2013; 35: 731–739

Do differentials in the support and advice

available at UK schools and colleges influence
candidate performance in the medical school
admissions interview? A survey of direct school
leaver applicants to a UK medical school
PAUL LAMBE1, CATHERINE WATERS2 & DAVID BRISTOW1
1
Plymouth University, UK, 2Truro College, UK

Abstract
Background: To our knowledge, nothing is known about whether differentials in support and advice during preparation for the
interview influence candidate performance and thereby contribute to bias in selection for medical school.
Aim: To assess if differences in advice and support with preparation for the medical school admissions interview given type
of school last attended influence interview score achieved by direct school leaver applicants to study on an undergraduate
UK medical degree course.
Methods: Confidential self-completed on-line questionnaire survey.
Results: Interview performance was positively related to whether a teacher, tutor or career advisors at the School or College
last attended had advised a respondent to prepare for the interview, had advised about the various styles of medical interview
used and the types of questions asked, and what resources were available to help in preparation. Respondents from Private/
Independent schools were more likely than those from State schools to have received such advice and support.
Conclusions: Differentials in access to advice on and support with preparation for the medical school interview may advantage
some candidates over others. This inequity would likely be ameliorated by the provision of an authoritative and comprehensive
guide to applying to medical school outlining admission requirements and the preparation strategy applicants should use in order
to best meet those requirements. The guide could be disseminated to the Principals of all UK schools and colleges and
freely available electronic versions signposted in medical school prospectuses and the course descriptor on the Universities
and Colleges Admissions Service.

Introduction Practice points

The medical school interview is a mainstay of the selection
. Selection to study medicine should be fair.
procedures used in the UK, North America and Australia
. This study reports that advice and support with prep-
(Parry et al. 2006; Basco 2008; Axelson et al. 2010;Griffin et al.
aration for the medical school admissions interview
2010; Mercer & Puddey 2011). However, most studies to
influence interview score.
date have focussed on the reliability and predictive validity of
. Private/independent schools pupils were more likely
medical school interview scores or, the attributes that should
than those from State schools to have received such
be assessed at interview (Powis & Rolfe 1988; Kulatunga-
13

advice and support.

Moruzi & Norman 2002; Lemay et al. 2007; Lambe & Bristow
20

2011). Research has shown that selection bias may occur

during the interview, emanating from interviewer tendencies
to be hawkish or lenient, stereotyping and interviewer
In respect of the Multiple Mini-Interview (MMI) it has
background (Edwards et al. 1990; Morris 1999, Griffin &
been found that prior access to interview questions has
Wilson 2010; Lumb et al. 2010). To our knowledge, nothing
no statistically significant effect on performance, indicating
is known about whether differentials in support and advice
that candidates accrue no gain, or ‘practice effect’, from the
during preparation for the medical school interview influence
opportunity to rehearse, or obtain advice on what constitutes
candidate’s performance and thereby contribute to a potential
an appropriate response (Reiter et al. 2006). Similarly, findings
bias in selection for medical school.
from a study of the effect of commercially provided coaching

Correspondence: Professor David Bristow, C413 Portland Square, Drake Circus, Plymouth PL4 8AA, UK. Tel: þ44(0) 1752 586754; fax: þ44(0) 1752
586788; email: [email protected]
ISSN 0142–159X print/ISSN 1466–187X online/13/90731–9 ß 2013 Informa UK Ltd. 731
DOI: 10.3109/0142159X.2013.801941
P. Lambe et al.

programmes on performance in the MMI suggest that ‘training at: (http://www1.plymouth.ac.uk/peninsula/research/camera/

