ERP for IB

Robustness test: analyzing the same hypothesis in different ways, f.e. age OR age groups
will give different results

Validity: Extent to which a measure correctly represents the concept of study

- Internal: how well is the study done
- External: generalizability of results

Reliability: Every time you measure you should get the same result

Accuracy: Measure is close to the actual value

1. You estimated the regression equation using a sample of 1250 firms in one country,
observed yearly for 6 firm characteristics, over the period between 2015 and 2020.
What type of data you use, explain.
- Panel Data bc mulitple years and multiple characteristics.

2. A researcher conducts a study about happiness. He wants to understand whether

having a side job has a relationship with subjective happiness among university
students in the Netherlands. He selects a random sample of 38,940 students, from all
universities in the Netherlands. Does this researcher run a risk of sample selection
bias? Motivate your answer.
- No because there is a large sample from all universities in the Netherlands and random
selection was done.

3. A team of researchers is interested in researching CEOsʼ transformational leadership

(TFL) behaviours and their impact on firm performance.
 CEO TFL behaviours are measured on a Likert scale from 1 to 7, where higher scores
correspond to more TFL behaviours. In addition to the data on CEO TFL behaviours, the
following two variables are also used in the research: Firm size and Founder CEO. Firm
size is measured as the number of full time employees in the firm. Founder CEO takes the
value of 1 if the CEO is also the founder of the firm, and 0 otherwise.

Describe the data types of the variables

i. CEO TFL behaviour variables from the questionnaire, ii. firm size, and iii. Founder CEO.
i. Ordinal the exact difference between categories is unknown
ii. Ratio, bc true zero
iii. Nominal bc only 2 catagories and dummy variable

Tutorial 1:

1. Explain the difference between primary and secondary data.

Primary is collected by yourself, and secondary is available on the internet.

2. In an IB research project on performance, how do you determine which type(s) of

performance measures to include?

 Defining research objectives and selecting key performance indicators (KPIs) that
align with these objectives, considering industry context, balancing quantitative and
qualitative measures, and ensuring data availability and reliability.
 depends on the research questions, there is no universal measure
 three levels, financial, operational, overall performance

3. Do we have to measure performance at all levels of analysis?

 Different levels: Firm, SBU, subsidiary, country

 Depends on the research objectives/ scope of the study

4. The following questions are based on a statement in the paper by Hult et al.
(2008), "...it is important to note while restricting the sample to firms adopting
harmonized standards may introduce selection bias issues: for example, if the
population of firms that adopt harmonized standards is fundamentally different
from those who do not, researchers must interpret the implications of their
findings conservatively".

4.1. Define selection bias (please discuss an example).

 Selection bias occurs when certain members of a population are systematically

excluded from a study's sample, leading to results that do not accurately represent
the entire population. For example, if a study on smartphone usage only surveys
people who own smartphones, it may not accurately reflect the preferences and
 behaviors of individuals who do not own smartphones, thus introducing selection
bias.
 F.e. when comparing south Korean risk taking to north Korean, but you only include
refugees from north korea (which are already more risk taking since they escaped
NK)

4.2. The following statement is taken from Du, Deloof and Jorissen (2015: 174): "We
design the questionnaire [...] In the initial questionnaire survey (conducted December
2008) and three follow-ups, 87 (21%) questionnaires are returned [...] We check the
nonresponse bias by comparing the size, industry, and headquarters country of our
respondent subsidiaries with the research population. Results from a t-test show that, on
average, the size of our respondent subsidiaries is not significantly different from that of
the research population (t = .73, p= .46)."

What is the difference between the approaches of Hult et al. (2008) and Du et al. (2015)?
Evaluate the difference.

 Hult et al:Comparing 2 sub-samples, not with the population (e.g. comparing group A
and B)
 Du et al: comparing a sub sample to the whole research population (e.g. comparing
group A to group A+B)

5.1. “The United States Steel Corporation once said that its employees’ average weekly
earnings went up 107 per cent between 1940 and 1948”. Does this mean there was a huge
wage raise?

No, it could have been either due to inflation as well as world war 2 which needed a lot of
steel for military purposes -> more working hours.

5.2. Discuss each of the following statements. What impression is given? Does it fit the
facts?

5.2.1. “27 per cent of a large sample of eminent physicians smoke Throaties - more than any
other brand.”

 it gives the impression that Throaties is a popular choice among physicians,

potentially influencing perceptions of the brand's quality or endorsement by medical
professionals

5.2.2. “More people were killed by airplanes last year than in 1910.”

