model data with functions, taking graphical, algebraic and
computational approaches to solving contextual and mathematical problems, supported
by technology. UDM stops this in year 12, while C, CT, UF and US courses increase the range
of functions. UF includes an understanding of rate of change, and US applies this to
optimization problems.

Solving spatial problems: all students have applied measurement concepts and skills to
solve spatial problems. UDM students do no further work. W courses focus on area and
perimeter in context and consider errors. C courses include optimisation problems, sine and
cosine rule. CT courses add solving problems using trigonometry in 3-D and circle
properties. UF introduces radian measure. US course introduces and applies vector
operations and applications (e.g. in 3-D space).

Use coding to solve problems: rolling out from 2020 onwards, all grade 1-9 students will
create computational representations by writing, or reading and altering, code and
executing it in order to inform and communicate decisions. At higher grades this involves
code to analyse numerical data. Choices about using plugged/unplugged approaches or
which technologies are left to the school boards.

April 2023 pg. 51

Benefits

 Throughout the curriculum, new concepts are introduced alongside their

applications in context.
 Opportunities to use technology are frequently integral to the expectations, e.g.
coding and investigating varying parameters of graphs
 The pathways in year 10-12 ensure all students have studied situations involving
linear and exponential functional relationships (whether numerically, graphically or
algebraically).
 They also allow specialisms in data management and calculus to prepare for further
study.
 There are online versions for the range of grade 10-12 courses allowing more
accessibility. Checking a sample of these suggests they do cover the whole
curriculum including assessing applications.93
 The relative stability of the curriculum means CPD has been focused on improving
pedagogy and low-achieving schools.

Barriers

 Needed to assure parents and community that fluent number skills were considered
important.

Adaptations

 There was a need for extensive CPD for elementary teachers around coding in the
new curriculum.
 Have created a digital curriculum tool so that resources and sample tasks for
teachers are all accessed in one place.94

93
https://ontariomath.weebly.com/grade-11---mbf3c.html
94
Christine Suurtamm et al., ‘Assessment in Mathematics Education’, in Assessment in Mathematics Education: Large-
Scale Assessment and Classroom Assessment, ed. Christine Suurtamm et al., ICME-13 Topical Surveys (Cham: Springer
International Publishing, 2016), 1–38, https://doi.org/10.1007/978-3-319-32394-7_1.

April 2023 pg. 52

8.2 ‘Mathematics and Statistics’ in Aotearoa New Zealand
Aotearoa New Zealand has a well-established ‘Mathematics and Statistics’ curriculum
which has been informed by research and is seen as progressive by the statistics
education community. The national curriculum is notable for taking a strong stance by
positioning statistics alongside mathematics:

Mathematics is the exploration and use of patterns and relationships in quantities,

space, and time. Statistics is the exploration and use of patterns and relationships in
data. These two disciplines are related but different ways of thinking and of solving
problems.95

The recent history and practice of statistics teaching in New Zealand illustrates the benefits
afforded by creating parallel ‘pillars’ for mathematics and statistics within the overall
subject area up to age 15 (as in Model 4b). A third ‘applied mathematics’ pillar is under
discussion for university preparation, with debate about whether/how its content can
combine applied mathematics (e.g. modelling) and data science96.

The National Curriculum in New Zealand

Schooling is compulsory for children in New Zealand between the ages of 6 and 16 (Years 1
to 11) and then optional until 18 (Year 13). Most students leave school around the age of 17 97,
but opportunities within employment and education can influence students to stay on.

The national curriculum is intended to be broad and coherent. Mathematics and Statistics
are compulsory areas of study until the end of Year 10. In Years 11, 12 and 13 student typically
work towards one of the three Levels of National Certificate of Educational Achievement,
New Zealand’s senior school qualification. For each NCEA Level students need to earn 80
credits, meeting standards and being graded through a combination of internal and
external assessments. As part of this, some students choose to specialise in Mathematics or
Statistics, but all students have to meet a mandatory numeracy requirement of 10 credits in
order to attain the NCEA Level 1 award (as in Models 7 and 8).

The latest version of the national curriculum was released in 2007, but there have been
refinements since. The New Zealand government has invested extensively and
collaboratively in curriculum development, for example by getting curriculum advisors and

95
https://nzcurriculum.tki.org.nz/The-New-Zealand-Curriculum/Mathematics-and-statistics
96
https://new.censusatschool.org.nz/wp-
content/uploads/2021/12/NZSA_Statistics%20and%20data%20science%20at%20CL7_8.pdf
97
https://www.education.govt.nz/our-work/our-role-and-our-people/education-in-nz/

April 2023 pg. 53

teacher participants to codevelop materials for both teaching and assessment. When
Statistics was introduced in its own right, there was a recognition of a “practice-facing
imperative”98, the importance of establishing teacher confidence, and the need to align
assessment practices with the direction of curricular change. As an example, all educators
were provided with sample tasks for internal assessments before the curriculum was
implemented.

Statistical Investigation and Statistical Literacy

The New Zealand statistics curriculum is divided into three strands: statistical investigation,
statistical literacy and probability99. The first two of these are most relevant to this
discussion.

From the start of the curriculum, Statistical investigation involves conducting investigations
using the ‘statistical enquiry cycle’. One Year 1 project100 starts with students going through
their lost property box, wondering what questions they could ask and answer. They sort and
classify the items, before organising them into a physical bar graph, then moving to (and
reflecting on) an iconic bar graph. Older primary school students engage with the data
provided by Census at School101, a database both of and for students from Year 5 upwards.
As students move into secondary school, they work with more sophistication, considering
sources of variation, cleaning data sets and making inferences. There is an emphasis on
doing real investigation with real data. At the top end, this strand involves content such as
stratified sampling, box plots, linear regression and informal confidence intervals. There is a
focus on simulation-based inference, so that students gain experience with sampling and
randomisation, as well as an informal appreciation of distributions. Correspondingly, some
formal statistics is postponed until university.

Statistical literacy encourages students to evaluate both the veracity of statistical

conclusions and the effectiveness or appropriateness of statistical methods. For younger
primary school students this might simply involve recognising when a pie chart is more
appropriate than a bar graph, but by the start of secondary school there is a move to
thinking beyond the data. A 2021 position statement102 exemplified this by discussing the

98
http://www.tlri.org.nz/sites/default/files/projects/TLRI_Project_Plus_Web_0.pdf (page 30)
99
https://new.censusatschool.org.nz/new-zealand-statistics-curriculum/
100
https://sdse.online/lessons/SDSE19-006/
101
https://new.censusatschool.org.nz/explore/
102
https://new.censusatschool.org.nz/wp-
content/uploads/2021/12/NZSA_Statistics%20and%20data%20science%20at%20CL7_8.pdf (see
page 13)

April 2023 pg. 54

use of ‘worry questions’, such as “Where does the data come from?”, “Is it reasonable to
generalise from this data?” and “Could there be alternative explanations for the findings?”
By the end of secondary school, a student might have worked to evaluate statistical reports
in the media, or had experience of critiquing causal-relationship claims. There is a definite
awareness that statistical literacy is a component of citizenship.

Technology

Technology features throughout these strands to aid both statistical work and
understanding. Students working in upper primary/lower secondary classes might use
Microsoft Excel or the free online data analysis platform, CODAP103 to analyse the Census at
School data. One widely used tool in secondary schools is iNZight104, a free piece of software
developed using R that allows students quickly to interrogate, summarise and graphically
present data sets. One unusual way in which the curriculum takes advantage of technology
occurs at the top end of the curriculum, where students can use bootstrapping to estimate
a population parameter in a sampling situation. This leverages the ability of the computer
to produce large numbers of samples quickly to arrive at a confidence interval empirically.

Assessment

The current assessment processes for the NCEA offer schools a lot of choice, but in brief
mathematics and statistics are both assessed through a mixture of internal and external
assessment. This allows educators to integrate experience with real data sets and
technology into the assessment process. Although it can have its own difficulties, internally
assessed work is considered important for bringing elements of validity beyond what would
be afforded by a simple written exam. The various ways into formal assessment can create
complications however, and there is a move to simplify practice at the time of writing, while
retaining both internal and external elements105.

Where Does Data Science Fit?

The New Zealand Statistical Association Education Committee released a statement in 2021,
giving their perspective on statistics and data science education at senior secondary

103
https://codap.concord.org/
104
https://inzight.nz/
105
Royal Society Te Apārangi Expert Panel. (2021). Pāngarau Mathematics and Tauanga Statistics in
Aotearoa New Zealand. Advice on refreshing the English-medium Mathematics and Statistics learning
area of the New Zealand Curriculum. Royal Society Te Apārangi.

April 2023 pg. 55

level106. They noted that, whilst the present curriculum already touches on some aspects of
data science, proper ‘data scientific thinking’ would integrate more computational aspects,
as well as a greater awareness of data ethics, responsibilities, and ownership. Some of the
societal implications of modern technology are covered elsewhere in the New Zealand
curriculum under ‘Digital Technologies’, an area which is compulsory up to year 10, but this
course does not touch on topics such as machine learning and the use of modelling
approaches involving testing and training data.

At the time of writing there is some discussion as to whether the curriculum could integrate
a third strand of mathematics provision, ‘applied mathematics’, and how data science
could feature - perhaps heavily - within this. The same statement argues that a specific
approach to data science would allow for targeted pedagogy (and learning experiences)
that combined statistical, computational, and societal aspects of the field. For example, a
student could be challenged to develop an interactive dashboard which summarised and
reported on dynamic time series data, identifying trends and making predictions.

Benefits

 Distinguishing mathematics and statistics throughout the curriculum has afforded

more space for development in the pedagogy and practice of statistics teaching.

 The curriculum structure gives time and attention to statistical literacy as well as
statistical calculations and methods, and there is deliberate interest in promoting
critical citizenship.

 Teachers and students have access to free, statistical software, fit for purpose.

 Conversations with New Zealand based researchers have suggested to us that the
present curriculum supports higher level study for many, preparing students well for
a popular first year statistics course at a well-known university.

Barriers

 Mathematics teachers can lack a strong statistical background; this and other
factors can lead to a gap between the intended and enacted curriculum.
 The added distinction between the two areas of study contributes to some tensions
between mathematics and statistics.
 The emphasis on statistics and various pathways available to students can hinder a
student’s ability to specialise in higher education; some universities have needed to

106
https://new.censusatschool.org.nz/wp-
content/uploads/2021/12/NZSA_Statistics%20and%20data%20science%20at%20CL7_8.pdf

April 2023 pg. 56

 provide first year calculus courses to support choice of STEM majors. There is a need
to consider carefully pathways from high school to university, and how and when to
teach the mathematics that is required to support further study in both
mathematics and statistics.
 Common concerns about teaching statistics and general education continue under
this model, such as how to strike the balance between context and content when
conducting investigations, and how to promote achievement across diverse groups
of students. This last point is of particular interest to New Zealand educators as they
work to decolonise the curriculum, support students from indigenous backgrounds
and enable all students to bring their worldviews to their education.

Adaptations

 Curriculum development was informed by both teacher feedback and educational

research, both at the outset and in review.
 Teacher development continues to be recognised as important.
 Significant effort has also been put into task design and the production and sharing
of resources including those for assessment. It was recognised that teachers would
want ‘tried and tested’ resources from the start.
 New Zealand is a relatively small country, where the state closely informs curriculum
development and educational agencies tend to be centralised. Whilst a lot may be
gleaned from New Zealand’s approach, some of the initiatives and positions taken
by New Zealand may be more challenging to enact in the United Kingdom, where
there are multiple awarding organisations and interests; at the very least the larger
number of stakeholders may reinforce or multiply barriers.
 In PISA 2018 New Zealand scored 494 in mathematics, slightly higher than the
average of 489 in OECD countries107. However, there has been some decline in
performance since New Zealand has been involved in PISA, and there is also gender
disparity and meaningful achievement differences between cultural and
socioeconomic groups.

107
https://www.oecd.org/pisa/publications/PISA2018_CN_NZL.pdf

April 2023 pg. 57

8.3 ‘Analysis’ and ‘Applications’ pathways in the
International Baccalaureate
The International Baccalaureate (IB) is a globally recognised provider of international
education. The IB takes a broad and balanced approach to curriculum design, requiring
students to work both within and across a range of curriculum areas, and placing a high
value on ‘learner profile’ capacities such as communication, risk-taking and reflection.
Although the IB has extended its provision since its 1960s origin, introducing Middle Years
and Primary Years Programmes in the 1990s, it is best known for its Diploma Programme
(IBDP), a two-year pre-university course of study and qualification presently offered
worldwide in over 5600 schools in 159 countries.

A student undertaking the standard IBDP must choose one subject to study from each of six
subject groups: language and literacy, language acquisition, individuals and societies,
sciences, mathematics, and the arts. Typically, a student will study three subjects at higher
level (HL), each of which corresponds to 240 hours of study, and the other three at standard
level (SL), which each require 150 hours. To qualify for the diploma each student must also
fulfil the requirements of three core elements: studies in the theory of knowledge, a
‘creativity, activity, service’ project, and an independent, self-directed piece of writing called
the ‘extended essay’ (EE). The IB promotes continuity between all these components.

Developing the IBDP Curriculum for Mathematics

The IBDP curriculum for each subject is reviewed in seven-year cycles, with curriculum
change informed in part by the changing needs and demands of higher education and
employment. The mathematics curriculum was last revised in 2019 (for first assessment in
2021) and instituted some major changes.

In the previous iteration, students could fulfil their mathematics requirement by choosing
one of four options: HL (with or without Further mathematics HL), SL and Mathematical
Studies SL. Mathematical Studies was presented as a course that was appropriate for
students with a wide variety of mathematical abilities. It emphasised the application of
mathematics, and contained a large number of statistical techniques. Whilst the IB
internally considered Mathematical Studies to be an SL course which met diploma
requirements, it was looked on less favourably by some university admissions teams, and in
practice it was ultimately perceived as an easier option by many schools and students.

In the new iteration, there is a clearer separation of the domain of the courses and the level
of difficulty. Students now choose from two strands of mathematics, known as Analysis and

April 2023 pg. 58

Approaches (AA) and Applications and Interpretation (AI), each of which is offered at either
SL or HL. All four courses contain common Level 3 topics such as introductory calculus,
sequences, and graphs of functions, ensuring a firm base for university-level study, but the
strands later diverge to reflect differences in students’ interests and career aspirations.

The AA strand serves as a pre-university route for mathematically intensive disciplines, and
as such can be thought of as more of a standard ‘pure mathematics’ strand, including (at
HL) topics such as proof by induction and Maclaurin series. In contrast, the AI strand focuses
more on statistical literacy and mathematical modelling skills and incorporates ‘applied
mathematics’ topics such as algorithms on graphs and Voronoi diagrams. A student might
favour AI if they were planning on work or further study in a field such as business, or
psychology, where quantitative reasoning and statistical/technological fluency were more
relevant than deductive and algebraic proof. The AI course contains much more statistics
than the AA one; although an AA student will meet linear regression and the normal
distribution, an AI student will gain additional experience with, for example, hypothesis
testing, t-tests and (at HL) confidence intervals. Technology permeates teaching and
assessment in both strands, but especially so for AI students.