is ineffective for improving MMI scores’ (Griffin et al. 2008). Documents/IntPrepSurvey.pdf).
The typical MMI comprises a series of structured interviews,
each designed to assess personal qualities (Eva et al. 2004).
In contrast, research in the field of occupational psychology Sample and study protocol
has found preparation and coaching to be associated
The sample comprised direct school leaver applicants
with enhanced interview performance and rating particularly
who had been invited to a selection interview for a place
in respect of the more structured and formulaic type of
on an undergraduate entry Bachelor of Medicine, Bachelor
traditional employment interview (Maurer & Solomon 2006;
of Surgery (BMBS) course (2011–2012 entry) at a UK medical
Tross & Maurer 2008). Although the format of the traditional
school. The aims of the study and assurance of confidentiality
admissions interview may differ among medical schools and
was explained in both an initial and non-response follow-up
vary in duration, panel composition, content and scoring
email. Applicants were asked to participate in the study after
method, in the main UK medical school interviews are
they had been interviewed, and completed the questionnaire
structured (Parry 2006) and designed to measure attributes
before they knew the outcome of whether they had been
aligned with the outcomes desired for graduates as outlined
offered a place or not. Respondents’ interview scores were
by the General Medical Council (GMC 2009). The UK medical
matched to their respective questionnaires by the school’s
school admissions interview typically probes a limited number
Registry Services and the anonymised questionnaires then
of common themes which include inter alia – motivation,
given over to the authors for analysis.
enthusiasm, understanding of and commitment to pursuing
a medical career; knowledge of the course; possession of
personal qualities appropriate to medicine; possession of skills
Statistical analysis
appropriate to medicine; awareness of the issues facing the
health sector, work experience and tolerance of ambiguity Latent class analysis (LCA) was conducted using the software
(Lambe & Bristow 2010). The use of the traditional interview, Latent GOLD Version 4.0. (Statistical Innovations, Inc.,
either structured or semi-structured, based on a set of well- Belmont, MA 02478). LCA is a non- parametric statistical
defined criteria and rated on pre-determined rating scales, method (does not assume linear relationship, normality of data
is widespread in countries such as New Zealand, Australia, distribution or homogeneity of variance) for identifying
Canada and the USA (Mercer 2009). Although the weight subtypes of related cases, typically from categorically scored
given to interview rating in the selection process may vary survey data (Hagenaars & McCutcheon 2002; Lambe & Bristow
considerably, interview performance is a key determinant of 2011). The classes are defined by the criterion of conditional
the likelihood of an offer of a place at medical school. independence: conditional on latent class membership, the
Whilst acknowledging that selection bias may occur at any observed variables are mutually independent of one another.
stage of the selection process, it is possible that some Thus the variables within any one class are uncorrelated.
applicants to study medicine are disadvantaged by factors Posterior membership probabilities (maximum likelihood esti-
extrinsic to the interview process. We hypothesised that mates based on the patterns of scores on the observed
respondents who occupied a relatively subordinate position variables) assign cases to homogeneous latent classes. Cases
in terms of ‘habitus’, defined as the level of support and access in a particular latent class are similar to one another because
to material and non-material resources at the school or college their response patterns are generated by the same probability
they last attended, would perform less well at the interview distribution.
(Bordieu 1977; Bordieu 1997). Two useful parameters are produced: latent class prob-
The aim of this survey study was to identify if there are abilities and conditional probabilities. Latent class probabilities
differentials in access to advice and support in preparation indicate the relative size of each class and whether the sample
for the medical school interview depending on the type of population is relatively evenly distributed among the classes.
school or college last attended, and if so, what effect, if any, Within each class there is a set of conditional probabilities
this may have on the interview performance. relating to each variable (survey items in the model).
The conditional probabilities represent the probabilities of an
individual in a particular class being at a particular response
Methods level for a particular variable and thereby enable characterisa-
tion of the nature of the types (typologies) defined by each
The questionnaire of the latent classes. Analogous to factor loadings in factor
The questionnaire comprised 20 questions and asked about; analysis, the conditional probabilities provide the measure-
preparation for the medical school interview, advice and ment structure that defines the latent classes.
support received, resources used, resources considered most Specification of the basic LCA model does not involve the
helpful, demographics and how candidates thought they could issue of causal ordering. In these analyses, our interest is in
be better helped to prepare. The survey was conducted before the underlying association between variables and we make
respondents knew the outcome of their medical school no distinction between what is dependent and what is
interview. The questionnaire, informed by a pilot study with explanatory. Used here as a data reduction technique,
first year students at this medical school, was developed the method empirically establishes the dimensions of any
and refined by the authors. The questionnaire can be accessed underlying associations.
732
 Do differentials in support and advice influence interview performance?