 There are way more planes now than in 1910

5.2.3. “The death rate in the Navy during the Spanish-American war was nine per thousand.
For civilians in New York City during the same period, it was sixteen per thousand.”

 Suggests that being in the navy was safer than being a civilian in New York during
this time period
6. With the information provided on pages 1325-26 (where the variables in the model are
described) of the article by Aybar and Ficici (2009), answer the following questions:

6.1. Which variables in the analysis are dichotomous or dummy variables? Define the
category that takes a value equal to 1.

6.2. Which variables are continuous variables?

- interval, ratio

BONUS QUIZ QUESTION 1: In a research about wage differences across industries, how
would you measure wage? Whose wage information should be collected?

- Only entry level wage average

- Divide per industry

BONUS QUIZ QUESTION 2: A team of researchers examines the implications of CEO

foreignness for a firm’s corporate social performance (CSP), and how firm’s profitability is
associated with this relationship.

For the research, CSP information is obtained from a data source, where firms’
environmental, social and governance actions are scored. Researchers use “Social Pillar
Score”, which measures a firm’s capacity to generate trust and loyalty with its workforce,
customers, and society, through its use of good management practices. Each firm’s score is
indicated on a scale from 1 to 5 (i.e., higher scores correspond to better CSP).

Firm profitability information is computed using data form firm financial tables as the return
on assets (ROA), corresponding to earnings before interest and taxes, divided by total
assets.

Describe the data types of the variables used in the hypotheses. Explain your answer by
discussing the characteristics of the respective data types.

i. Firm profitability (ROA) Ratio scale, because it is a number which can

potentially be zero
ii. ii. CSP (social pillar score) Ordinal scale, because it is on a scale from 1-5
Tutorial 3:

BONUS QUESTION 1

Answer the following questions (based on Table 2 in Aybar and Ficici,

2009).

i. Column 5 in Table 2 gives the kurtosis values of the

variables introduced in the table. What does the kurtosis
measure tell us? What can you conclude by looking at the
kurtosis values provided in Table 2 (provide the name of
variables)?
i. Kurtosis is a statistical measure that describes the shape, or
peakedness, of a probability distribution's tails in relation to
the normal distribution. In simpler terms, it measures how
much data is concentrated around the mean, compared to the
tails.
ii. FS/TS, total assets and foreign assets have lighter tails
Total sales has a heavy tail
Foreign sales and FA/TA has a relatively normal tail

ii. The last column of Table 2 provides the skewness values of

the different firm size variables. What does the skewness
measure tell us? What can you say about the skewness of
variable foreign sales?
i. Skewness is a statistical measure that describes the
asymmetry of a probability distribution around its mean. It
indicates whether the data points are concentrated more on
one side of the mean compared to the other.
ii. FS/TS is relatively normally skewed, the others are all
substantially positively skewed

1. Which main hypotheses did you find in the Kleinknecht,

Kwee, and Budyanto (2016) paper? (Hint: the question is not
about the control variables)

Using higher shares of temporary workers, manpower agency

workers and self-employed (freelance) workers is related to heavier
management bureaucracies.

2. Answer the following questions (based on Table 3 in

Kleinknecht, Kwee, and Budyanto, 2016)

2.1. What is the number of observations?

1458-2861, varies from variable to variable
 2.2. Predict the maximum number of observations in
the analyses (using Table 3).
the lowest decides, 1458

2.3. Compare your prediction with Table 4. Do they

match?
yes

2.4. In Table 3, is there any number you think is unrealistic

(or which figure in the table strikes you and why)?
maximum of 180% makes no sense bc 100 should be the maximum

3. See Table 4 showing regression results. Which regression

technique was used in the paper? Please provide reasons for
your answer (mention what is the dependent variable).
1. Ordinary least square regression, size of management
bureaucracies
4. What are the model characteristics, as seen in Table 4?
Define them and explain Table 4 [hint. Consider the
statistics reported in the table, the conclusions you make
from them, and the statistics missing in the table.]
1. Use of dummy variables, in each model one extra dummy is
added to see if there are any synergy effects

 R squared tells us the percentage of variance in the outcome that is

explained by the linear combination of variables
 Increases as we keep adding variables
 To avoid this, use adjusted R squared, which takes into account the
number of independent variables and the sample size
 When comparing, look for the model with the highest adjusted R
square
 F-statistics tests whether all coefficients jointly equal zero (the null
hypothesis)/ or jointly differ from zero (the alternative hypothesis) -
> should be significant

5. How is Table 4 organized? (i.e., what are the different models, and
why do the authors provide different models?). What is the most
important model for the paper?
1. Each adds a new industry
6. Which industry dummy is missing in Table 4 and why?
1. Agriculture, as it is the baseline for the model
 7. Interpret the coefficients of ‘public administration’ and ‘flexible
labour’ in Model 4 of Table 4.