The SL courses maintain the same foci as their HL counterparts but have less content. A
student might choose AA SL if they enjoyed the more abstract, logical and problem-solving
aspects of mathematics but did not want to take it at HL because of their other interests or
career plans. Topics within AA SL include the laws of logarithms and how to solve quadratic
equations in sin x, cos x and tan x. A student might choose AI SL to see how mathematics
could be used in various contexts. Topics within AI SL include amortization and annuity,
fitting mathematical models to data, and chi squared tests.

When it was first introduced, the AI course was heavily promoted as it was a significant
change to what had come before. Although the AI HL course is academically demanding
and suits those going into a range of university courses, there was a danger that
stakeholders would see AI as the ‘easy’ option. The IB assuaged some concerns by
researching and sharing how university admissions teams would perceive this new option.
The new AI course has proved popular. In May 2022, out of 104,000 candidates, 56% studied
the Analysis and Approaches course (19% took HL and 37% SL) and 44% the Applications and
Interpretations course (7% took HL and 37% SL).108 Overall, the uptake of HL has increased to

108
https://www.ibo.org/globalassets/new-structure/about-the-ib/pdfs/final-statistical-bulletin-
dp-cp-may-2022.pdf

April 2023 pg. 59

26% from 15% in 2019. Teachers’ mindset regarding the new courses continues to shift as the
AI strand becomes more established, and the IB continue to refine their presentation.

Assessment and Practical Issues

The assessment of these four courses involves externally assessed written examination
papers. For AA, one of the papers is non-calculator, but AI students use calculators as
needed. There is also an independent project which is internally assessed (but centrally
moderated) and worth 20% of the final grade. This is intended as a chance for students to
choose and develop a piece of mathematics that is significant and of interest to them.
Some students find that the mathematics they need stretches slightly beyond the course,
or across strands (for instance, an AA student might find that their data does not have a
linear association and learn about Spearman’s rank correlation coefficient, which is only in
the AI strand.) The resulting projects are very diverse, but a fair number of students use
some element of statistical modelling within their projects and get to deploy important
data science skills such as sorting and cleaning data in practice. The assessments are valid
but the IB recognises that they incur significant workload for teachers. Even though they
allocate 30-hours of curriculum time, plus guidance and toolkits for developing research
skills, the production of the projects can be challenging.

HL candidates in both strands have a third written examination paper which is one hour
long and consists of two questions. This is intended to test sustained reasoning. Although
not mandated, generally, it is common for one of the two questions on the AI HL paper to
have a statistics focus, allowing for the testing of more involved statistical work under
examination conditions. However, this does not encompass the skill of working with large
data sets, which the England example shows can be problematic to integrate into
externally assessed examinations.

Since 2012, the IB also offers an alternative career-related programme109, where students
take subjects from a minimum of two DP areas and carry out a career related study with an
external provider. It is interesting that there has been significant uptake of the AI strand of
mathematics by students on this programme, even at the challenging HL level.

What Can We Learn from The Changes to the IB?

The recent changes to the IBDP mathematics courses demonstrate that it is possible to
arrange choice pathways for pre-university students (our model 7) within a baccalaureate

109
https://www.ibo.org/programmes/career-related-programme/what-is-the-cp/key-facts-
about-the-cp/

April 2023 pg. 60

structure (model 8). The four combinations of AA/AI and SL/HL go a long way towards
disentangling the level of difficulty from the focus of the curriculum, and this has
undoubtedly given students more options and broadened the appeal of HL mathematics. A
student who is good at mathematics but wants to study psychology or business, for
example, can now study a course which is more relevant to their goals but still
appropriately challenging. There was some concern amongst different groups of
stakeholders when these courses were first introduced which had to be managed.

The assessment model of the IBDP highlights both the potential and problems of extended
project work (as in model 5). The IA can result in authentic work of a very high level, and
many students successfully work with statistics for this element of the course. However, this
can place a large demand on teachers’ time and attention. Mathematical writing, and the
integration of technology into mathematical work involves new skills, and the assessment
rubric must recognise and balance the mathematics with how it is communicated and
structured. On a related note, when students choose their topic areas for the core extended
essay, mathematics is one of the least popular topic areas, partly because schools do not
share a consensus about what constitutes critical thinking with and about mathematics.

What Might the Future Hold for Data Science within the IB?

The closest relation to data science within the IBDP at the time of writing is mathematics,
particularly the AI strand, as this contains the statistical elements. Further, all students can
take a data science approach to their individual projects within the mathematics
component. The IB is already undertaking its review for the 2026 mathematics curriculum,
and one of the aspects it is looking at is how data science and modelling are handled. A
relevant question is whether mathematical modelling (with an extra-mathematical
component) may in future be considered separately from problem solving. Another is
whether elements of data science might be integrated elsewhere, for instance in the ‘digital
society’ course that was introduced in 2022. The ability to place at least some elements of
data science outside of mathematics is an advantage of a diploma structure and, with
coordination, has the benefit of reinforcing to students how uncertainty is a real-world
feature. Although a diploma programme does provide common structure to students’
learning, in practice it is still difficult to write good examination questions that involve
concepts from other subjects. These are usually very carefully scaffolded.

The IB is also currently studying how data science competencies track through their
programmes for primary and middle year students, using a continuum moving from data
literacy to data handling to data science.

April 2023 pg. 61

Benefits

 The provision of four mathematics courses, arranged across two parallel strands
and matched at two levels of difficulty, disentangles the choice of content from the
level of challenge.
 The new suite of courses is more flexible, so that the mathematical study better
aligns with students’ future plans (both in terms of education and careers), as well
as supporting students entering IB study from different mathematical backgrounds.
 The inclusion of an internally assessed project (for all students) and sustained
reasoning paper (HL only) offers valid opportunities for students to apply statistical
and data science skills, and new avenues for assessment.
 Technology is recognised and integrated throughout the courses of study.
 The mathematics curriculum is regularly revisited in light of changing international
expectations.

Barriers

 The initial introduction of the courses met with some concern and confusion from
various stakeholders, including universities, teachers, parents and students.
 The independent project is introduced, managed and assessed internally by schools,
and this can have a impact on teacher workload.
 Whilst large data sets may feature in individual students’ projects, they do not
feature in the wider curriculum or assessment yet.
 All changes and developments in assessment and curriculum have to sit within and
match the wider IB qualifications and framework.
 A limitation: the IB cohort largely consists of those preparing for university.

Adaptations

 The IB conducts significant liaison with stakeholders prior and during the teaching of
the new courses.
 The IB has significant investment in ongoing teacher professional development,
running courses and conferences worldwide, as well as providing substantial online
content. Each school also has a designated ‘IB coordinator’ who not only has
administrative responsibility but is tasked with pedagogical leadership and support.

April 2023 pg. 62

8.4 ProDaBi – Project Data Science and Big Data in German
schools
Project Data Science and Big Data at School (ProDaBi)110 is a project that has been run since
2018 by computer science education and mathematics education researchers at
Paderborn University. The project is funded by the foundation Deutsche Telekom Stiftung.

ProDaBi has been developing and trialling data science education materials (and some
associated professional development) intended to raise students’ awareness of data as a
model of the world, data processing, exploration of data, and to teach them how to use
data-driven machine learning (ML) for making predictions and decisions. The project
started with a year-long trial with Grade 12 (Year 13) students in two schools in Paderborn,
with three-hour weekly sessions, run over seven months in the school year 2018/2019 111. In
subsequent years materials have been developed and trialled lower down the age range,
first with teaching modules for lower secondary aged students, and then into the top end of
primary school. Lessons have been typically taught by computer science teachers.

ProDaBi offers an example of how a computing-focus approach to data science is

presently being enacted (our Model 2 for adding content to computing). It also embodies
the paradigm that machine learning is a fundamental component of data science which
can be communicated to school-aged students.

Predictive modelling as a fundamental idea in data science education

A starting premise for the ProDaBi project is that the industrial processes of data science
are distinct from the stages of the statistical modelling cycle in many important ways.
Significant amounts of data may already exist which have not necessarily been collected to
a plan; the domain/business context is a crucial starting point; data preparation and
cleaning are separate, time-intensive steps; and training and testing a model with data is a
distinct and important phase of the data science cycle 112. Instead of reaching a statistically
rigorous conclusion, now the goal often is to develop, validate and deploy a model which
can be used for context-informed classification or prediction. This in turn raises new and
important issues of social responsibility and ethics.

ProDaBi developers list three aims for their activities – demystifying the data-based
machine learning (ML) model building process, investigating ML by developing their own

110
https://www.prodabi.de/
111
https://publikationen.bibliothek.kit.edu/1000127944/97090875
112
Please see slides and video at https://iase-web.org/Webinars.php?p=230307_2000

April 2023 pg. 63

models, and reflecting on responsible use of ML. One reason that ML is seen as integral is
that data sets (and real-world contexts) typically involve multiple variables and tracking
the resulting multitude of correlations by hand is simply not possible. Students could be
introduced to simple classification models as an example of machine learning, for example
deciding whether an image is of a cat or a dog, or determining whether a lizard has come
from a natural or disturbed habitat based on attributes including its mass, length of limbs,
tail length, head depth and toe-pad width 113.

Pedagogical and computational models

The ProDaBi project uses the notion of decision trees to develop and test its predictive
algorithms, where at each stage of a multi-stage process one uses a particular attribute
(and a threshold) to categorise a candidate. Decision trees serve as a transparent model
for students, helpful for understanding the ‘machine room’ of artificial intelligence (AI),
compared to, say, artificial neural networks (ANN). Decision trees provide an opportunity for
teachers and students to discuss potential advantages and disadvantages of automation,
and the developing role of humans in AI 114.

Students meet decision trees for the first time at the top end of primary school, where they
work with physical cards that name an object and list some of its attributes. This
‘unplugged’ approach doesn’t introduce them to the mathematics or computation of data-
based ML but the students get to explore the heuristics and properties of the process,
recognising that different algorithms can result in different outcomes and having an
opportunity to compare and discuss the accuracy and biases of their choices.

The resources provided for early secondary school students involve the students using
larger data sets with semi-automatic support, specifically the computing environment
Common Online Data Analysis Platform (CODAP) 115 which facilitates ‘drag and drop’ for
articulating commands. At the top end of secondary school students can use Jupyter
notebooks (cell-based environments that support creation or adaptation of Python code)
or Python programming to access and work with large data sets, with Python libraries to
create decision trees and related visualisations. Students are also tasked to grapple with

113
Example from GAISE II report, p. 97 https://www.amstat.org/asa/files/pdfs/GAISE/GAISEIIPreK-
12_Full.pdf
114
Slide 28 at https://iase-web.org/Webinars.php?p=230307_2000
115
https://codap.concord.org/

April 2023 pg. 64

assessing rules using misclassification rates, issues of overfitting data, false positives and
false negatives, growing and pruning decision trees, and evaluating their algorithms 116.

Meaningful contexts to engage students

As well as giving students experience of machine learning processes, the ProDaBi resources
demonstrate the importance of context in Data Science and are based around
recognisable scenarios from the real world.

At the top end of primary school ProDaBi uses the example of choosing/recommending
food items (based on nutritional information). Students are given 55 data cards and green
and red paper clips to label the cards. Two example cards are given below:

Figure 5 Data cards – Nutritional facts of food items (Slide 56, at https://iase-web.org/Webinars.php?p=230307_2000 )

In this activity, students get to manually create a decision tree as a rule system for
classifying food items, test their tree with test data, compare different trees/approaches
and explore how a machine learning method might create a decision tree.

116
Biehler, R., & Fleischer, Y. (2021). Introducing students to machine learning with decision trees
using CODAP and Jupyter Notebooks. Teaching Statistics, 43(S1), S133-S142.
https://doi.org/10.1111/test.12279

April 2023 pg. 65

 Figure 6 A decision tree for food items (Slide 64, at https://iase-web.org/Webinars.php?p=230307_2000 )

One of the contexts in the materials for older students is based around social media
platforms’ use of online gaming behaviours to decide which advertisement is shown to a
player, with data fields including the type of device someone uses to play games, the online
platforms they use, the types of game, how often they plan, and so on. An authentic data
set to support this resource was created by the ProDaBi developers working with secondary
schoolchildren.

Figure 7 A student's solution creating a decision tree with CODAP (Biehler & Fleischer, Teaching Statistics. 2021;43:S140 37)

Students can use the Jupyter Notebooks to focus on one menu-based action at a time or
move on to use more versatile code-based actions; they can also use the notebooks to
add explanations and narrative.

April 2023 pg. 66

Benefits

 The unplugged/embodied, menu-driven and Python code-based implementations

allow different age groups to explore ML.
 Relevant and authentic contexts support student engagement and sense-making.
 Students get to appreciate the role and need for computational tools, automation
and ML.
 Students see and grapple with the ethical implications of ML.

Barriers

 The mathematical demand of some of the constructs (confusion matrices / two-

way tables) relating to the decision trees used in some of the resources can be
problematic – students need to be able to use these.
 Teaching was done by computing teachers and there are related teacher supply
issues.
 Some teachers may have difficulties filling in student gaps in mathematical
knowledge.

Adaptations

 Team teaching could be considered to draw in complementary knowledge across

disciplines.
 The developers are experimenting with different sized data sets to draw out different
aspects of ML.

April 2023 pg. 67

 8.5 Scotland’s post-15 National Progression Awards in Data
Science
Scotland’s “Curriculum for Excellence” 117 lays out a programme for students from ages 3 to
18, consisting of a broad general education followed by a senior phase from the ages of 15
to 18, during which students can begin to specialise and attain formal qualifications.

Scotland is notable for having recently added a set of three national qualifications to the list
of options: National Progression Awards (NPA) in Data Science 118. Although these have had
limited uptake to date, their construction and outworking speaks to how an additional
distinct qualification in data science can exist alongside other qualifications in
mathematics. (This approach supplements rather than replaces existing qualification so is
similar to Model 5)

The Data Science NPA

There are three levels of NPA in Data Science, labelled 4, 5 and 6 after the matching levels of
award in the Scottish Credit and Qualifications Framework (SCQF) 119. The qualifications are
intended to provide students with a foundation of knowledge in data science, developing
both applicable skills and an “awareness of the societal aspects of this important,
emerging technology” 120.

The three Data Science NPA sit against a background of related national mathematics
qualifications, such as the Higher ‘Applications of Mathematics’ (also SCQF 6) which
includes statistical content such as regression and hypothesis testing. However, the NPA
have a specific focus on data science and have been designed with flexibility in mind;
students can work towards these qualifications in school or in college, such that they might
also be appropriate for adults wishing to retrain.