The survey items included as binary indicators in the model Table 1. Demographics and mean interview score of
Direct School Leaver applicants (DSL) for 2011/12 entry to this
were: (1) candidate was advised to prepare for the interview medical school, compared to the study sample.
by a tutor, teacher or career advisor; (2) candidate was advised
at school/college last attended on the styles of medical school
interview and types of questions asked; (3) candidate was Applicants to this
DSL medical school Study sample
advised at school/college last attended on what resources
Male (%) 53 45
are available to help prepare for the test; (4) candidate was
Female (%) 47 55
provided with a medical school interview course at the school/ Interview score Mean ¼ 21; Mean ¼ 21;
college last attended. Two active covariates were also included Standard Deviation ¼ 4.3; Standard Deviation ¼ 4.0;
minimum ¼ 6; minimum ¼ 10;
in the model: quartile interview score (scored 1 ¼ bottom maximum ¼ 29 maximum ¼ 29
through 4 ¼ top quartile); and school type (1 ¼ Private/
Independent, 0 ¼ State). Private/Independent was defined
as Grammar fee-paying, Private and Independent schools,
State school was defined as Comprehensive, Grammar no fees, Latent class analysis
Sixth Form College and Further Education College.
Seven further survey items were included in the model as A sequence of models was fitted to the data with different
categorically scored inactive covariates. Unlike active covari- numbers of classes and a two-class model was found to
ates, inactive covariates do not influence the model parameter provide the best fit to our data (Table 2). This was confirmed
estimates and are included in the model to enable further by a bootstrap test which showed that the three-class model
description of the typologies of candidate interview prepar- did not provide a significant improvement in fit over the two-
ation. They were; number of medical school interviews class model. A good model fit is indicated by a low-likelihood
attended, time spent on preparation, amount of money spent ratio statistic (L2), not substantially greater than the model
on preparation resource, most helpful preparation resource degrees of freedom, a p value 4 0.05, a low Bayesian
used, school/college last attended, applicant part of a men- information criterion (BIC) a low Aikake information criterion
toring/widening access to medicine scheme and gender. (AIC) and a dissimilarity index (D) of 50.1 (Hagenaars &
To triangulate the findings of the latent class analysis, McCutcheon 2002).
respondents’ interview score was regressed on respondents’ The two-class model’s indicators, and the active covariates
posterior probability of latent class assignment (maximum included in the model, which all had Wald statistic p values
likelihood estimates). Additionally, an Index of Support was of 50.05, indicated that each contributed significantly
created based on respondent’s answers to the following binary towards the ability to discriminate between the classes
(scored yes ¼ 1, no ¼ 0) survey items: advised to prepare for (Table 3). Thus, both classes contained a homogeneous
the medical interview by a teacher, tutor or career advisors; group of respondents who shared common characteristics
advised by a teacher, tutor or careers advisor about the styles and behaviours with respect to the observed indicators of
of medical school interviews used and the types of questions preparation for a medical school interview. The LCA identified
two distinct classes, or typologies of preparation for the
asked; advised by a teacher, tutor or careers advisor on what
medical school interview. Class I comprised 57% of respond-
resources were available to help prepare for the interview; and
ents, and Class II 43% of respondents (Latent Class Probability
School provided an interview preparation course. A general-
of 0.57 and 0.43, respectively; Table 3).
ised analysis indicated that the reliability of the Index of
Respondents in Class I were highly likely to have been
Support was at an acceptable level (G-coefficient ¼ 0.7,
advised to prepare for the medical school interview by a
Standard Error of Measure ¼ 0.19) (Streiner et al. 2008).
teacher, tutor or career advisor at the school/college they
last attended (Conditional Probability of 0.88; Table 3).
In contrast, respondents in Class II were much less likely to
Ethical considerations have been so advised at the school/college they attended
Ethical approval for this study was sought from and granted (0.38). Indeed, respondents in Class I were 12 times as likely
by Peninsula College of Medicine and Dentistry Ethics to have been advised to prepare for the medical school
Committee. interview by a teacher, tutor or career advisor at the school/
college they attended than those in Class II (odds ratio ¼ 12.02,
Confidence Interval 10.47 to 13.78). Class I respondents were
also highly likely to have been advised about the styles of
Results
medical school interviews used and the types of questions
The survey response rate was 55%, with 218 of the 400 asked (0.89), whilst those in Class II were highly unlikely
direct school leaver applicants to the course (2011/12 entry to have been so advised (0.01). Additionally, respondents
cycle) who were interviewed in December 2010 completing in Class I were much more likely to have been advised
questionnaires. The interview score distributions revealed on what resources were available to help them prepare
no difference between the population of direct school for the medical school interview than those in Class II
applicants for entry to this medical school in 2011/12 (0.48 versus 0.06), and their school/college was more
and the self-selected sample drawn from this population likely to have provided an interview preparation course
(Table 1). (0.39 versus 0.03).
733
P. Lambe et al.

Table 2. Goodness-of-fit statistics of latent class analysis models containing different numbers of classes or ‘typologies’.

Latent Class Models X2 L2 d.f. Npar BIC AIC D

I 240.00 197.48 116 4 411.77 34.51 0.36
II 70.65 72.66 109 11 499.83 145.33 0.08
III 62.04 61.22 102 18 460.29 130.53 0.13
IV 51.63 57.77 95 25 441.25 132.28 0.17
BIC ¼ Bayesian information criterion statistic; AIC Aikake information criterion statistic

Classification error rate for the 2-class model ¼ 0.04, and Lambda ¼ 0.89 (measure the closeness of association between the latent classes and the indicator
variables, 0 ¼ no association to 1 ¼ perfect association).

Table 3. Outline of conditional and latent class probabilities for the two-class model,
(statistically significant indicators and covariates: Wald statistic p values 5 0.05).