DV: management ratio, Public administration = dummy, flexible labor:

factor score on labor flexibility

4.74% points lower than agriculture, significant at 5% level

One unit higher flexible labor will lead to a 1.68% point higher
management ratio, significant at 1% level
=5

a. Can you determine the presence of outliers in this sample,

using the standard deviation rule?
1. yes
b. While answering (a) using the standard deviation rule, which
of the calculated values are not feasible and why?

Because it is about investments negative values are not possible because

you cannot invest negative money
8. The standard deviation method can be described as ‘quick
and dirty’. Why that qualification?
1. It assumes the data follows a normal distribution which is not
always the case, and it is very subjective and context
dependent.
9. Upon examining the data in detail, suppose you find that the
data of firms in the oil industry constitute the outlier in the
sample. You then face the problem of how to deal with it.
Construct a decision tree of acceptable ways how to handle
the problem. What is the first thing to determine in such a
case? Proceed from here and figure out the research
strategy for the relevant situations. What is a general rule in
such a case?
1. Lecture slides
10.In the paper by Lin, Xu and Xie (2023),

1. What is the research question in the paper by Lin, Xu and

Xie, and what hypotheses do they test? Show the
hypotheses in a conceptual model (draw a figure).

RQ: How does cultural diversity of semi cultural teams influence

the performance of semi-virtual teams?

H1: The relationship between cultural diversity and the strategy

quality of a semi-virtual team is more likely to be positive when
 the game rules change shortly before a game commences.

H2: The relationship between cultural diversity and the strategy

quality of a semi-virtual team is more likely to be positive when
team members on average control more virtual characters.

H3: The relationship between cultural diversity and the strategy

quality of a semi-virtual team is more likely to be positive when a
multicultural team is playing at home court.

2. How do they measure ‘cultural diversity’?

We used Blau’s heterogeneity index to measure a team’s cultural

diversity (Blau, 1977):

Do they check their method with alternative approaches? no

Tutorial 4
Answer the following questions on the sample used:

1.1. Describe the sample used in this analysis.

Survey data from a sample of MNE subsidiaries operating in Belgium with headquarters
in 14 different countries
 50 employees, non financial industries

1.2. What is the final size of the sample? And the respondents’ sample?

Final size is 428, the respondents’ is 87 but 4 are deleted so 83

Answer the following questions (based on Table 2):

2.1. What is the number of observations used per variable?
83, except for HQ country director, which is 60

2.2. Can you identify the dummy variables?

Local implementer, wholly owned, anglo-saxon headquarters, Japanese headquarters

2.3. What does the value of the ‘mean’ represent in the case of dummy variables?
proportion of the cases where the variable is 1

2.4. Which role does that the sample of subsidiaries’ boards seem to perform the most?

Independent variable, coordination role

Answer the following questions (based on Table 3):

3.1. Explain the value 0.35*** between subsidiary size and service role.
positive and significant correlation between subsidiary size and service role. As the
subsidiary increases the service role also increases

3.2. Is there any indication of multicollinearity? Explain how you checked for it.

no because there is no correlation > 0.7 except for the dependent variable

Answer the following questions (based on Table 4):

4.1. Explain the setup of Table 4.

The table is setup to show the differences between each role, horizontal is dependent variable,
vertical is independent
BONUS QUESTION 1

Write down the estimated regression equation corresponding to

Model 2 in Table 4.

Y^=1.091X1+1.149X2−0.108X3−0.654X4−0.036X5+2.783X6

Answer the following questions regarding heteroscedasticity:

5.1. Is there evidence of heteroscedasticity in the analysis?

They correct for possible heteroscedasticity so we don’t know

5.2. Is there something the authors do to correct for

heteroscedasticity?

They use robust standard errors to correct for heteroscedasticity

5.3. Do you expect the results reported in a regression table

(such as, for example, Table 4) to differ if heteroscedasticity is
corrected for, compared to the case in which heteroscedasticity is
present yet not corrected for? If so, how do you expect the
results to differ and why?
  Yes, I would expect more precise coefficient estimates when it is corrected for.
OLS assumption (of constant error variance) violated
 ● Biased standard errors,
 ● unreliable t-statistics,
 ● unreliable significance tests
 ● Misleading conclusions about significance

BONUS QUESTION 2

The following questions are based on Aybar and Ficici (2009).

i. Can you explain what the difference is between the columns of the
results table (Table 6)?