The level 4 NPA consists of two units: data citizenship and data science. These cover basic
statistical techniques and data visualisations, but also require students to reflect upon how

117
https://education.gov.scot/education-scotland/scottish-education-system/policy-for-
scottish-education/policy-drivers/cfe-building-from-the-statement-appendix-incl-btc1-
5/what-is-curriculum-for-excellence
118
https://www.sqa.org.uk/sqa/91458.html
119
These are comparable with levels 1, 2 and 3 in the UK’s Regulated Qualifications Framework.
120
https://www.sqa.org.uk/sqa/91458.html

April 2023 pg. 68

the ways in which data is collected, processed, and presented can have consequences and
make a difference, both to individuals and in society.

The level 5 and level 6 awards each contain more advanced versions of these core units as
well as a third module from a choice of: data security, computer programming, machine
learning, a data science project, data science statistics, and (at level 6 only) an additional
statistics option121. Within these modules students might engage with primary data from
environmental sensors, see how machine learning uses testing and training datasets, or
develop their programming skills. The choice of modules again reflects the flexibility of the
qualification; it is envisioned that the overall award could be taught across departments,
with teachers from mathematics, computer science or geography teaching towards their
own strengths and interests. Similarly, whilst students can develop programming skills in a
language such as Python if they choose, each NPA can be attained using only Microsoft
Excel or Google Sheets.

At the time of writing there has been limited but developing take up of the qualifications; an
estimated 1000 students have studied the NPA since it was introduced in 2020.

Delivery Models, Assessment and Teacher Support

The adaptability of the Data Science NPA has been promoted as a strength. The educator’s
guide122 is written with an awareness of the fact that many teachers have to work with
classes where students are working at different levels of the NPA, or even on different
qualifications. Even where students are all studying together, they might encounter the
Level 4 NPA as an intense enrichment experience within their mathematics lessons, or
tackle all three levels over multiple years with a view to progressing to progressing
ultimately to a degree in Data Science. There is even overlap with other qualifications; for
instance, the data security module also counts towards the NPA in Cybersecurity.

Schools can set their own assessments for their students, although these need to be
checked against unit specifications and approved before use. Alternatively, the core units
can be assessed through a combination of computer-based tests and centrally provided
projects which are marked by teachers then submitted online.

The courses are strongly supported by work carried out by Data Education in Schools 123, a
group of researchers, educators and consultants based out of Edinburgh University. They

121
https://dataschools.education/wp-content/uploads/2020/11/educators-guide_Oct2020.pdf
122
https://dataschools.education/wp-content/uploads/2020/11/educators-guide_Oct2020.pdf
123
https://dataschools.education/

April 2023 pg. 69

provide an extensive array of materials 124 developed with the data science organisation
Effini 125, and specifically tailored to the NPA.

Future Developments

Although the NPA is only of direct relevance to senior secondary school students, there is
growing interest in developing data science content and data literacy throughout the
curriculum stages. For example, Data Education in Schools also provides resources for
younger students. These link to the curricula of multiple subject areas, touching on many
recognisable contexts, and integrating technology at various points.

Benefits

 Having a separate qualification allows a broad range of data science content and
skills to be covered.
 The flexibility built into the qualifications allows students, schools and further
education colleges to engage with data science in a number of ways, promoting
access and ultimately supporting equity.
 The modular approach to the qualification encourages participation of staff from
different curriculum areas, hence using the cross-curricular nature of data science
as a strength.

Barriers

 The qualification exists in a crowded field.

 It can be challenging to balance the relative difficulty of various modules.

Adaptations

 To increase the relevance of the qualification it has been designed and promoted so
that it sits comfortably across the school/further education boundary.

 Teachers are supported with some high-quality resources and there are opportunities
for professional development (although it is recognised that more are needed.)

124
https://dataschools.education/npa-course-materials/
125
https://effini.com/

April 2023 pg. 70

8.6 MEI supplementary Data Science Courses
MEI is a charity organisation concerned with advancing mathematics education. They see
data science as an important interdisciplinary field which is gradually filtering down into
schools. Since 2020 they have supported schools by delivering online data science courses
aimed at students in Years 12 and 13, designed to sit alongside either A level Mathematics or
Core Mathematics. Such courses could be the basis of credit-bearing extra qualifications
(our Model 5)

The MEI data science courses

MEI presently offer two online courses. The shorter course is a free, open access self-study
programme designed as an introduction to data science 126. Over six lessons, students
cover topics including cleaning data, data visualisation, and a brief introduction to
machine learning. The course also involves work with the A level Mathematics large data
sets, and teachers can recommend it to their students as a way of increasing their
understanding of working with data.

The longer course is a term-long taught programme. MEI runs each presentation of the
course with limited spaces; teachers apply on behalf of their students, so one cohort might
involve about 150 students from 30 different schools. Each week students take part in a
one-hour live session and support this with independent study. This course is formally
assessed through a combination of a practical task and a one hour written assessment,
each worth 50%. Although schools must administer the assessment, MEI set and mark the
work, and award certificates. There are three levels of award: pass, merit and distinction.
Students who submit the assessments but do not achieve at least a pass in both are sent a
certificate of attendance.

The vast majority of the students involved with the longer course are A level students, but
the mathematics involved is accessible to both A level and Core Mathematics students,
covering topics such as linear functions, scatterplots, correlation and regression. There has
been a conscious decision to circumvent probability within the course and to emulate the
processes of data science; the goal is to create predictive models rather than explanatory
ones, processes such as linear regression are presented as ‘black boxes’ enacted by code,
and relative judgements are based on metrics, not probabilities.

Both courses involve some element of programming. During development, MEI considered
various tools but ultimately chose to use the freely available, web based Kaggle platform

126
https://mei.org.uk/introduction-to-data-science/

April 2023 pg. 71

which allows students to combine text and coding in Python (which is already taught in
many schools as a foundation language and is widely used in industry). However, MEI is
clear that the data science course is not a programming course. Students mainly import
libraries for data analysis, visualisation, and machine learning, and alter routines using the
commands in those libraries. By providing students with sample blocks of code to copy,
adapt and build on, the courses intend to recognise the relevance of programming to data
science and support transferrable skills without assuming too much at the point of entry.

Data science sensibilities

The courses were developed in collaboration with computing educators and influenced by
the International Data Science in Schools Project (IDSSP) 127. The courses have also been
reviewed by data scientists in industry and academia. Their content and delivery both
reflect certain key data science sensibilities.

The course positions technology as central to data science. The course developers believe
that data at scale needs technology, and that meaningful, authentic work can rarely be
done with pencil and paper. They also recognise the import of representations in data
science, for example when students are exploring distributions, or presenting their findings.
Students need access to a range of visualisations, including some dynamic ones, and the
affordances of these visualisations are unavailable without technology.

Both courses also evince the importance of context, or domain knowledge in the practice of
data science. The contexts covered within the courses are varied and have been picked so
as to be accessible and of interest to 17-year-olds. They include hotel cancellations, house
prices, engine sizes and emissions, climate data and basketball players. Students access
these contexts via historical, open-source data sets that are closed to revision, and which
may require some data cleaning. Context is also used explicitly to inform method; for
instance, students might be required to make a sensible context-based decision as to
whether an outlier constitutes an error or a relevant piece of data.

The courses touch on some aspects of ethical issues surrounding data. There is no
discussion about data stewardship or privacy, possibly due in part to the time available, but
there is exploration of the implications of using models in context, and how bias
(conceptualised as a mismatch of the deployment of the model to the data) can be
introduced to a model through the choice of training data.

127
http://www.idssp.org/

April 2023 pg. 72

The short course integrates insight into career opportunities in data science, including
video interviews with working data scientists.

Machine Learning, AI and the relationship Between mathematics, computing and data
science

Both courses integrate aspects of machine learning and AI. A pyramid shared with students
in the short course starts with data sources and pre-processing at the base, moves up
through data exploration and analysis, then goes onto machine learning before finally
reaching decision making. Practical experience with training and testing data also features
in some of the tasks; a large data set might be split, so that 80% can be used to build a
model whilst the remaining 20% is held back to test it.

However, a representative of MEI made it clear to us in conversation that they saw data
science as part of the mathematics and statistics curriculum. They offered three main
arguments for this: that mathematical and statistical analysis and skills form the
foundation of data science and artificial intelligence; that data science is pedagogically
informed by statistics education; and that whilst data analytics is grounded in computing,
data science places a higher import on context than computing has traditionally done.
Notwithstanding, they also recognised that computing had an important part to play, and
that teachers moving from mathematics might be uncomfortable with some aspects of
computational thinking, including the acceptance of ‘black box’ procedures and the
readiness to engage with complicated, multivariate contexts.

It was also suggested that placing data science closer to mathematics and statistics than
computing might steer around some of the significant gender gap present in A level
computing; 37% of the students covered by the post-pilot survey 128 were female, compared
to 15% of A level computing candidates in summer 2022 129.

Benefits

 Positioning these courses as separate, supplementary study means that there are
fewer demands and restrictions on the content.
 All the staff involved in delivering the course are specialists with good knowledge
and understanding of data science as a discipline.

128
https://mei.org.uk/app/uploads/2021/08/mei-ds-short-course-pilot-report.pdf
129
https://www.jcq.org.uk/wp-content/uploads/2022/08/A-and-AS-Results-Summer-2022-
Compared-to-2021.pdf

April 2023 pg. 73

  Authentic, large-scale data is used within the courses, and the assessment for the
taught course involves both a practical task and a written paper; the examination
and certificate promote involvement and add cachet.
 Successful integration of user-friendly but commercial-grade technology

Barriers

 The courses have been popular, and the longer course has always had a large
number of students applying, but it would be both practically and economically
challenging to reproduce this collaborative approach at scale.
 The short lengths of the courses limit both the width and the depth of the content
compared to other curriculum domains, and there is no provision at all for younger
students.
 It can be hard to find good quality data sets in contexts which 17-year-olds will
understand enough to appreciate the relevant attributes, whilst still be interesting
enough for a student to be motivated genuinely to ask and answer a question.

Adaptations

 MEI have balanced or adapted the statistical and programming content so that the
course is accessible to a wider range of students – in particular, the course is
suitable for interested students studying either A level or Core Mathematics.
 The courses involve collaboration between MEI and schools, from the point of
application to the administration of the examination.

April 2023 pg. 74

8.7 Geography and Core Mathematics Project
The concepts and skills taught in school mathematics have relevance and application
within other curricula. The 2017 Smith Report 130 for instance notes that mathematical and
quantitative skills form part of the formal assessment of a third of non-mathematics A
levels. This intersection is particularly evident through the teaching of Core Mathematics,
with its emphasis on applied mathematical skills and authentic cross-curricular contexts.

The Geography and Core Maths Project 131 is a DfE funded partnership between the
Advanced Mathematics Support Programme (AMSP) and the Royal Geographical Society
(with the Institute of British Geographers) (RGS) intended to consolidate one such
connection. Through a combination of continuing professional development (CPD) courses
and online resources this initiative works to support the approximately 80% of A level
Geography students who are not studying A level Mathematics 132, and to support their
teachers in delivering quantitative and statistical content. It provides a single-subject
precedent for our Model 6 (current KS 5 structure with content added across subjects) and
speaks to how joined-up approaches inform future avenues for professional development.

Mathematical and statistical skills in geography

Throughout secondary school, geography lessons regularly incorporate aspects of

mathematical and statistical reasoning. This can call on skills that are well covered by the
mathematics curriculum, such as finding a percentage change or reading a compound
bar chart, but it may also involve more domain specific work, for instance working with
geospatial data or understanding a kite diagram. A level Geography also includes some
inferential statistical methods which go beyond the content of GCSE Mathematics. Notably
at A level, the geography programme of study includes a non-examined assessment (NEA)
worth 20% of the grade, for which each student must conduct some fieldwork and write up
a report, including sections of data processing and analysis.

In this way geography teachers are required to touch on a significant amount of

mathematical and statistical content, some of which is likely to be new to their students, or
even to them. This challenge is particularly pronounced for some when facilitating the NEA,
as there is less support from textbooks and prepared examples. Projects often involve
primary quantitative data drawn from observation or the use of tools such as
anemometers and pH gauges, and in each case geography teachers must first steer their

130
I’m assuming we already have this reference, but Smith (2017) para 113
131
https://www.rgs.org/geomathsbriefing/
132
https://www.rgs.org/geomaths/

April 2023 pg. 75

students towards appropriate statistical methods for their chosen context, and then
support them in reaching conditional or qualified conclusions.

The data-related elements of the present geography curricula are consonant with the
mathematical and data literacy competencies presented elsewhere in this report. For
instance, the fieldwork component of the A level course includes some discussion of data
gathering and ethics, although there is limited discussion about cleaning data or data
protection issues. Teachers can also bring large data sets into the geography classroom,
for instance by accessing the NOMIS data on the labour market 133. Data trends can be
identified using online analysis and visualisation packages such as Gapminder 134 and
students can work with location-linked data via representations such as heat maps and
flow maps via software such as ArcGIS Online135.

The RGS-AMSP Collaboration

Since 2019 AMSP, in liaison with the RGS, has delivered tailored CPD sessions for geography
teachers working with mathematics. The broad goals of the project are to develop
geography teachers’ data skills and confidence, and to promote the parallel study of
geography and core mathematics 136. These sessions generated immediate interest,
particularly as recent changes to geography specifications had increased the expectation
for geography teachers to integrate data skills into their teaching. The CPD provision was
adapted over the pandemic period, and at the time of writing AMSP are offering smaller
sessions which focus on specific content or processes, such as chi-squared tests,
Spearman’s rank correlation coefficient, or how to engage with big data sets. These
presently involve hundreds of teachers, with more teachers participating each year.
Sessions are recorded for those who are unable to attend, with a certificate for those
attending four sessions or more out of a set of six. AMSP also offers dedicated online
support via the MEI Integral platform. Feedback from the sessions has been positive;
comments from participants suggest that this programme has developed geography
teachers’ knowledge of statistics and how to teach it, as well as bolstering their confidence
and willingness to integrate technology into their teaching.

133
https://www.nomisweb.co.uk/
134
https://www.gapminder.org/
135
https://schools.esriuk.com/
136
Similar provision is also offered for teachers of biology and psychology. Although this has also
had some success, it is not detailed here for reasons of space, and because the respective subject
organisations have been less closely involved.

April 2023 pg. 76

Benefits

 For students, cross-curricular teaching of statistics can create learning situations

where the questions and the answers are perceived as being meaningful; abstract
teaching of statistics can lean on repeated calculation and ersatz contexts.

Barriers

 The developing prominence of statistical methods within A level Geography has

moved some teachers outside of their comfort zones, leaving them feeling
challenged and necessitating space for and provision of professional development.
Whilst AMSP is addressing this need for some, it can be hard for teachers to access
CPD.
 It can be challenging to assess meaningfully some statistical/data analysis skills
through timed written assessment. Context-based learning of methods such as
Spearman’s rank correlation coefficient may end up being reduced to cloze-type
completion questions.
 Cross-curricular delivery across siloed subject areas meets barriers due to small
differences such as vocabulary, as well as larger difficulties due to specification
changes. For instance, geography teachers have noted that students cannot now be
assumed to be familiar with spreadsheet packages since Information and
Communications Technology (ICT) has been superseded by Computing in many
schools.