Conditional probabilities

Indicators Class I Class II Wald p Value

School/College advised on styles of medical school interviews & types of questions 0.89 0.01 11.73 0.00
School/College advised on what resources available to help prepare for medical school interview 0.48 0.06 13.94 0.00
School/College provided a medical school interview preparation course 0.39 0.03 16.42 0.00
Was advised to prepare for the medical school interview by a teacher, tutor, or career advisor 0.88 0.38 33.87 0.00
Active covariates
Quartile interview score (1 ¼ bottom)
1 0.19 0.32 4.23 0.01
2 0.22 0.26 4.55 0.03
3 0.27 0.25
4 0.32 0.17
Interview score (continuous) 0.05 0.04
Public/Independent school 0.23 0.13 1.83 0.02
Inactive covariates
Number of interviews
1 0.45 0.52
2 0.39 0.24
3þ 0.16 0.23
Time preparing
0–10 hrs 0.49 0.51
11–20 hrs 0.36 0.37
21 þ hrs 0.15 0.12
Money spent on preparation
£0 0.37 0.25
£1–50 0.51 0.61
£51–250 0.10 0.13
£251–500 0.01 0.00
£500þ 0.01 0.00
Most helpful practise resource
Teacher, Tutor 0.13 0.01
Current medical student 0.08 0.14
Online resources 0.05 0.14
Online discussion forums 0.03 0.14
Books 0.18 0.32
Practise interviews 0.49 0.21
Mentoring scheme 0.00 0.00
Attend interview prep-course 0.03 0.04
School/college last attend
Comprehensive 0.19 0.26
Grammar no fees 0.27 0.15
Grammar fees 0.06 0.00
Public/Independent 0.18 0.14
Sixth-form college 0.26 0.41
Further education college 0.04 0.04
Part of WP scheme 0.10 0.01
Female 0.58 0.47
Latent class probability 0.57 0.43

734
 Do differentials in support and advice influence interview performance?

Inspection of the conditional probabilities associated

with the inactive covariates (Table 3) revealed little difference
Advised to prepare by a tutor, teacher or career advisor between the two classes in the number of medical school
interviews attended, in time spent on preparation and the
amount of personal money spent on preparation resources.
School /college advised on interview styles & questions
These probabilities can be interpreted as percentages. Thus,
49% of those assigned to Class I considered practise interviews
School /college advised on preparation resources available the most helpful compared with 21% in Class II. A teacher or
tutor was considered the most helpful resource used in
School /college provided a preparation course
preparation for the interview by 13% of respondents in Class I,
compared to 1% in Class II. Conversely, a greater proportion
0 20 40 60 80 100 of Class II than Class I considered current medical students
Conditional probability
or books or online resources as the most helpful resources
Class I Class II
in their preparation. This differential may have reflected the
disparity between the two classes in the probability of support
Figure 1. Bar graph contrasting latent classes by the and advice in preparation from a teacher, tutor or career
conditional probabilities of a respondent having been advised advisor at the school/college a respondent last attended.
to prepare for the medical school interview by a tutor, teacher When contrasted by type of school last attended, 24% of
or career advisor, advised by the school/college about respondents in Class I had attended Private/Independent
interview styles and questions, advised by the school/college schools compared to 14% in Class II. Conversely, 41% of Class
about preparation resources available, and whether the II attended a Sixth Form College compared to 26% in Class I.
school/college provided a preparation course. Overall, 76% of Class I attended State schools compared to
86% of respondents in Class II. Respondents who had been
Importantly, respondents in Class I were less likely to have part of a widening access to medicine or mentoring scheme
scored in the bottom quartile at interview (0.19 versus 0.31) made up 10% of Class I and only 1% of Class II. Class I
and more likely to have achieved a top quartile score at comprised 58% females compared to 47% female in Class II.
interview than their counterparts in Class II (0.32 versus 0.17).
Indeed, those in Class I were over twice as likely to have
Advice and support
achieved a top quartile score at interview than those in
Class II (odds ratio ¼ 2.35, Confidence Interval 2.04 to 2.70). The survey results revealed that 68% of respondents were
Conversely, those in Class II were almost twice as likely advised to prepare by a teacher, tutor or career advisor, but
to have been awarded a bottom quartile interview score only 8% of respondents thought that the most helpful resource
(odds ratio ¼ 1.87, Confidence Interval 1.63 to 2.14). in preparing for the medical school interview was a teacher,
When interview score was included in the LCA model as tutor or career advisor. Advice on the styles of medical
a continuous level variable, the mean interview score for admission interview and types of questions asked was given to
respondents assigned to Class I was 22 points and that for 52% of respondents, and advice on the resources available
Class II 20 points (Interview score Wald statistic ¼ 5.41, to help prepare for the interview to 30% of respondents at the
p ¼ 0.02 Standard Error 0.31). An independent two-samples school/college they attended. A medical school interview
t-test rejected the null-hypothesis of no difference between preparation course was provided at schools and colleges
mean interview scores given latent class assignment attended by 24% of respondents. Commercially available
(t ¼ 2.61, df ¼ 189, Pr(|T| 4 |t|) ¼ 0.01). interview preparation courses were attended by 15% of
Furthermore, respondents who had attended a Private/ respondents. When asked if they had any comments on how
Independent school had over twice the likelihood of belong- candidates could be better helped to prepare for the medical
ing to Class I than to Class II (odds ratio ¼ 2.26, Confidence school interview 28% of respondents entered comments into a
Interval 1.92 to 2.59). free text box. The following are typical: ‘More explanation of
When the conditional probabilities that characterised each the interview itself in the prospectus. . ..’
of the two classes were graphically contrasted the disparity ‘Be more forward in encouraging interviewees to
between the two groups in respect of the likelihood of having prepare, maybe
received advice and support in their preparation for the including ideas as to how they could do it’
medical school interview was starkly illustrated (Figure 1). ‘Possibly have an area on the UCAS/University
Differentials in the actual percentage of respondents who website about how to
were advised to prepare by a teacher, tutor or careers prepare . . . typical questions asked, how inter-
officer, advised on the types of interview used, advised views . . . differ’
about preparation resources available, and whose school ‘Schools should be better informed on how to
provided an interview preparation course, are apparent help students prepare. . .
when contrasted by type of school/college last attended. resources such as books and websites should be
It would appear that overall, Public/Independent and easier to find’
Grammar (fee-paying) schools offer more relevant advice ‘Colleges . . . should give . . . more advice earlier
than other institution types (Figure 2). on about work experience
735
P. Lambe et al.