Adaptations

 For teachers, access to professional development which is jointly devised and then
revised by both a subject association and a mathematics education organisation
warrants relevant, considered content and sensitivity to the demands of both areas.
 A formal relationship between the RGS and AMSP has supported long-term provision
which has developed in response to evolving teacher needs, combining live and
asynchronous learning.
 Some awarding bodies (for example Pearson 137) have provided additional
documentation, outlining the nature of the curriculum overlap, differences in
approach and vocabulary, and noting common misconceptions.

137
https://qualifications.pearson.com/content/dam/pdf/A%20Level/Geography/2016/teaching-
and-learning-materials/Maths_for_geographers_guide.pdf

April 2023 pg. 77

8.8 Cross-curricular work in Civic Statistics
The term ‘Civic Statistics’ was coined by an international collaborative group (ProCivicStat)
funded by Erasmus+. A product of the ProCivicStat project is a collection of teaching and
learning materials designed to support statistics teaching that engages with current social
issues. 138 The team behind ProCivicStat offer on their website that:

We believe that students should see the usefulness of statistics in understanding

evidence, and should engage with data about important social phenomena, to
support their development as active and empowered citizens.

Social phenomena are complex, and democracies need citizens who can explore,
understand, and reason about information of a multivariate nature. Most statistics
courses fail to teach these skills 139.

This position has been illuminated by the recent pandemic, where citizen engagement with
statistics involved contested concepts and metrics, multivariate phenomena, and non-
linear relationships 140. The ProCivicStat authors go on to offer an eleven-facet framework
for teaching civic statistics141 which is designed to ensure fair coverage of engagement and
action, knowledge, and enabling processes.

Civic Statistics in the Curriculum and the Importance of Citizenship

Although the ProCivicStat resources do not constitute a full tried-and-tested curriculum,

and there is limited research on the efficacy of this approach, the decades of experience
and thinking of the movement behind ProCivicStat offer some insight into how citizenship
aspects of data science might be integrated into other subjects (Model 3). Together the
resources are designed to promote civic engagement through statistics, although not
necessarily in mathematics lessons. Materials cover a wide range of topics such as income
inequality, migration, and ageing societies. Teacher guidance is provided in multiple
languages and linked to outside sources of data and support materials. Many contexts sit

138
These materials are hosted on the website of The International Statistics Literacy Project (ISLP)
(which in turn is a subgroup of the International Association of Statistics Education (IASE), the
education section of the International Statistics Institute (ISI)).
139
https://www.iase-web.org/islp/pcs/
140
J. Ridgway, Covid and data science: Understanding R0 could change your life, Teaching
Statistics 43 (2021), S84–S92. https://doi.org/10.1111/test.12273
141
https://iase-web.org/islp/pcs/documents/Conceptual_framework_brief.pdf

April 2023 pg. 78

more comfortably in subjects such as geography or politics, where students might be
better prepared to ask and investigate meaningful questions of interest to them.

One contemporary starting point for work in Civic Statistics which was shared with us was
the OECD Better Life Index 142. This website offers an easy-to-use graphical interface where
students can adjust the relative importance of eleven indices and rank the quality of life in
various countries, and illustrates well how teachers might adopt different but
complementary approaches. A mathematics teacher might use this to support work on
proportional reasoning and weighted averages, and a data science educator could get the
students to download the dataset, edit it as necessary for specific purposes and then
explore other representations. A critical citizenship perspective, which might be adopted by
an economics teacher, would further bring in questions such as “what are the benefits and
limitations of indicator systems such as this?” and “how can you measure democracy?”
These questions are consonant with further data science sensibilities as students begin to
interrogate the source and handling of data, as well as associated ethical issues.

The term ‘civic statistics’ might suggest that it should be taught within citizenship, a
foundation subject at KS 3 and KS 4. However, an academic interviewed in associated with
the project noted that the teaching of citizenship in UK schools is not always valued, even
though it invokes a wide range of issues with substantial impact on students. They therefore
felt that coordinating civic statistics across established curriculum areas such as English
and Business Studies would be both more effective and more authentic than separating it
as a component of taught citizenship. Teachers should be encouraged to challenge with
their students the statements concerning their domain that they see in the media, perhaps
by engaging with reports from fact-checking organisations such as FullFact 143 in lessons or,
for example, demonstrating how choosing the position of the origin on time-series data for
salaries or inflation can steer the implications of a graph.

Civic Statistics and Data Literacies

Whilst they are primarily built around the goal of getting students involved in social issues,
the ProCivicStat resources regularly take an approach which resonates with many data
science principles and practices. Context is paramount and data visualisation often forms
an integral part of the work, which can be invaluable in helping students make sense of the
data. Technology affords students the chance to work with real data sets, and the main PCS
resource database is searchable not only by statistical topic and level of difficulty, but also

142
https://www.oecdbetterlifeindex.org/
143
https://fullfact.org/

April 2023 pg. 79

by the applications or platforms which the students will be using. These include Gapminder,
INZight, R and CODAP. Notwithstanding, civic statistics itself does not involve any
compulsory use of explicit or original programming.

Future Developments in Civic Statistics

The nature and approaches of civic statistics continue to be discussed 144 against a
background of pressing issues including climate change, diversity, and social equality.
Developments in content include an increased awareness of machine learning processes:
common techniques such as decision trees might be modelled with students through
activity and analogy, and students of all ages can be expected to engage with and
evaluate the output of AI, and to reflect on the role it might come to play in society.
Advances in pedagogy include a consideration of how civic statistics might be assessed
through project-work or other non-traditional means.

Benefits

 A civic statistics approach in principle offers an authentic enhancement of multiple

curricula in a way that intersects and resonates with many aspects of data science.
 The broad and flexible scope of civic statistics could appeal to a lot of stakeholders
and meaningfully contribute to progress with regard to some equality, diversity and
inclusion issues in education.
 Civic statistics could be integrated into multiple curriculum models; for example, it
could be delivered as part of a citizenship-facing component within a
baccalaureate model or across multiple subjects in secondary school with the help
of an in-school coordinator.
 Developments in technology and the interest associated with current affairs mean
that there are opportunities for educators outside of mathematics and computing to
engage with data science in a way that entails low buy in but high payoff.

Barriers

 Teachers would require professional development to learn how to think in terms of

civic statistics, how to use new technologies, and (potentially) how to coordinate the
delivery of civic statistics across and between subjects.

144
Ridgway, J. (ed.) (2022) Statistics for Empowerment and Social Engagement: Teaching Civic
Statistics to Develop Informed Citizens. Springer ISBN: 978-3-031-20748-8

April 2023 pg. 80

  Formal assessment of civic statistics could be complicated by the need to balance
validity and standardisation.
 Incautious development or presentation of civic statistics could make it appear as
an ‘easier’ form of statistics, rather than a demanding approach that draws on
knowledge and sensitivities from multiple disciplines.
 The contemporary focus of civic statistics would require resources to be constantly
renewed, or for local parties to be trained and empowered to develop their own
datasets and resources.

Adaptations

 Stakeholders could be invited in from multiple levels – not just from within the
education sector but beyond, for instance official statistics providers and the media
could contribute to training, the production, and the updating of resources.
 The website is designed to be user-friendly and accessible for educators from a
wide variety of backgrounds and starting points, mapping content to statistical
topics and tools. This could be developed even further so that subject specialists
could more easily structure pathways for their students that attended to key skills
and approaches from civic statistics.

April 2023 pg. 81

8.9 Bootstrap integrating computing with other subjects
Bootstrap 145 is a US-based initiative, housed at Brown University, that provides free
curricular materials and professional development for integrating Computer Science and
Data Science into mathematics, science, business and social studies classes for students
from US grades 5 to 12 (expected ages 10 to 19). The programme has expanded significantly
since its origin: it has received multiple grants from charitable and corporate organisations,
and is now well established in the US, such that in the academic year 2021/22, it reached
nearly 30,000 students. Further, according to their website, “nearly 50% of our students are
African-American or Latin(x), and nearly 45% are girls and young women.” 146 The Bootstrap
website foregrounds three principles: equity, scale and rigour. 147

Bootstrap offers an example of how a programme of study can be developed explicitly

around the intersection of computing and mathematics (and later computing and data
science). It is also notable for promoting a model of provision which does not rely heavily on
specialist computer science educators; according to their 2015 evaluation, approximately
three quarters of the teachers did not have a computer science degree, and the same
fraction did not have computer science as their main teaching assignment. 148

Bootstrap:Algebra

Bootstrap:Algebra was originally devised as an after-school programme, but has since

evolved into a nationally recognised curricular resource which is suitable for in-class use.
The current version is promoted as a 25-hour course where students get to create a video
game, whilst working with mathematical content and computer science concepts. 149 Game
design is seen as an engaging starting point from which students can connect to key ideas
from mathematics and programming.

The Bootstrap curriculum is underpinned by several pedagogic and design principles

intended to support students in making connections and engaging with cross-curricular
work. Notably, Bootstrap:Algebra uses the functional programming language Racket, which
treats variables and functions in a similar way to mathematics. The intent is to smooth the

145
This case study is based on content and research papers available via the Bootstrap website
https://www.bootstrapworld.org/ . Many of the research papers, on the development and trial of
various phases of Bootstrap have Kathi Fisler and Shriram Krishnamurthi, computer science
professors at Brown University and co-founders of Bootstrap, as co-authors.
146
https://www.bootstrapworld.org/index.shtml
147
https://www.bootstrapworld.org/
148
https://www.bootstrapworld.org/impact/EvalReportDec2015Medium.pdf (page 6)
149
https://www.bootstrapworld.org/materials/algebra/Bootstrap-Algebra.pdf

April 2023 pg. 82

connection between algebraic thinking and programming abstractions, anticipating that
“functional languages reinforce – rather than undermine – the math that students already
need to learn in school.” 150 151 Once students have built their initial understanding, they can
move on to other languages in other Bootstrap modules (see below).

The Bootstrap curriculum is also designed to be flexible and helpful to teachers: students do
not need to spend all 25 hours in a computer lab, the materials are designed in a modular
way to aid organisation and delivery. Work is aligned to US State curricula and standards to
encourage teacher buy-in, and professional development and the programme also runs
workshops and professional development. This flexibility extends to assessment: using
Bootstrap does not necessitate any particular model of assessment, allowing teachers to
attend to the demands of their own context.

Expansion and Bootstrap: Data Science

The Bootstrap provision has been extended to include further modules which integrate
computing into science lessons (Bootstrap:Physics) and social science lessons
(Bootstrap:Data Science). There is also a fourth course, Bootstrap:Reactive, which focuses
on developing more advanced programming skills.

The materials are presented in a highly flexible way, such that a teacher can use them to
support anything from a one-week to a one-year course. 152

Bootstrap:Data Science is a relatively new module which, like Bootstrap:Algebra, is designed

for embedded or integrated delivery. The course authors have detailed the thinking behind
their curriculum design in some detail 153, proposing that a “strong and equitable data
science curriculum combines four ingredients”. These are listed as: domains of study,
statistics and mathematics, computing and civic responsibility. Instead of developing a
computer game, students combine data science content (such as linear regression) with
computing content (such as Boolean logic) to pose and answer questions; the course

150
https://www.bootstrapworld.org/blog/software/FunctionalLanguages.shtml
151
The effectiveness of this approach is examined in Wright, G., Rich, P. & Lee, R. (2013). The
Influence of Teaching Programming on Learning Mathematics. In R. McBride & M. Searson
(Eds.), Proceedings of SITE 2013--Society for Information Technology & Teacher Education
International Conference (pp. 4612-4615). New Orleans, Louisiana, United States: Association for
the Advancement of Computing in Education (AACE). Retrieved April 5, 2023
from https://www.learntechlib.org/primary/p/48851/.
152
https://www.bootstrapworld.org/materials/data-science/
153
https://cs.brown.edu/~sk/Publications/Papers/Published/spddplfk-integ-ds-desn-assm-
curric/paper.pdf

April 2023 pg. 83

materials provide 26 large data sets drawing on contexts such as politics, sport and the
environment. Some aspects of data science, such as data cleaning and working with newer
forms of data visualisation are not included in the course due to space constraints.

Equity, Access and Variability

The Bootstrap materials appear to be used widely and with thousands of students across
the United States. It offers a proof of concept for integrating some elements of computer
science into mathematics (and other such integrations) in a way that supports both
domains. Further, the embedded design approach may be helping to promote equity and
support access by bypassing bottleneck issues associated with a deficit of specialist
computer science teachers. Nonetheless, it is relevant to note that the small sample sizes in
the main evaluation document (where the participants could skew towards enthusiastic
adopters) 154 make it hard to quantify some aspects of its success or reflect meaningfully
on how well similar work might translate to the UK context.

Benefits

 The courses have been developed to appeal to and support teachers, particularly
those with limited computer science experience. This has had benefits in terms of
equity and access, and positively impacted the scalability of the programme.
 The Bootstrap materials seek to integrate knowledge and approaches from each
involved discipline in an authentic way which has mutual benefit. This has extended
to the combined pedagogy, for instance when error messages while programming
are seen as opportunities for a student to test their understanding of abstract
structure, rather than simply as mistakes or typos. 155
 Meaningful and accessible large data sets connect statistics and programming with
domains of interest and with civic responsibility.

Barriers

 The goal of establishing clearer links between subject areas could limit the included
content from each discipline and impose design choices (for instance, some
computer scientists might prefer programming languages which were more widely
used over functional programming options).
 Some presentations of the courses rely on a community of volunteers (college
students, specialist teachers, and software professionals) which cannot be assumed.

154
https://www.bootstrapworld.org/impact/EvalReportDec2015Medium.pdf
155
https://bootstrapworld.org/blog/software/Errors.shtml

April 2023 pg. 84

  The flexibility built into the Bootstrap model might not transfer comfortably to a
context with more constraints concerning curriculum and assessment.

Adaptations

 As more school districts and teachers are coming on board, Bootstrap can increase
their pool of ‘ambassadors’ and professional development offerings.
 There is an increased focus on evaluation and iterative improvement of the
materials.
 The Bootstrap team have been developing adaptations for students with visual
impairments or other accessibility issues.

April 2023 pg. 85

8.10 Urban Data School
Data literacy is often taught in schools using small, personally collected data which is
easier for students to relate to. This does not give the pupils the experience of critically
querying and deriving insights from larger, complex data that is collected at scale.

The Urban Data School study trialled a method for teaching from complex data, collected
through a smart city project. It showed that existing data principles from the literature can
be adapted to design data literacy activities that help pupils understand complex data
collected by others and form interesting questions and hypotheses about it.

The study was carried out over two years in three primary and secondary schools in Milton
Keynes, using datasets from the MK:smart project. The lesson plans were co-created with
three teachers (maths, geography and triple science) and used with 67 students aged
between 10–14 years, across 4 separate sessions, each of 2-3 lessons over a few weeks. It is
a small-scale, longitudinal example of our Model 3, adding content across subjects).