1: School/College advised on interview styles & types of questions 2: School/College advised on preparation resources available

10 20 30 40
20 40 60 80
Percentage

Percentage
ve

e)

ve

e)

e
en

en
es

es
g

g
fe

fe
si

si
le

le

le

le
(fe

nd

(fe

nd
en

en
ol

ol

ol

ol
o

o
pe

pe
(n

(n
C

C
ar

ar
eh

eh
de

de
rm

rm

n
ar

ar

m
pr

pr
io

io
m

m
m

m
In

In
Fo

Fo
om

om
at

at
ra

ra
m

m
e/

e/
uc

uc
C

ra

ra

h
at

at
xt

xt
Ed

Ed
G

G
iv

iv
Si

Si
Pr

Pr
er

er
rth

rth
Fu

Fu
3: School/College provided preparation course 4: Adivised to prepare by a tutor, teacher, career advisor

20 40 60 80 100
20 30 40 50
Percentage

Percentage
ve

e)

ge

ge

ve

e)

ge

ge
en

en
es

es
fe

fe
si

si
le

le

le

le
(fe

nd

(fe

nd
en

en
ol

ol

ol

ol
o

o
pe

pe
(n

(n
C

C
ar

ar
eh

eh
de

de
rm

rm

n
ar

ar

m
pr

pr
io

io
m

m
m

m
In

In
Fo

Fo
om

om
at

at
ra

ra
m

m
e/

e/
uc

uc
C

ra

ra

h
at

at
xt

xt
Ed

Ed
G

G
iv

iv
Si

Si
Pr

Pr
er

er
rth

rth
Fu

Fu
Figure 2. Contrasted by type of school/college attended the percentage of respondents given advice at the school/college they
attend (1) on the styles of interview and types of questions asked at a medical school interview; (2) on what resources were
available to help prepare for the medical school interview; (3) were provided with a medical school interview preparation course;
(4) were advised to prepare for the medical school interview by a tutor, teacher or career advisor. Only 1 respondent attended
a type of school or college not listed in the options), n ¼ 217.

Table 4. Model 1, fit statistics of the linear regression of latent class posterior probability on interview score (F(1,189)) ¼ 6.93, p ¼ 0.01)
R2 ¼ 0.03. Model 2, fit statistics of the linear regression of Index of Support on interview score (F(1,189)) ¼ 6.26, p ¼ 0.01) R2 ¼ 0.03.

Coefficient Std. Error t P 4 |t| 95% Confidence Interval Beta

Model 1
Posterior probability of Class I assignment 1.6219 0.61 2.63 0.01 0.4068 2.8370 0.19
Constant 19.87 0.44 44.69 0.0000 19.0999 20.8640

Model 2
Index of Support 0.5567 0.22 2.50 0.01 0.1177 0.9956 0.18
Constant 19.85 0.48 40.70 0.000 18.89 20.82

and types of things that you can talk about during rho ¼ 0.19, p 5 0.05), interview score and score on the
interview, especially at Support Index (Spearman’s rho ¼ 0. 18, p 5 0.05).
state colleges as there are so many students’ A respondent’s posterior probability of latent class assign-
ment (a maximum likelihood estimation from 0 to 1 of being
assigned to latent Class I, given individual response pattern to
survey items in the LCA model) was found to be a significant
Multivariate regression
independent predictor of interview score (F(1,189) ¼ 6.93,
Interview score, score on the Index of Support, and latent class Prob 4 F ¼ 0.01, R2 ¼ 0.03) (Model 1; Table 4). In a separate
assignment (1 ¼ Latent Class I, 0 ¼ Latent Class II), were linear regression model score on the Index of Support was
significantly correlated; latent class assignment and score found to be a significant independent predictor of interview
on the Support Index (Spearman’s rho ¼ 0. 86, p 5 0.01), score (F(1,191) ¼ 6.26, Prob 4 F ¼ 0.01, R2 ¼ 0.03) (Model 2;
latent class assignment and interview score (Spearman’s Table 4 and Figure 3). In Model 2, the intercept was 19.85, and
736
 Do differentials in support and advice influence interview performance?