Data literacy activity design principles

The study authors synthesised from the literature the following principles for designing data
literacy activities.

P1 Inquiry principle: Lead the students first in a guided inquiry, followed by an open inquiry
when they are more familiar with the data and the approach.

P2 Expansion principle: Start from a representative small part of the data set and expand
out, rather than starting with the full, large data set and focusing in. This helps students
orient themselves within the data, before navigating across it.

P3 Context principle: Teach in a context the student understands, using data from their
own environment or somehow related to them.

P4 Foundational competences principle: Focus on developing foundational competencies

rather than practical skills, for example how to ask ‘good’ scientific questions from data.

P5 STEAM principle: Take a STEAM approach by working collaboratively on creative

activities alongside practical ones.

P6 Personal data collection principle: Students should engage with data they collected
themselves. When analysing an external data set, they should be given additional activities

April 2023 pg. 86

that help them understand what it is like to collect that type of data. This is to support them
in contextualising and interpreting the external data.

The main change to existing literature is in P6: personal data collection can complement
(rather than replace) interpretation of existing data.

Note that the principle ‘P4 Foundational competences principle: Focus on developing
foundational competencies rather than practical skills’ is not demoting the role of skills, but
is intended to clarify that the skills build on and need the foundational competencies. All
students should obtain the foundational competences of interpreting, critiquing and asking
questions of data. This is especially important if the data set is not available, as in adverts
or media reports. Data-literate citizens with a critical mind will more easily spot fake news,
unwarranted conclusions and other ways of manipulating public opinion or prospective
customers with data. The ability to ask relevant and feasible questions builds the
foundation for the choice of the data cleaning, analysis and presentation methods and
tools. Without the foundational competences, the practical skills (like applying statistical
and visualisation methods) may be misguided and thus contribute to misinformation and
erroneous conclusions.

Activities

The three main data literacy activities are described next. Their goal was to develop the
students’ skills for asking questions that they were personally interested in from the data.
Therefore, all activities followed principles P1 and P4. Any additional principles followed are
given in the individual activity descriptions. The first two activities were each done with and
without computers, in different sessions.

Smart meter energy data: The students were shown anonymised data about whole house
consumption, individual appliance consumption and generation of solar energy from ~70
houses in Milton Keynes (P3). Students were first given data about a single household over
a single day and then for more households and over a time period (P2). Students were
shown data on paper and using an interactive app.

Potential for solar energy production: The students were shown aerial photographs and an
interactive Google-like aerial map of Milton Keynes, including of their school and
neighbourhood (P3), with information about the potential annual energy yield if solar
panels were installed on roofs, based on their direction, pitch and size.

Be a LiDAR device: This lesson followed the previous lesson to explain the LiDAR technology
used to obtain the rooftop dataset. Students worked in groups to build plasticine houses

April 2023 pg. 87

(P5) and measure their height for each square centimetre (P6). Students swapped their
data to figure out the shape of the other group’s house from the data alone.

The activities drew on mathematical and digital literacies to help pupils better understand
an externally sourced complex dataset. Doing their own measurements and using
interactive visualisations led pupils to more critical questions and insights, like whether the
solar panel yield takes into account nearby obstacles (e.g. tall trees) and if smart metres
are always measuring correctly. The design principles were effective in helping students
make sense of the data and of the wider topic (energy consumption and generation).

Benefits

 Interactive data visualisations can support school children in asking valid questions
from data.
 Snapshots of the data can help school children understand how to frame questions
around the extended dataset, along dimensions such as space and time.
 The learning of data skills lends itself to cross-curricular learning and can begin with
primary school students.
 When engaging with externally sourced data it can be useful to act in the role of a
data collector to understand better where errors can creep into the data and to
develop better data scepticism.

Barriers

 Even though the teachers co-created and co-delivered the lessons, they didn’t feel
confident in doing such activities by themselves after the project finished.
 Bespoke interactive visualisations may be needed, with simple user interfaces.

Adaptations

 The activities were refined and adapted by each teacher to their classroom and
context.
 While many open data sets exist, they require curation for the local context of pupils.

April 2023 pg. 88

 9. Overview of case studies
Each of the case studies in Section 8 offers a particular perspective on how mathematical
and data literacy education - or at least certain aspects of it - might be conceptualised,
organised, and delivered. Although each account is necessarily a product of its own
context, it offers an example of what is possible, and collectively the studies point towards
several barriers and adaptations which are relevant to this analysis. The following
discussion collects some comparisons and commonalities which speak to form and
feasibility.

Through the discussion, we evaluate the effect of each curriculum model on the following
criteria. Judgements are qualitative, based on our analysis of the case studies and the
lessons learnt. The numerical scales are included to facilitate overall comparisons, and are
not the result of any systematic quantitative process.

 Benefits for mathematical and data literacy: will the model allow the competencies
to be achieved for all/many/specialist students, with the corresponding societal,
personal and employment benefits? Scale +10 (allow) to -10 (prevent).

 Equitable access: will the model affect gaps in attainment or participation in STEM
due to gender, social class or ethnicity? Scale +1 (reduce) to -1 (widen).

 Assessment: how extensive are the changes to assessment structure as compared

to current provision? Scale 0 (none) to -2

 Teacher professional development: what extent of professional development is

needed and how many teachers are affected? (It is relevant here that 6,500
secondary teachers taught some computing in 2021, compared to 16,500 for
geography, 36,000 for mathematics, 50,000 for science and 222,000 primary and
nursery teachers 156.) Scale 0 (none) to -5 (highest need)

 Teacher workforce: will this model result in secondary teachers teaching outside
their specialist subject or need a non-trivial increase in the teaching workforce?
Scale 0 (no effect) to -2

156
DfE, School workforce in England 2021. https://explore-education-
statistics.service.gov.uk/find-statistics/school-workforce-in-england#dataBlock-c26f37c7-
2bca-4019-a55d-a26f58713754-tables

April 2023 pg. 89

9.1 The case for change
The case studies repeatedly speak to the import and relevance of data-driven
mathematics. Amongst other features, the advancing place of statistical/data knowledge
and skills in the workforce is recognised within MEI’s courses for students and the work with
geospatial data in geography; the team behind Civic Statistics make a strong argument for
data science as a critical component of contemporary citizenship; and the developing
curricula and qualifications in Scotland, New Zealand and the IB offer examples of how
statistics/ data science is already part of the international educational landscape. Together
the cases make multiple, intersecting arguments for the deliberate inclusion of
mathematical and data literacy in secondary education, and against keeping the current
curriculum structure with no changes.

Failing to make the change is negative rather than neutral in its outcome because the
demands of employers for T-shaped, broad and connected, knowledge 157 are already not
met under the siloed nature of our current system. Moreover there are pedagogic
implications in teaching mathematics content that is increasingly distant from its use. As
technology increasingly disguises societal practices such as making change or using
paper maps, these are not available as experiences which students can mathematise158.
Instead, informal, out-of-school discussions become critical in making connections
between classroom mathematics and its actual use. This has effects on equity, as it has
been shown that such experiences are not uniformly distributed and affected by social
class and linguistic capital 159. In addition, teaching mathematics for meaning-making and
for connections to realistic uses of mathematics supports the participation and
engagement all students, but especially girls 160. Finally, there is an ongoing PD cost for new

157
Kispeter, ‘What Digital Skills Do Adults Need to Succeed in the Workplace Now and in the next
10 Years?’
158
Jenefer Golding, ‘Mathematics Education and Financial Literacy: A Valued but Vulnerable
Symbiosis’, in Proceedings of the British Society for Research into Learning Mathematics 40 (1)
March 2020, ed. Rachel Marks, 2020, http://www.bsrlm.org.uk/publications/proceedings-of-day-
conference/ip40-1/.
159
Robyn Zevenbergen, ‘Mathematics, Social Class, and Linguistic Capital: An Analysis of
Mathematics Classroom Interactions’, in Sociocultural Research on Mathematics Education, ed.
Atweh, Bill, Helen Forgasz, and Ben Nebres (Mahwah: Lawrence Erlbaum Associates, 2001), 201–15.
160
Cathy Smith and Jennie Golding, ‘Schools’ Strategies for Promoting Girls’ Participation in
Mathematics’, in Proceedings Of the 42nd Conference of the International Group for the
Psychology of Mathematics Education (Vol 4) Umeå, Sweden., ed. Ewa Bergqvist et al., 2018, 211–
18; Cathy Smith, ‘Gender and Participation in Mathematics and Further Mathematics A Levels: A
Literature Review for the Further Mathematics Support Programme’, 2014,
https://core.ac.uk/download/pdf/79498409.pdf.

April 2023 pg. 90

teachers who must understand, justify and motivate applying mathematics in inauthentic,
outdated contexts in order to teach the curriculum.

Model 1 No change

Benefits for mathematical and data literacy -5

Equity -1

Teacher Professional Development (Maths teachers only) 0

No effect on Assessment or Teacher Workforce 0

9.2 Promoting data skills through computing and across

subjects
We have found no examples of data skills being added to computing curricula. Rather the
opposite, since coding has recently been added to primary and early secondary
mathematics curricula in Ontario and also in Sweden 161. Several case studies take a
proactive role in demonstrating the role of technology within mathematical and data
literacy, including the Scottish NPA model and ProDaBi. Both projects aim for flexible models
of delivery but have been crucially supported by computing teachers. As such they can
inform Model 2, that is retaining the current structure for KS 3 and KS 4 with content added
to Computing. ProDaBi was particularly notable for aiming to allow upper-primary aged
students to explore and reflect upon some basic elements of machine learning through
activity and analogy. Scotland’s National Professional Awards in data science sit against a
background of related academic and vocational qualifications for over-15s.

The fit with computer science does have problems. GCSE Computer Science is optional so
adding to that content would not deliver the benefits for all, or even many. Through our
reading and conversations, we encountered different opinions about whether students
should be provided with software that automated key processes or (eventually) be
required to develop their coding skills. Even without programming, any expansion of content
in Computing as a foundation subject would require more time in KS 3 and, problematically,
at KS 4 where the existing demands are already enacted inconsistently, in part because of

161
C. Kilhamn et al., ‘Variables in Early Algebra: Exploring Didactic Potentials in Programming
Activities.’, ZDM - Mathematics Education 54, no. 6 (November 2022): 1273–88; Susanne
Kjällander et al., ‘Elementary Students’ First Approach to Computational Thinking and
Programming.’, Education Sciences 11, no. 2 (February 2021): 80.

April 2023 pg. 91

the low profile of non-assessed subjects. Thus this model without any change to
assessment and curriculum structure would not achieve the full aims.

The ProDaBi example suggests that computing teachers may well require less extensive PD
than teachers of other subjects to start to teach data skills, and there are fewer of them.
However there is already a shortage of qualified computing teachers, with only 53% of
lessons being taught by a teacher with a relevant post A level qualification 149; extending the
subject time would further affect supply.

Teaching data science with a heavy emphasis on computing also invokes issues of equity
and access. Not all students have equal access to technology or previous experience with
technology, such that an overreliance on computing as an entry point could form a barrier
to some students engaging with other, less technical, data skills. The MEI case study shows
that positioning data science closer to mathematics and statistics than computing can
reduce gender gaps in participation, with 37% of their students being female 162 compared
to 15% of A level computing candidates in summer 2022 163.

Model 2 Current Structure for KS 3 and KS 4 with content added to Computing

Benefits for mathematical and data literacy +4

Equity -1

Teacher Professional Development (Computer teachers only) -0.5

Assessment 0

Teacher Workforce -2

The example of ProDaBi also touches on some wider issues. First, it suggests the value of
teaching data science over time; the higher level ProDaBi project is quite ambitious, but it
relies on previous exposure. A supporting comparison can be drawn with the New Zealand
curriculum where (amongst other skills) students develop their statistical literacy from
primary school, asking more sophisticated questions and gradually developing higher-
order skills of critique and evaluation. Scotland is likewise working to develop the teaching
of data science in primary school. Introducing and developing data science sensitivities
over time advances both the potential depth and width of provision within later Key Stages.

162
https://mei.org.uk/app/uploads/2021/08/mei-ds-short-course-pilot-report.pdf
163
https://www.jcq.org.uk/wp-content/uploads/2022/08/A-and-AS-Results-Summer-2022-
Compared-to-2021.pdf

April 2023 pg. 92

The Bootstrap programme from the United States offers something of a counterpoint;
through careful curriculum and resource design it aspires to empower non-specialists to
teach computer science in conjunction with their own disciplines, including mathematics,
physics and social science, thus working towards its goals of equity, scale and rigour. The
provision is for c. 25-hour courses, with associated teacher PD, that can be dropped into
existing curricula flexibly as suits teachers or school boards. This is enticing, and it speaks to
the import and power of proper teacher support, but it is unclear how well this approach
might generalise to more extensive slices of subject curricula, or to the contexts such as
England where curriculum time and assessment are more standardised.

Mathematical and data literacy’s central activity of posing and answering questions in
context naturally suggests the potential for cross-curricular work (which is an advantage of
the Intersectional framework proposed in Section 5 of this paper). The merits of this
approach are recognised not only in ProDaBi, Scotland’s NPA and Bootstrap case studies
but also in Civic Statistics, Urban Data School and the PD collaboration between Geography
and Core Maths. The approach laid out in the Civic Statistics case study foregrounds
citizenship elements and data literacy with a global perspective. ‘Drop-in’ materials for
teachers cover a wide range of topics such as income inequality, migration, and ageing
societies that appeal to many stakeholders and allow informed discussion of equality,
diversity and inclusion issues. The Urban Data School, aimed at primary/ lower secondary
students, captured similar principles of engaging students through inquiry into data from
their own environment and included experiences of collecting such data though smart city
technologies.

While there is certainly an opportunist element of innovative education projects taking any
available teaching space, this consensus suggests that adding content across subjects
provides a viable way of achieving the benefits of mathematical and data literacy. The
change also affects the teacher workforce equally, with no specific demands on shortage
subject teachers. It is particularly welcomed as a way of introducing mathematical and
data literacy to those with mathematics anxiety, and of broadening its appeal to girls
through enhancing meaning-making 160. Moreover, the example of Geography shows that A
level teachers in England have, over time, and when incited by relevant drivers such as
assessment, taken advantage of professional development opportunities to develop
students’ mathematical, statistical and technological skills that are authentically useful in
their own subject.

KS 3 and 4 subjects (apart from mathematics) that were repeatedly suggested as

particularly suitable for enhanced data skills were sciences, geography and religious
education. At KS 5 this is broadened to include psychology, sociology and politics. We have

April 2023 pg. 93

seen no trials of embedding in the arts or languages. This does expose a weakness of any
model that embeds mathematical and data literacy across several subjects but retains
current free subject choice at 14 or 16 (such as Model 6): there is no guarantee which
combination of subjects individuals will study, and thus there would be no mechanism for
ensuring all had met the desired competencies, reducing the equity advantage. Further
investigation is needed to understand how/whether data literacy could be embedded in a
wider range of subjects. In addition, changes to assessment would be needed, since the
pressure of high-stakes examinations means that teachers would be likely to omit any
non-assessed content in the subjects chosen.

Model 6 Current KS 5 Structure with content added across subjects

Benefits for mathematical and data literacy +7

Equity +0.5

Teacher Professional Development (in Humanities, Science + ?) -2.5

Assessment -0.5

Teacher Workforce 0

Model 3, of enhancing some subjects at KS 3 and 4 (while leaving KS 5 unchanged) raises

further issues of progression. The case studies suggest that compromises are inevitable –
even desirable – in developing resources that promote competencies across domains.
Students’ experiences in Geography, for example, would not be the same as those in RE. A
potentially uneven pattern of experience at KS 4 would not provide a reliable foundation for
specialists to build their skills during KS 5. (A similar effect was noted in the Geography case
study, where changes to KS 3 computing mean that GCSE teachers cannot rely on students’
spreadsheet skills.) Reteaching could be needed, reducing curriculum time, or this could in
effect push existing inequities arising from subject choice at 16 down to age 14.

Model 3 Current Structure for KS 3 and KS 4 with content added across subjects

Benefits for mathematical and data literacy +8

Equity +0.5

Teacher Professional Development (in Humanities; Science + ?) -3

Assessment -0.5

Teacher Workforce 0

April 2023 pg. 94

Two issues which came up in conversation with experts during the writing of the case
studies was how teachers in other subjects might be supported in bringing data science
into their lessons, and how cross-curricular approaches to mathematical and data literacy
might be aligned when subjects are siloed. One suggestion was that schools might
designate a ‘data science coordinator’, although it was also noted that previous efforts to
have ‘statistics coordinators’ in UK schools had not been fruitful.

9.3 Introducing strands within mathematics

Several case studies site mathematical and data literacy within the mathematics
curriculum. Among these, Ontario is unusual in that their approach to teaching or assessing
data-driven problems is not marked or ring-fenced. In Ontario, most content strands are
roughly balanced between expectations of developing a new mathematical skill and those
of applying it in domain contexts, many of which are specified in the curriculum documents.
Thus (although it extends to KS 5) it speaks to Model 4a of adding content to the
compulsory mathematics curriculum at KS 3 and Foundation level GCSE. In England’s
current framework this is the only way of ensuring that students meet the competencies for
all. With care, Higher level GCSE could be tailored to address the competencies for many.
Without further change at KS 5, this model would not embed the specialist skills.

One (flawed) argument for Model 4a is that it involves minimal disruption: teaching with
attention to data and applications is not precluded in the current curriculum, indeed is
already some teachers’ practice. Schools have moved towards spending two (rather than
3) years on KS 3, suggesting some space in that KS. However, this ignores the way that
high-stakes assessment drives teachers to frame students’ opportunities to learn 164. GCSE
Mathematics questions are set in a mathematical context 165 or in a simple context
involving units specified in the curriculum (e.g. money, time, length, mass, litres, speed,
density), and teaching is largely restricted to those. Embedding this extra content would
involve teacher PD and increased curriculum time, with a subsequent effect on workforce
capacity.

164
ACME, ‘Problem Solving in Mathematics: Realising the Vision through Better Assessment’, 2016,
https://royalsociety.org/~/media/policy/Publications/2016/problem-solving-in-mathematics-
06-2016.pdf.
165
Candia Morgan and Sarah Tang, ‘To What Extent Are Students Expected to Participate in
Specialised Mathematical Discourse? Change over Time in School Mathematics in England’,
Research in Mathematics Education 18, no. 2 (3 May 2016): 142–64,
https://doi.org/10.1080/14794802.2016.1174145.

April 2023 pg. 95

A further consideration from Ontario is that assessment of students is done by teachers,
following a national reporting system. Assessment is able to follow what is taught, rather
than drive it. 70% of the assessment must occur during the course and 30% at the end,
drawing on teacher-designed tests and tasks. Provincial tests at grades 3, 6 and 9
separately monitor school performance against expectations. This system leaves a lot of
responsibility to teachers in a very respected education ecosystem. The 10-year curriculum
design cycle acts to influence the enacted curriculum through PD and through some
internal structure; for example the 2020 move to designate modelling as a ‘strand’ in
algebra is to ensure that it is assessed.

Model 4a Current Structure for KS 3 and KS 4 with content added to a Mathematics

course

Benefits for mathematical and data literacy +5

Equity 0

Teacher Professional Development (in Maths) -2

Assessment -0.5

Teacher Workforce -0.5

The New Zealand case study takes a stronger approach of ensuring attention to data
literacy within mathematics by creating curriculum strands (or pillars). It draws a
distinction between equally-weighted ‘Mathematics’ and ‘Statistics’ strands both of which
must be taught and assessed throughout the primary and secondary curricula until the
final, optional pre-university year of school when students may choose one. New Zealand is
thus similar to Model 4b in which assessment is extended to two GCSEs (although NZ
strands extend beyond KS 4). Previous arguments for a two-pillar or two-subject approach
to teaching mathematics have often separated applications of mathematics from
developing new mathematical tools (e.g. Use of Maths 166 or Essential Maths 167). In the case
of New Zealand, both strands involve applications to real-word contexts. There was
considerable PD involved in educating mathematics teachers about pedagogy for
statistics, and this was supported by extensive pre-trialling of curriculum materials and
sample tasks. Much assessment is teacher-designed with some post-16 students able to

166
Paul Hernandez-Martinez et al., ‘Students’ Views on Their Transition from School to College
Mathematics: Rethinking “Transition” as an Issue of Identity’, Research in Mathematics Education
13, no. 2 (2011): 119–30, https://doi.org/10.1080/14794802.2011.585824; Andy Noyes et al., ‘Evaluating
Mathematics Pathways’ (Department of Education, 2011).
167
MEI, ‘Curriculum and Assessment in Mathematics in England at KS4 and KS5 - Position Paper’.

April 2023 pg. 96

choose between studying ‘standards’ (of which three must involve numeracy) that have
external assessments and those that are internally moderated.

This approach has been successful in increasing fidelity to the intended curriculum, and a
recent expert panel review 168 has suggested it be retained, indicating its strength. There are
concerns about inequitable attainment patterns in New Zealand, but the same review
considers those to result from school differences and not to curriculum structure.

Again, without further change at KS 5, this model’s benefits are limited since it would not
embed the specialist skills.

Model 4b Additional content at KS 3 and 4 creates two parallel, or sequential,

Mathematics GCSE courses

Benefits for mathematical and data literacy +8

Equity +0.5

Teacher Professional Development (in Maths) -2

Assessment -1.5

Teacher Workforce -1

9.4 Introducing pathways within mathematics

The approach taken at KS 5 in several of the case studies is to introduce pathways at 16-18.
New Zealand and Ontario both set a minimum requirement of mathematical credits to
obtains school graduation certificates and provide a range of courses to suit students’
interests and needs. The International Baccalaureate (IB) requires students to choose
between two pathways through mathematics, which can loosely be thought of as ‘pure’
and ‘applied’ mathematics routes. Their overall approach is seen as effective in providing
benefits for all, through the minimum requirements, and in allowing specialisation.

In Ontario, the pathways are described in terms of preparing for school leavers’ next
destinations: the workplace, college or university. Within college preparation, a distinction is
made between technology-related programmes and others. Within university preparation,
the distinction is between degree majors needing mainly statistical data management

168
Royal Society Te Apārangi Expert Panel, ‘Pāngarau Mathematics and Tauanga Statistics in
Aotearoa New Zealand. Advice on Refreshing the English-Medium Mathematics and Statistics
Learning Area of the New Zealand Curriculum’ (Royal Society Te Apārangi, 2021).

April 2023 pg. 97

such as social sciences (which has been popular among girls 169); those needing
knowledge of functions such as business, health science and those also needing calculus
such as STEM or economics. Online courses are provided to mitigate problems of accessing
teaching. Pathways are overlapping, but it is recognised that early choice has equity issues
and the entry point has recently been moved up by de-streaming grade 9.

In New Zealand, students have extensive choice in selecting standards to study, which may
in practice be restricted by schools. The present curriculum supports higher level study,
encouraging and preparing large numbers of students to choose a first-year statistics
course at university. There is an accompanying concern that the emphasis on statistics
may distract potential STEM undergraduates from specialising in calculus standards.

The two-pillar approach within IB mathematics has been well received by many and has
turned out to be a useful way of grouping content for many diploma programme and
career programme students. Students choose from two strands of mathematics, known as
Analysis and Approaches (AA) and Applications and Interpretation (AI), each of which is
offered at either SL or HL. There is core common content (including some calculus) ensuring
that all four forms of IB Mathematics meet certain standards allowing university-level study.
The AI versions of the curriculum could resonate with the Intersectional competencies
suggested for all and for many school leavers (with little, however from the computational
toolkits). All four courses include a non-examined but assessed project based on asking
and answering questions.

These case studies show that pathways are an effective structure for ensuring an
appropriate range of mathematical and data literacy competencies, and also that they
have the flexibility to accommodate regular updates as mathematical needs evolve,
speaking to Models 7 and 8.

For Model 7, adapting the current A/AS level and Core Maths structure, the issue recurs that
there is no minimum mathematics requirement, and indeed there is currently little
enthusiasm for compulsory Core Maths 170. However, the case studies reveal no dissent
once established as part of a baccalaureate system with minimum requirements for
numeracy and literacy (as in the IB and as being developed pre-16 for the Welsh

169
Jennifer Hall, ‘Gender Issues in Mathematics: An Ontario Perspective’, Journal of Teaching and
Learning 8, no. 1 (5 January 2012), https://doi.org/10.22329/jtl.v8i1.3004.
170
Andrew Noyes and Michael Adkins, ‘Studying Advanced Mathematics in England: Findings
from a Survey of Student Choices and Attitudes’, Research in Mathematics Education 18, no. 3 (1
September 2016): 231–48, https://doi.org/10.1080/14794802.2016.1188139.

April 2023 pg. 98

Baccalaureate 171). The Core Maths structure does allow further specialisation that could
include the statistics/data science and computation skills that are not currently included. In
addition, increasing the numbers studying mathematics would have a large impact on
teacher workforce. Relevantly, the RGS/Core Maths case study has started to develop non-
mathematics teachers towards teaching current Core Maths although uptake is not known.
The examination-based assessment remains problematic, given that all relevant case
studies found the need for project-work to make valid assessments.

Model 7 Current KS 5 Structure with additional content creating choice of mathematics

pathways (examined as A/AS level or Core Maths)

Benefits for mathematical and data literacy +7

Equity 0

Teacher Professional Development (Maths + others) -2

Assessment -0.5

Teacher Workforce -0.5

Changing the curriculum so as to make not only a space, but an entitlement, for developing
mathematical and data literacy is ambitious. Models 1-7 have all accepted the constraints
of the current system where, after KS 3, students can choose not to study some subjects.
The rarity of this has been commented on in mathematics 172, but applies even more to
subjects where data-driven questions might be asked and answered. This feature
repeatedly means that England cannot follow other national curricula in planning minimum
cross-curricular entitlements for all.

It is beyond the scope of this paper to investigate different baccalaureate options. Some,
such as New Zealand, have an implicit baccalaureate where students are required to take a
range of options with certain limiting requirements but a great deal of choice; others, such
as the International Baccalaureate are more formal, based around a template where
students choose within subject areas and meet some synoptic requirements. It is also
worth recalling once more the new Curriculum for Wales where a coherent and broad
approach to education is established throughout secondary school. Each of these options
would afford the benefits of embedding mathematical and data literacy, either through a

171
https://hwb.gov.wales/curriculum-for-wales
Jeremy Hodgen, Is the UK an Outlier?: An International Comparison of Upper Secondary
172

Mathematics Education (London: Nuffield Foundation, 2010).

April 2023 pg. 99

coordinated, subject-based approach, or via a synoptic element. This is discussed further
in the next section.

Model 8 A new baccalaureate system at KS 5 and below (requiring study in core

subjects)

Benefits for mathematical and data literacy +10

Equity +1

Teacher Professional Development -2

Assessment -2

Teacher Workforce -1

These case studies indicate that any move to a pillars pathways system would need to be
thought through carefully, balancing foci and attending judiciously to the needs of
students, higher education, and employers. Any partition of mathematics will inevitably be
imposed and imperfect, and it is entirely possible that incautious labelling may lead to
content being neglected or even lost; the New Zealand case study reminds us for example
that statistical modelling is distinct from, and does not supersede, mathematical modelling.
The IB case further warns that when multiple versions of mathematics are on offer at the
same level, it is natural for some stakeholders to perceive one as ‘easier’ than the other, and
without handling and pre-emption this will affect provision and uptake.

9.5 Pitfalls and possibilities of mathematical and data

literacy as a separate curriculum area
At first glance, two of the case studies successfully deliver data science content as an
adjunct to other curricular provision. The MEI courses for A level and Core Mathematics
students can be considered as enrichment, and the Scottish NPA in Data Science has been
established independently of any other qualifications. Arguably, this approach has led to
fewer restrictions on the content of these courses and awards them greater freedom to
develop and evolve over time.

However, both cases implicate some significant feasibility issues surrounding scaling,
equity and access. One of the strengths of the MEI courses is the presence of specialists
with a deep knowledge of, and appreciation for data science; this would be hard to scale
up at present. Equally, whilst the Scottish NPA offers a considered and detailed approach to
teaching data science, it exists as one of many possible qualifications, and relies on the

April 2023 pg. 100

presence of willing and capable staff for in-school delivery. If data science is positioned
only as a separate and specialised discipline, it is likely that current workforce shortages
and individuals’ curriculum choices will mean that fewer students will have the opportunity
to develop data science competencies.

A related but contrasting approach would be to introduce data science via a project-
based component that has its own presence within the curriculum but is explicitly taught
and positioned with reference to other curriculum areas. This model is not precisely
exemplified within the case studies, but it can be illuminated by some comparisons with
other examples of project-based qualification and interdisciplinary study.

One of the most common project-based qualifications in the UK at the moment is the
Extended Project Qualification and a Level 3 data science project could exist in a similar
way, requiring students to use their independent study and research skills to carry out some
data-focused work for themselves. However, whilst such an approach might capture well
the investigative nature of data science, an optional qualification would not reach all
students, and it would rely on a level of knowledge of and experience with data science that
Key Stage 5 students do not typically possess at present. There are also practical concerns
surrounding independent work: MEI, for example MEI, note the difficulty for experienced
teachers of finding large open access data sets suitable for 17-year olds.