Adjusted Predictions with 95% CIs questions, suitable responses and feedback on mock interview
performance (Maurer et al. 1998; Maurer et al. 2001, 2008). The
23

latter study, which examined the influence of an intervention

Linear Prediction Interview Score

designed to ‘focus interviewees on interview-relevant content

22

and familiarise them with the means for identifying and

conveying the content accurately in the interview’, concluded
that such interventions can positively influence interview
21

scores (Maurer et al. 2008). They postulate that the explanation

of how coaching and preparation interventions influence
20

interview behaviour might be found in Bandura’s social

cognitive theory, particularly the construct of self efficacy
19

(Bandura 1986; Maurer et al. 2008). They reason that coaching

0 1 2 3 4 and preparation interventions influence candidates’ self-
Index of Support
efficacy or perceived control over performance in the
interview. Self- efficacy is defined as an individual’s belief in
Figure 3. Linear prediction of interview score given score
his or her ability to face challenging situations competently and
on Index of Support (n ¼ 192).
the choices a person is most likely to make (Luszczynska &
Schwartzer 2005). We speculate that such a mechanism
underpins the association found in this study between advice
the coefficient for Index of Support was 0.55, which means on and support with preparation received at school or college
that the predicted mean for interview score when the Index and respondents’ subsequent performance in the medical
of Support as 0 was 19.85 and that for every unit increase school interview.
in the Index of Support interview score increased by 0.55 The findings of this survey study suggest that, irrespective
points (Table 4). of whether state or private sector, some schools and colleges
provided more support to aspiring medical students than
others. This is in agreement with the conclusion reached by
Discussion the British Medical Association, that a ‘lack of guidance in
Bordieu’s theoretical concept of ‘institutional habitus’ proved applying to medicine is not . . . a problem found exclusively
a useful framework in which to test our hypotheses. Bordieu among lower socio-economic groups’ (BMA 2009). It is also
defined social capital as networks of influence and support, in general agreement with a longitudinal study which
and cultural capital as the material and non-material resources examined sources and availability of support and advice for
that are linked to the acquisition of knowledge and skills careers and future learning in UK schools. The study found
(Bordieu 1977; 1997). How much social and cultural capital a a substantial reduction in the impact and availability of
respondent could access in the ‘field’ of UK school or college specialist advice (Wiggins & Coe 2009). Although a recent
last attended, was found to be associated with interview study found an increase in the self- confidence of applicants
performance (Robb et al. 2007; Brosnan 2010). who had received advice and support on a scheme to facilitate
The findings suggest that differences in interview score are progression from UK schools and colleges to university, it also
in part explained by differences in access to advice and found that improvement in support for students with prepar-
support at the school or college a respondent attended during ation for the university admissions interview was much needed
the application stage for a place at medical school. The (Lamont et al. 2011).
analysis identified two distinct subgroups which differed in Although we consider that selection to UK medical schools
their probability of achieving a top quartile interview score is based on ‘valid, open, objective and fair selection proced-
according to measures of advice and support that they had ures’ this study indicated there is a variation in the quality and
received at the school or college they attended. Interview quantity of advice and support for aspiring medical school
performance was found to be positively related to whether applicants. Therefore, the profession’s intention to be fair to all
a teacher, tutor or careers advisor at the School or College candidates (General Medical Council 2009) may be under-
last attended had advised a respondent to prepare for the mined at the preparation stage prior to the medical school
interview, had been advised about the various styles of selection interview. Importantly, as pointed out by Griffin
medical interview used and the types of questions asked, et al., the issue of whether the predictive validity and reliability
and had been advised on what resources were available to of interview scores are changed by practice effects may need
help in preparation. to be resolved (Griffin et al. 2010). However this is beyond the
Despite the widespread use of interviews in selection scope of this study.
for employment little research has been conducted on the
effects of coaching and preparation. However, the findings
Conclusion
of this study align with those of the few studies which
have examined the influence of various levels of ‘coaching’ Differentials in access to advice on and support with prepar-
practices on employment interview behaviour, ranging ation for the medical school interview may advantage
from a general description of the process to participation some candidates over others. This inequity would likely
in and observation of interview role plays including sample be ameliorated by the provision of an authoritative and
737
P. Lambe et al.