An alternative approach would be to combine a project-based output with taught lessons,

and to make it compulsory for all students. This is resonant with the delivery of Theory of
Knowledge (TOK) within the International Baccalaureate diploma programme. Students
have specific TOK lessons where they meet key ideas and are assessed via an exhibition
and a short essay. There is also an expectation that subject teachers support the delivery of
TOK by making explicit connections to ideas from the course. TOK plays a meaningful but
bounded role within the wider diploma; compared to the main components it involves
fewer curriculum hours and has fewer points attached to it. Teachers typically deliver TOK in
addition to their regular subject specialism, reducing workload effects.

Model 5 Current KS 5 structure with the addition of credit-bearing project-based work.

Benefits for mathematical and data literacy +7.5

Equity +0.5

Teacher Professional Development -2

Assessment -0.5

Teacher Workforce -0.5

April 2023 pg. 101

It is possible to map mathematical and data literacy into this model: for instance, students
could meet new statistics/data science or computational content in two focused lessons
each week, be supported in making connections with their other areas of study, and then
complete an individual project which could play to their strengths and interests. This could
be done discretely, or as a required component within a baccalaureate model, possibly
even as a smaller component of a wider citizenship requirement. A flexible, considered and
responsive form of this model could well serve many students on multiple trajectories, and
the potential benefits for developing mathematical and data literacy are substantial. Still,
quality delivery of such an approach would require teacher professional development, (just
as TOK teachers benefit from focused training and support,) and the quality and depth of
the output would be enhanced further if it was supported by the integration of data science
content and sensitivities into earlier key stages.

9.6 The challenge and potential of assessment

One element of provision that runs through the case studies and cuts across the suggested
models is assessment. Formal assessment practices not only codify expectations and
standards, but also have the potential to shape teachers’ impressions of what is truly
important within subject areas, and to steer classroom pedagogy.

Two of the main issues to consider in the context of this report are what place teacher
assessment might have, and the benefits and costs of project work as compared to
examinations.

The case studies illustrate an extensive range of assessment models, from 100% teacher
assessed work (Ontario) to a fully centralised assessment system, albeit on a small scale
(MEI). A mixed approach seems to be particularly common (for instance New Zealand) with
elements of support and moderation built in as considered necessary (as with the IB). It is
of note that the Scottish NPA allows teachers to be involved in writing assessments as well
as marking them; this presents an opportunity for teachers to attend to their contexts and
perhaps address diversity and inclusion issues.

The majority of the cases discussed also appeared to endorse a balance between project
work and written examinations; indeed, in the case of Ontario assessment could be fully
project-based. It is not difficult to argue that students’ facility with the linked, context-based
and empirical tools of data science might be better tested through a report or project,
rather than a formal written examination, and the work of the IB has the value and validity

April 2023 pg. 102

of adopting this approach across mathematics. However, written examinations can reduce
workload demands on teachers and offer some degree of standardisation, so a balance
might seem optimal. Too many avenues for assessment can complicate matters, as seen
in the case of New Zealand.

There were also differing approaches regarding the award and aggregation of marks; the
MEI short course awards a certificate at either fail, pass, merit or distinction, whilst the IB IA
mathematics project is marked out of 20 and then contributes 20% to the overall mark for
the mathematics component. The appropriateness of different scoring systems depends
on the model of provision they exist within; a pass/fail criterion might suit a civic statistics
project instituted as a compulsory component of a baccalaureate style qualification, but it
would not work well within the current A level system. Nonetheless, how assessment is
graded can impact on its reliability, as well as influencing teacher workload.

9.8 The important role of integrative, focused and ongoing

professional development
Another recurring aspect of the case studies was the importance of supporting teachers in
delivering high quality content. The present shortage of teachers in both mathematics and
computer science impacts heavily on the feasibility of many of the proposed options for
curriculum development. Whilst approaches such as Ontario’s online courses and initiatives
such as the AMSP/RGS training go some way towards ameliorating these deficits, it is
important to recognise that many teachers may not be mathematically confident – or may
even be mathematically anxious – so the scope of cross-curricular training and delivery
may be limited.

Teachers are at the pupil-facing side of any curriculum changes, and can be involved and
supported at various stages of curriculum development. This was observed within the case
studies, for example where:

 Teachers were involved in the creation and testing of resources before the statistics
curriculum was introduced in New Zealand;
 Teachers can attend ongoing professional development and training to address
concerns and shortfalls in their knowledge and previous experience, such as is
provided by the IB or the AMSP/RGS collaboration;
 Teachers have access to online resource collections which exemplify and support
good practice, such as is available for mathematics teaching in Ontario, or for the
Scottish NPA.

April 2023 pg. 103

The importance of timely, high-quality professional development and support for teachers
was also a recurring theme in the accompanying conversations.

9.8 Feasibility overview of models

The preceding discussion has provided qualitative judgements on the feasibility of each
model. Figure facilitates an overall comparison, of the potential of each model for
achieving the benefits of embedding mathematical and data literacy, together with the
effects on equitable access, and the impact on teacher PD, assessment and teacher
workforce capacity.

Model Feasibility Comparison

-8 -4 0 4 8 12

Model 1

Model 2

Model 3

Model 4a

Model 4b

Model 5

Model 6

Model 7

Model 8

Benefits Equitable access Teacher PD Assessment Teacher workforce

1 No change. Current structure in England RS 1

2 Current Structure for KS 3 and KS 4 with content added to Computing RS 2b
3 Current Structure for KS 3 and KS 4 with content added across subjects RS 3 (GCSE)
4a Current Structure for KS 3 and KS 4 with content added to Mathematics RS 2a
4b Additional content at KS 3 and 4 creates two parallel or sequential RS 4 (GCSE)
Mathematics GCSE courses
5 Current KS 5 structure with the addition of credit-bearing project-based work RS 5
e.g. EPQ.
6 Current KS 5 Structure with content added across subjects RS 3 (post-16)
7 Current KS 5 Structure with additional content creating choice of RS 4 (post-16)
mathematics pathways (examined as A/AS level or Core Maths)
8 A new baccalaureate system at KS 5 and below (requiring study in core RS 6
subjects)
Figure 8 Feasibility for models

April 2023 pg. 104

9.9 The value of a joined-up approach
One final recurring theme throughout the case studies and the attendant conversations
was how an approach involving multiple stakeholders and organisations can support
curriculum design and delivery. Sometimes connections were temporary, for instance when
teachers were provided with information about how universities were responding to
changes to the IB Mathematics curriculum, so that they could have informed discussions
with students and parents. Other actions were more permanent, such as in the case of the
NPA in Scotland which have been developed so that they can be delivered either by schools
or further education providers.

The cases of MEI and AMSP/RGS demonstrate how subject associations can support
teachers and enhance student outcomes; the involvement of RGS in particular has a history
which builds on previous, successful collaborative work 173. The cases of New Zealand and
Ontario both suggest an educational ecosystem where policy makers, researchers and
practitioners work together. It may be that the present UK climate, where awarding bodies
compete over examination entries, is less inclined towards collaboration.

Whilst all curriculum areas are developing, this is especially true of data science; neither the
pedagogy nor the content is settled. Cyclical models of curriculum development (such as
the seven-year review of the IB) go some way towards coping with these changes, but a
broad collaborative approach could also help keep data literacy materials up to date, for
instance by inviting media organisations to help develop and update material on civic
statistics. Finally, a joined-up approach seems apt: by engaging and involving stakeholders
and interested parties from across education and beyond, it helps to frame the problem of
mathematics, statistics/data science and computing education against its full domain,
and to carry out important work in an ethical space.

173
Nuffield Foundation, ‘Mathematics in A Level Assessments’ (Nuffield Foundation, 2012),
https://www.nuffieldfoundation.org/wp-content/uploads/2012/04/Mathematics-in-A-level-
assessments-A-report-on-the-mathematical-content-of-A-level-assessments-in-Business-
Studies-Computing-Economics-Geography-Psychology-and-Sociology.pdf.

April 2023 pg. 105

 10. References
ACME. ‘Problem Solving in Mathematics: Realising the Vision through Better Assessment’,
2016. https://royalsociety.org/~/media/policy/Publications/2016/problem-solving-
in-mathematics-06-2016.pdf.
ACME and the Royal Statistical Society (RSS). ‘Embedding Statistics at A Level: A Report on
Statistical Requirements and Assessment across A Level Courses in Biology, Business,
Chemistry, Geography, Psychology and Sociology’, 2015. http://www.acme-
uk.org/media/32719/embeddingstatistics-at-a-level.pd.
AfDSP. ‘Data Science Standards’. Alliance for Data Science Professionals, 2022.
https://alliancefordatascienceprofessionals.co.uk//standards.
Bargagliotti, Anna, Christine Franklin, Pip Arnold, Rob Gould, Sheri Johnson, Leticia Perez, and
Denise Spangler. ‘Pre-K-12 Guidelines for Assessment and Instruction in Statistics
Education (GAISE) Report II’. American Statistical Association and National Council of
Teachers of Mathematics., 2020. https://hdsr.mitpress.mit.edu/pub/cqncbp3l.
Biehler, Rolf, and Yannik Fleischer. ‘Introducing Students to Machine Learning with Decision
Trees Using CODAP and Jupyter Notebooks’. Teaching Statistics 43, no. S1 (July 2021).
https://doi.org/10.1111/test.12279.
Bond, Philip. ‘The Era Of Mathematics: An Independent Review of Knowledge Exchange in the
Mathematical Sciences.’ UKRI, 2018. https://www.ukri.org/publications/the-era-of-
mathematics/.
Brochu, Pierre, Maria Chuy, Marie-Anne Deussing, Koffi Philippe Houme, and Council of
Ministers of Education, Canada (CMEC). Measuring up, Canadian Results of the OECD
PISA Study: The Performance of Canada’s Youth in Mathematics, Reading and
Science: 2012, First Results for Canadians Aged 15. Toronto: CMEC, 2014.
http://site.ebrary.com/lib/celtitles/docDetail.action?docID=10831057.
Cheng, Jing, Jiansheng Bao, and Dianzhou Zhang. ‘From “Two Basics”, to “Four Basics” to
“Core Mathematics Competencies” in Mainland China’. In Beyond Shanghai and
PISA: Cognitive and Non-Cognitive Competencies of Chinese Students in
Mathematics, edited by Binyan Xu, Yan Zhu, and Xiaoli Lu, 1–13. Research in
Mathematics Education. Cham: Springer International Publishing, 2021.
https://doi.org/10.1007/978-3-030-68157-9_1.
Chernoff, Egan J. ‘The Canadian Math Wars’. Canadian Journal of Science, Mathematics
and Technology Education 19, no. 1 (1 March 2019): 73–76.
https://doi.org/10.1007/s42330-018-0037-9.
Cobb, George W. ‘The Introductory Statistics Course: A Ptolemaic Curriculum?’ Technology
Innovations in Statistics Education 1, no. 1 (12 October 2007).
https://doi.org/10.5070/T511000028.
‘Course Descriptions and Prerequisites’. Accessed 23 March 2023.
https://www.dcp.edu.gov.on.ca/en/course-descriptions-and-
prerequisites/mathematics.
Davies, Neville, and Neil Sheldon. ‘Teaching Statistics and Data Science in England’s
Schools’. Teaching Statistics 43, no. S1 (2021): S52–70. https://doi.org/10.1111/test.12276.

April 2023 pg. 106

———. ‘Teaching Statistics and Data Science in England’s Schools’. Teaching Statistics 43 (2
June 2021): S52–70. https://doi.org/10.1111/test.12276.
Department for Education. ‘Guide to the Post-16 Qualifications Landscape at Level 3 and
below for 2025 and Beyond’. Department for Education, 2023.
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/atta
chment_data/file/1128144/Guide_to_the_post-
16_qualifications_landscape_at_level_3_and_below_for_2025_and_beyond.pdf.
———. Mathematics Programmes of Study: Key Stage 4. National Curriculum in England,
2014.
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/atta
chment_data/file/331882/KS4_maths_PoS_FINAL_170714.pdf.
———. ‘Participation in Education, Training and Employment Age 16 to 18, Calendar Year 2021’,
2021. https://explore-education-statistics.service.gov.uk/find-
statistics/participation-in-education-and-training-and-employment.
Donoho, David. ‘50 Years of Data Science’. Journal of Computational and Graphical
Statistics 26, no. 4 (2 October 2017): 745–66.
https://doi.org/10.1080/10618600.2017.1384734.
Forbes, Sharleen. ‘The Coming of Age of Statistics Education in New Zealand, and Its
Influence Internationally’. Journal of Statistics Education 22, no. 2 (July 2014): 2.
https://doi.org/10.1080/10691898.2014.11889699.
Geiger, Vince, Peter Galbraith, Mogens Niss, and Catherine Delzoppo. ‘Developing a Task
Design and Implementation Framework for Fostering Mathematical Modelling
Competencies’. Educational Studies in Mathematics 109, no. 2 (February 2022): 313–
36. https://doi.org/10.1007/s10649-021-10039-y.
Geraniou, Eirini, and Morten Misfeldt. ‘The Mathematical Competencies Framework and
Digital Technologies’. In Mathematical Competencies in the Digital Era, edited by Uffe
Thomas Jankvist and Eirini Geraniou, 39–60. Mathematics Education in the Digital
Era. Cham: Springer International Publishing, 2022. https://doi.org/10.1007/978-3-031-
10141-0_3.
Golding, Jenefer. ‘Mathematics Education and Financial Literacy: A Valued but Vulnerable
Symbiosis’. In Proceedings of the British Society for Research into Learning
Mathematics 40 (1) March 2020, edited by Rachel Marks, 2020.
http://www.bsrlm.org.uk/publications/proceedings-of-day-conference/ip40-1/.
Golding, Jennie, Cathy Smith, and Mick J Blaylock. ‘Learning to Teach Contextualized
Problem-Solving in a Non-Calculus Mathematics Pathway’. Teaching Mathematics
and Its Applications: An International Journal of the IMA 37, no. 2 (4 June 2018): 69–
83. https://doi.org/10.1093/teamat/hry009.
Gould, Robert. ‘Toward Data-Scientific Thinking’. Teaching Statistics 43, no. S1 (July 2021).
https://doi.org/10.1111/test.12267.
Grover, Shuchi, and Roy Pea. ‘Computational Thinking in K–12: A Review of the State of the
Field’. Educational Researcher 42, no. 1 (1 January 2013): 38–43.
https://doi.org/10.3102/0013189X12463051.