comprehensive guide to applying to medical school outlining CATHERINE WATERS, PhD, is a Lecturer of Community Learning at Truro
College, Cornwall.
admission requirements and the preparation strategy appli-
cants should use in order to best meet those requirements.
Declaration of interest: The authors report no conflicts
The guide could be disseminated to the Principals of all UK
of interest. The authors alone are responsible for the content
schools and colleges and freely available electronic versions
and writing of the article.
signposted in medical school prospectuses and the course
descriptor on the Universities and Colleges Admissions
Service.
We are not aware of any studies that have investigated the References
effect of preparation and availability of resources on perform- Axelson R, Kreiter C, Ferguson K, Solow C, Huebner K. 2010. Medical
ance in the traditional medical school interview. This study School Preadmission Interviews: Are structured interviews more
has increased our knowledge and understanding about the reliable than unstructured interviews? Teach Learn Med 22(4):241–245.
equitability of an important element in this medical school’s Basco W, Lancaster C, Gilbert G, Carey M, Blue A. 2008. Medical school
application interview score has limited predictive validity for perform-
selection and admissions process.
ance on a fourth year clinical practice examination. Adv Health Sci
Educ Theory Pract 13:151–162.
Bandura A. 1986. Social foundations of thought and action: A social
Limitations cognitive view. Englewood Cliffs, NJ: Prentice Hall.
Bordieu P. 1977. Outline of a theory of practice. Cambridge: Cambridge
The use of non-probability sampling prevents evaluation of the University Press.
reliability of the resulting estimates and thus limits knowledge Bordieu P. 1997. The forms of capital. In: Halsey A, Lauder H, Brown P,
of how much confidence can be placed in the interpretation Wells A, editors. Education:Culture, Economy, Society. Oxford: Oxford
University Press. pp 241–258.
of our survey findings (Lemeshow & Levi 1999). Furthermore, British Medical Association. 2009 Equality and diversity in UK medical
the response rate of 55% raises the possibility of non-response schools. BMA. [Accessed 15 February 2011] Available from http://
bias (i.e. non-responders differ from responders in their socio- www.bma.org.uk/images/eocreport_tcm41-192128.pdf
demography, behaviour or attitudes). Given the anonymity of Brosnan C. 2010. Making sense of differences between medical schools
the survey we cannot quantify the effect of non- response through Bordieu’s concept of field. Med Educ 44:645–652.
Edwards J, Johnson E, Molidor J. 1990. The interview in the admission
on the generalisability of our findings to the wider population process. Acad Med 65:167–177.
of medical school interviewees. Elam C, Andykowski M. 1991. Admissions interview ratings: Relationship
Additionally, we assume that respondents have answered to applicant academic and demographic variables and interviewer
honestly. However, we appreciate that numerous factors can characteristics. Acad Med 67:865.
contribute to a response bias which can result in a question- Eva K, Rosenfeld J, Reiter H, Norman G. 2004. Admissions OSCE:
The multiple mini-interview. Med Educ 38:314–326.
naire survey over- or under-estimating the prevalence of an General Medical Council. 2009. Tomorrow’s Doctors. London: General
attitude or behaviour. Sources of response bias include Medical Council. [Accessed 14 May 2011] Available from www.gmc-
misunderstanding of the question, recall of past behaviours, uk.org/education/undergraduate/tomorrows_doctors.asp
attitudes and beliefs, and the tendency to give socially Griffin B, Harding D, Wilson I, Yeomans N. 2008. Does practice make
desirable answers (Schwartz & Oyserman 2001; Paulhus perfect? The effect of coaching and retesting on selection tests used for
admission to an Australian medical school. Med J Aust 189(5):270–273.
2002). To ameliorate the impact of response bias we have Griffin B, Wilson I. 2010. Interviewer bias in medical student selection.
avoided ambiguous terms, used simple language and pre- Med J Aust 193:343–346.
tested the questionnaire in order to determine what the Hagenaars J, McCuthcheon A. (eds.) 2002. Applied Latent Class Analysis.
respondents thinks is being asked. Bias may also have been Cambridge: Cambridge University Press.
Kulatunga-Moruzi C, Norman G. 2002. Validity of admissions measures
introduced because the results were based on the narrow
in predicting performance outcomes: The contribution of cognitive and
research design of a cross-sectional survey of an applicant noncognitive dimensions. Teach Learn Med 14:34–42.
cohort to a single institution. Lambe P, Bristow D. 2011. Predicting medical student performance from
attributes at entry: A latent class analysis. Med Educ 45:308–316.
Lambe P, Bristow D. 2010. What are the most important non-academic
attributes of good doctors? A Delphi survey of clinicians. Med Teach
Glossary of terms 32(8):e347–e354.
Lamont E, Mehta P, Nicholas J, Aston H. 2011. An evaluation of the reach
Predictive validity: The effectiveness of one set of test or for excellence programme: Cohort three and school college links view.
research results as a predictor of the outcome of future National Foundation for Educational Research. Slough NFER.
experiments or tests. Lemay J, Lockyer J, Collin T, Brownell K. 2007. Assessment of non-
cognitive traits through the admissions multiple mini-interview.
Reference: http://medical-dictionary.thefreedictionary.com
Med Educ 41:573–579.
Lemeshow S, Levy P. 1999. Sampling of Populations: Methods and
Applications. New York: Wiley.
Lord Browne. 2010. Securing a Sustainable Future for Higher Education
Notes on contributors in England: An Independent Review of Higher Education Funding &
Student Finance, 12 October,2010. [Accessed 20 May 2011] Available
PAUL LAMBE, PhD, is a selection and admissions psychometrican at from www.independent.gov.uk/browne-report
Plymouth University Peninsula Schools of Medicne and Dentistry. Lumb A, Homer M, Miller A. 2010. Equity in interviews: Do personal
DAVID BRISTOW, PhD, is vice dean of Medical Educationat Plymouth characteristics impact on admission interview scores? Med Educ
University Peninsula Schools of Medicne and Dentistry. 44:1077–1083.
738
 Do differentials in support and advice influence interview performance?