April 2023 pg. 107

Guo, Meize, and Anne Ottenbreit-Leftwich. ‘Exploring the K-12 Computer Science Curriculum
Standards in the U.S’. In Proceedings of the 15th Workshop on Primary and Secondary
Computing Education, 1–6. Virtual Event Germany: ACM, 2020.
https://doi.org/10.1145/3421590.3421594.
Hall, Jennifer. ‘Gender Issues in Mathematics: An Ontario Perspective’. Journal of Teaching
and Learning 8, no. 1 (5 January 2012). https://doi.org/10.22329/jtl.v8i1.3004.
Hernandez-Martinez, Paul, Julian Williams, Laura Black, Pauline Davis, Maria Pampaka, and
Geoff Wake. ‘Students’ Views on Their Transition from School to College Mathematics:
Rethinking “Transition” as an Issue of Identity’. Research in Mathematics Education 13,
no. 2 (2011): 119–30. https://doi.org/10.1080/14794802.2011.585824.
Hodgen, Jeremy. Is the UK an Outlier?: An International Comparison of Upper Secondary
Mathematics Education. London: Nuffield Foundation, 2010.
Hong, Yanjun, and Uffe Thomas Jankvist. ‘Comparing Two Mathematical Competencies
Frameworks from China and Denmark’. Asian Journal for Mathematics Education 1,
no. 4 (2022): 494–514. https://doi.org/10.1177/27527263221144193.
Hoyles, Celia, ed. Improving Mathematics at Work: The Need for Techno-Mathematical
Literacies. 1st ed. New York, NY: Routledge, 2010.
Institute of Analytics. ‘IoA Analyst Competency Framework’, 2022. https://ioaglobal.org/wp-
content/uploads/2022/06/The_IoA_Analyst_Competency_Framework.pdf.
International Data Science in Schools Project (IDSSP) Curriculum Team. ‘Curriculum
Frameworks for Introductory Data Science’, 2019.
https://www.idssp.org/pages/framework.html.
Jensen, Jens Hojgaard, and Uffe Thomas Jankvist. ‘Disciplinary Competence Descriptions
for External Use’. Nordic Studies in Mathematics Education 23, no. 2 (2018): 3–24.
Kilhamn, C., K. Bråting, O. Helenius, and J. Mason. ‘Variables in Early Algebra: Exploring
Didactic Potentials in Programming Activities.’ ZDM - Mathematics Education 54, no.
6 (November 2022): 1273–88.
Kilpatrick, Jeremy. ‘Competency Frameworks in Mathematics Education’. In Encyclopedia of
Mathematics Education, edited by Stephen Lerman, 110–13. Cham: Springer
International Publishing, 2020. https://doi.org/10.1007/978-3-030-15789-0_27.
Kispeter, Erika. ‘What Digital Skills Do Adults Need to Succeed in the Workplace Now and in
the next 10 Years?’ Report. London: Department of Culture, Media and Sport, June
2018.
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/atta
chment_data/file/807831/What_digital_skills_do_adults_need_to_succeed_in_th
e_workplace_now_and_in_the_next_10_years_.pdf.
Kjällander, Susanne, Linda Mannila, Anna Åkerfeldt, Fredrik Heintz, João Piedade, and Nuno
Dorotea. ‘Elementary Students’ First Approach to Computational Thinking and
Programming.’ Education Sciences 11, no. 2 (February 2021): 80.
Kullberg, Angelika, Ulla Runesson Kempe, and Ference Marton. ‘What Is Made Possible to
Learn When Using the Variation Theory of Learning in Teaching Mathematics?’ ZDM
49, no. 4 (1 August 2017): 559–69. https://doi.org/10.1007/s11858-017-0858-4.

April 2023 pg. 108

Lazarus, Jill. ‘Secondary Curriculum Over the Last Decade’. Gazette - Ontario Association for
Mathematics 60, no. 3 (2022): 52–54.
MEI. ‘Curriculum and Assessment in Mathematics in England at KS4 and KS5 - Position
Paper’. Mathematics in Education and Industry, 2022. https://mei.org.uk/curriculum-
and-assessment-in-mathematics-in-england-at-ks4-and-ks5-position-paper/.
Morgan, Candia, and Sarah Tang. ‘To What Extent Are Students Expected to Participate in
Specialised Mathematical Discourse? Change over Time in School Mathematics in
England’. Research in Mathematics Education 18, no. 2 (3 May 2016): 142–64.
https://doi.org/10.1080/14794802.2016.1174145.
Mou, Haizhen, and Michael M. Atkinson. ‘Want to Improve Math Scores? An Empirical
Assessment of Recent Policy Interventions in Canada’. Canadian Public Policy /
Analyse de Politiques 46, no. 1 (2020): 107–24.
National Academies of Sciences, Engineering, and Medicine. Data Science for
Undergraduates: Opportunities and Options. Washington, D.C.: National Academies
Press, 2018. https://doi.org/10.17226/25104.
news.ontario.ca. ‘Ontario Dedicating $60 Million for Renewed Math Strategy’. Accessed 23
March 2023. https://news.ontario.ca/en/release/36364/ontario-dedicating-60-
million-for-renewed-math-strategy.
1. Niss, M., R. Bruder, N. Planas, R. Turner, and J. A. Villa-Ochoa, ‘Conceptualisation of the Role
of Competencies, Knowing and Knowledge in Mathematics Education Research’, in G.
Kaiser (ed.), Proceedings of the 13th International Congress on Mathematical
Education, (Cham: Springer International Publishing, 2017), pp. 235–48
Niss, Mogens, and Tomas Højgaard. ‘Mathematical Competencies Revisited’. Educational
Studies in Mathematics 102, no. 1 (1 September 2019): 9–28.
https://doi.org/10.1007/s10649-019-09903-9.
Noyes, Andrew, and Michael Adkins. ‘Studying Advanced Mathematics in England: Findings
from a Survey of Student Choices and Attitudes’. Research in Mathematics Education
18, no. 3 (1 September 2016): 231–48. https://doi.org/10.1080/14794802.2016.1188139.
Noyes, Andy, P. Drake, G. Wake, and Murphy. ‘Evaluating Mathematics Pathways’.
Department of Education, 2011.
Nuffield Foundation. ‘Mathematics in A Level Assessments’. Nuffield Foundation, 2012.
https://www.nuffieldfoundation.org/wp-content/uploads/2012/04/Mathematics-in-
A-level-assessments-A-report-on-the-mathematical-content-of-A-level-
assessments-in-Business-Studies-Computing-Economics-Geography-Psychology-
and-Sociology.pdf.
———. ‘Mathematics in A Level Assessments’. Nuffield Foundation, 2012.
https://www.nuffieldfoundation.org/wp-content/uploads/2012/04/Mathematics-in-
A-level-assessments-A-report-on-the-mathematical-content-of-A-level-
assessments-in-Business-Studies-Computing-Economics-Geography-Psychology-
and-Sociology.pdf.
OECD. Education Policy Outlook 2019: Working Together to Help Students Achieve Their
Potential. Education Policy Outlook. OECD, 2019. https://doi.org/10.1787/2b8ad56e-en.

April 2023 pg. 109

———. ‘PISA 2021 Mathematics Framework (Second Draft)’, 2018.
https://www.oecd.org/pisa/sitedocument/PISA-2021-mathematics-framework.pdf.
O’Grady, Kathryn, Marie-Anne Deussing, Tanya Scerbina, Yitian Tao, Karen Fung, Vanja Elez,
and Jeremy Monk. ‘Measuring up: Canadian Results of the OECD PISA 2018 Study’.
Council of Ministers of Education, Canada, 2019.
Ontario Ministry of Education. The Ontario Curriculum, Grade 9: Mathematics (2021), 2021.
https://www.dcp.edu.gov.on.ca/en/curriculum/secondary-
mathematics/courses/mth1w.
———. The Ontario Curriculum, Grades 1-8: Mathematics (2020), 2020.
https://www.dcp.edu.gov.on.ca/en/curriculum/elementary-mathematics.
———. The Ontario Curriculum, Grades 11 and 12: Mathematics (2007), 2007.
https://www.edu.gov.on.ca/eng/curriculum/secondary/math1112currb.pdf.
Research to Inform Elementary Mathematics Curriculum Revision, 2020.
https://www.youtube.com/watch?v=AmZtIpAHZs4.
Ridgway, Jim. ‘Implications of the Data Revolution for Statistics Education: The Data
Revolution and Statistics Education’. International Statistical Review 84, no. 3
(December 2016): 528–49. https://doi.org/10.1111/insr.12110.
Rittle-Johnson, Bethany, Michael Schneider, and Jon R. Star. ‘Not a One-Way Street:
Bidirectional Relations Between Procedural and Conceptual Knowledge of
Mathematics’. Educational Psychology Review 27, no. 4 (December 2015): 587–97.
https://doi.org/10.1007/s10648-015-9302-x.
Royal Society Te Apārangi Expert Panel. ‘Pāngarau Mathematics and Tauanga Statistics in
Aotearoa New Zealand. Advice on Refreshing the English-Medium Mathematics and
Statistics Learning Area of the New Zealand Curriculum’. Royal Society Te Apārangi,
2021.
Royal Society’s Advisory Committee on Mathematics Education’s (ACME) Post-16 Contact
Group. ‘Mathematics for the T Level Qualifications: A Rationale for General
Mathematical Competences (GMCs)’, 2019.
https://royalsociety.org/~/media/policy/topics/education-
skills/Maths/Mathematics%20for%20the%20T%20Level%20Qualifications%20-
%20a%20rationale%20for%20GMCs.pdf?la=en-GB.
Schoenfeld, A. ‘Learning to Think Mathematically: Problem Solving, Metacognition and Sense
Making in Mathematics’. In Handbook of Research on Mathematics Teaching and
Learning, edited by D. Grouws, 335–70. New York: Macmillan, 1992.
SCORE. ‘Mathematics within A Level Science 2010 Examinations’. SCORE (Science Community
Representing Education), 2012.
https://www.stem.org.uk/resources/elibrary/resource/25956/mathematics-within-
level-science-2010-examinations#&gid=undefined&pid=1.
Sinitsky, Ilya, and Bat-Sheva Ilany. Change and Invariance: A Textbook on Algebraic Insight
into Numbers and Shapes, 2016.
https://search.ebscohost.com/login.aspx?direct=true&scope=site&db=nlebk&db=nl
abk&AN=1429092.

April 2023 pg. 110

Smith, Cathy. ‘Gender and Participation in Mathematics and Further Mathematics A Levels:
A Literature Review for the Further Mathematics Support Programme’, 2014.
https://core.ac.uk/download/pdf/79498409.pdf.
Smith, Cathy, and Jennie Golding. ‘Schools’ Strategies for Promoting Girls’ Participation in
Mathematics’. In Proceedings Of the 42nd Conference of the International Group for
the Psychology of Mathematics Education (Vol 4) Umeå, Sweden., edited by Ewa
Bergqvist, Marcus Österholm, Carina Granberg, and Lovisa Sumpter, 211–18, 2018.
Smith, Cathy, and Candia Morgan. ‘Curricular Orientations to Real-World Contexts in
Mathematics’. The Curriculum Journal 27, no. 1 (2 January 2016): 24–45.
https://doi.org/10.1080/09585176.2016.1139498.
Suurtamm, Christine. ‘Elementary Mathematics Curriculum, Teaching, And Learning: The
Past Ten Years’. Ontario Mathematics Gazette 60, no. 3 (2022): 19–24.
Suurtamm, Christine, Denisse R. Thompson, Rae Young Kim, Leonora Diaz Moreno, Nathalie
Sayac, Stanislaw Schukajlow, Edward Silver, Stefan Ufer, and Pauline Vos.
‘Assessment in Mathematics Education’. In Assessment in Mathematics Education:
Large-Scale Assessment and Classroom Assessment, edited by Christine Suurtamm,
Denisse R. Thompson, Rae Young Kim, Leonora Diaz Moreno, Nathalie Sayac,
Stanislaw Schukajlow, Edward Silver, Stefan Ufer, and Pauline Vos, 1–38. ICME-13
Topical Surveys. Cham: Springer International Publishing, 2016.
https://doi.org/10.1007/978-3-319-32394-7_1.
Tight, Malcolm. ‘Twenty-First Century Skills: Meaning, Usage and Value’. European Journal of
Higher Education 11, no. 2 (3 April 2021): 160–74.
https://doi.org/10.1080/21568235.2020.1835517.
Tukey, John W. ‘The Future of Data Analysis’. The Annals of Mathematical Statistics 33, no. 1
(1962): 1–67.
Wake, Geoff. ‘Making Sense of and with Mathematics: The Interface between Academic
Mathematics and Mathematics in Practice’. Educational Studies in Mathematics 86,
no. 2 (June 2014): 271–90. https://doi.org/10.1007/s10649-014-9540-8.
Wal, Nathalie J. van der, Arthur Bakker, and Paul Drijvers. ‘Which Techno-Mathematical
Literacies Are Essential for Future Engineers?’ International Journal of Science and
Mathematics Education 15, no. 1 (1 May 2017): 87–104. https://doi.org/10.1007/s10763-
017-9810-x.
Watson, Anne. Key Ideas in Teaching Mathematics: Research-Based Guidance for Ages 9-
19. Oxford, United Kingdom: Oxford University Press, 2013.
Wolff, Annika, Michel Wermelinger, and Marian Petre. ‘Exploring Design Principles for Data
Literacy Activities to Support Children’s Inquiries from Complex Data’. International
Journal of Human-Computer Studies 129 (September 2019): 41–54.
Zevenbergen, Robyn. ‘Mathematics, Social Class, and Linguistic Capital: An Analysis of
Mathematics Classroom Interactions’. In Sociocultural Research on Mathematics
Education, edited by Atweh, Bill, Helen Forgasz, and Ben Nebres, 201–15. Mahwah:
Lawrence Erlbaum Associates, 2001.

April 2023 pg. 111

 Appendix A Expert interviews
We gratefully acknowledge the following experts who contributed to this report, providing
knowledge of professional and education policy, curriculum initiatives, research and
evaluations within their fields of mathematics, statistics, data science and computing.

Christian Bokhove, Professor in Mathematics Education, University of Southampton

Tom Button, Maths Technology Specialist, MEI

Sophie Carr, Vice President for Education and Statistical Literacy, Royal Statistical Society

Neville Davies, Emeritus Professor, University of Plymouth

Kate Farrell, Director of Curriculum Development, Data Education in Schools, Scotland

Anna Fergusson, Professional Teaching Fellow, Dept of Statistics, University of Auckland

Matt Forshaw, Senior Advisor for Skills, Alan Turing Institute

Eirini Geraniou, Associate Professor of Mathematics Education, UCL

Rachel Hilliam, Chair, Alliance for Data Science Professionals

Jenni Ingram, Associate Professor of Mathematics Education, University of Oxford

Rhys Jones, Associate Dean, Faculty of Health and Medical Sciences, University of Surrey

Christina Leslie, Computational Biologist, Sloan Kettering Institute, USA

Darren Macey, Framework Design, Cambridge Mathematics

Simon Pinfield, Manager: Resources, Projects and Partnerships, Royal Geographical Society

Tom Rainbow, Advanced Mathematics Support Programme

Jim Ridgway, Emeritus Professor in the School of Education, Durham University

Judy Robertson, Chair in Digital Learning, University of Edinburgh

Neil Sheldon, Chair, Teaching Statistics Trust

Deborah Sutch, Curriculum Manager, Mathematics, Diploma Programme, IBO

Christine Suurtamm, Professor of Mathematics Education, University of Ottawa

Clare Walsh, Head of Education, Institute of Analytics

April 2023 pg. 112

Appendix B Royal Society’s indicative
curriculum models

April 2023 pg. 113

View publication stats