Luszczynska A, Schwartzer R. 2005. Socila cognitive theory. In: Connor M, Robb N, Dunkley L, Boynton P, Greenhalgh T. 2007. Looking for
Norman P, editors. Predicting Health Behaviour. 2nd ed. Buckingham, a better future: Identity construction in socio-economically
England: Open University. pp 127–169. deprived 16-year olds considering a career in medicine. Soc Sci Med
Maurer T, Solamon J, Troxtel D. 1998. Relationships of coaching with 65:738–754.
performance in situational employment interviews. J Appl Psychol Schwartz S. 2005. Admissions to Higher Education Review, Fair admissions
83:128–136. to higher education: Recommendations for good practise. Admissions
Maurer T, Solamon J. 2006. The science and practise of a structured to Higher Education Steering Group. Bristol: HEFCE.
employment interview program. Personnel Psychol 59:433–456. Schwarz N, Oyserman D. 2001. Asking questions about behaviour:
Mercer A. 2009. Who get in and who doesn’t? Selecting medical Cognition, communication, and questionnaire construction. Am J
students: An Australian case study. Education Research and Educ 22:127–60.
Perspectives 36(1). Streiner D, Norman G. 2008. Health Measurement Scales: a practical guide
Mercer A, Puddey I. 2011. Admission selection criteria as predictors of to their development and use. Oxford: Oxford University Press.
outcomes in an undergraduate medical course: A prospective study. Supporting Professionalism in Admissions (SPA). 2010. Defining the
Med Teach 33:997–1004. Applicant Experience. [Accessed 18 May 2011] Available from
Morris J. 1999. The value and role of the interview in the student admissions www.spa.ac.uk/documents/applicant _experience/
process: A review. Med Teach 21:473–481. Tiffin P, Dowell J, McLachlan J. 2012. Widening access to UK medical
Nicholson S. 2005. The benefits of aptitude testing for selecting medical education for under-represented socio-economic groups: Modelling the
students. BMJ 331:559–560. impact of the UKCAT in the 2009 cohort. BMJ 344:E1805–9.
O’Brien A, Harvey J, Shannon M, Lewis K, Valencia O. 2011. A comparison Tross S, Maurer T. 2008. The effect of coaching interviewees on subsequent
of multiple mini- interviews and structured interviews in a UK setting. interview performance in structured experience-based interviews.
Med Teach 33:397–402. J Occup Organ Psychol 81:589–605.
Parry J, Mathers J, Stevens A, Parsons A, Lilford R, Spurgeon P, Thomas H. Turner R, Nicholson S. 2011. Can the UK Clinical Aptitude Test (UKCAT)
2006. Admissions processes for five year medical courses at English select suitable candidates for interview? Med Educ 45:1041–1047.
schools: review. BMJ 332:1005. Universities and Colleges Admission Service (UCAS). 2012. Individual
Paulhus D. 2002. Socially desirable responding; The evolution of a institution and all institutions level Management Statistics UCAS,
construct. In: Braun H, Jackson D, Wiley D, editors. The role of Corporate Data, Policy and Research, UCAS. [Accessed 3 April 2012]
constructs in psychological and educational measurement. Mahwah, Available from http://www.ucas.ac.uk/about_us/stat_services/stats_
NJ: Lawrence ERLBAUM. pp 49–69. online/annual_datasets_to_download
Powis D. 1998. How to do it: Select medical students. BMJ 1317:1149–1150. Wiggins A, Coe R. 2009. Careers guidance and participation in education:
Reiter H, Salvatori P, Rosenfeld J, Trinh K, Eva K. 2006. The effects of Findings from the YELLIS dataset. Centre for Evaluation and Monitoring,
defined violations of test security on admissions outcomes using Durham University, UK. Available from http://www.suttontrust.com/
multiple mini-interviews. Med Educ 40(1):36–42. research/careers-guidance-and-participation-in-education/

739
Copyright of Medical Teacher is the property of Taylor & Francis Ltd and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.