They each account for different windows, from larger to smaller size

ii. Looking at the results provided in panel A of Table 6, interpret the

coefficient of the variable TSTATUS in the specification of the model
with the dependent variable SCAR(- 5,+5).
The linear regression value of 0.0135*** indicates that a one-unit change
leads to an increase of 0.0135 which is statistically significant at the 1%
level

iii. Can you select the best model specification with the information
provided in Table 6?

No bc there is no F Rsquared or adjusted Rsquared

 Tutorial 5

1. In their paper, Kleinknecht, Kwee, & Budyanto, (2016) acknowledge

that they “have to be cautious about causal inferences”.

1.1. Why do the authors need to be cautious about their inferences in this
analysis?

 The paper acknowledges the need for caution due to the absence of
a time lag between dependent and independent variables. This lack
of temporal precedence makes it difficult to determine the direction
of causality.

1.2. How did the authors address the causality issue mentioned in the
above statement?

 They used firm-level survey data from the Netherlands, which

allowed them to control for various influential factors within a single
country context.
 The authors employed regression analysis to estimate the
relationships between variables while controlling for potential
confounding factors.
 They discussed the limitations of their approach, including the
absence of a time lag between dependent and independent
variables, and conducted sensitivity analyses.

1.3 What are the requirements for causation in general?

 Temporal precedence: The cause must precede the effect in time.

 Covariation: There should be a consistent relationship between the
cause and effect.
 Elimination of alternative explanations: Other potential factors that
could explain the relationship between the cause and effect should
be ruled out or controlled for.
 Mechanism: There should be a plausible explanation or mechanism
underlying how the cause leads to the effect.
 Consistency: The relationship between the cause and effect should
be observed consistently across different studies, populations, and
conditions.

2. BONUS QUESTION

A researcher in the ministry of education would like to test the following

hypothesis: “Children from high-income families perform better in middle
school exams.” She collected data for 6000 children. They were randomly
sampled from all children who were in their first year of middle school
education in 2018. The study variables are:

Math: Score (in points, from 0 to 100) of the child in the first-year middle
school’s final math exam in year 2018. (Note: the exam is centrally
organized and the questions are the same for all middle schools.)

Gender: Gender of the child in administrative records where 0 = Boy, 1 =

Girl. Height: Height of the child (in centimeters).
Lnincome: Net family income (in logarithm).
Leaves: Days of leave in the school year.
Sports: Time spent in doing sports (average number of hours per week).

1. a) Write down the estimation equation for regression analysis.

Yi=β0+β1X1i+β2X2i+β3X3i+β4X4i+β5X5i+ϵi

Where:

 YiYi represents the dependent variable (math score).

 X1iX1i, X2iX2i, X3iX3i, X4iX4i, and X5iX5i represent the independent variables
(gender, height, lnincome, leaves, sports respectively) for the iith observation.
 β0β0 is the intercept term.
 β1,β2,β3,β4,β5β1,β2,β3,β4,β5 are the coefficients of the independent variables.
 ϵiϵi is the error term.

2. b) One month after the initial research on math scores, the

researcher managed to obtain the permission from the ministry of
education to access the final exam scores of Science and Literacy
for the same 6000 children in 2018. She adds these two scores to
her existing dataset to create a new dataset for her future research.
What type of data (time series, cross- sectional, or panel) does she
have in this new dataset? Explain.

 Cross-sectional data is collected at a single point in time from multiple individuals,

units, or subjects. In this case, the researcher collected data on math, science, and
literacy scores for 6000 children in the year 2018. The data does not involve
observations over time (time series) nor does it track the same individuals over time
(panel data). Instead, it provides a snapshot of the performance of these children
across different subjects at a specific point in time. Therefore, it qualifies as cross-
sectional data.

3. The following questions are based on a dataset used in the book

"Principles of Econometrics" by Hill, Griffiths, and Lim. The dataset
provides survey data on people's wages, education, experience, and
personal and US regional characteristics. Table 2 below provides
information on the main variables of the dataset, and Table 3 on the
descriptive statistics.

Looking at the information provided in Tables 2 and 3 answer the

following questions:

3.1. Classify the variables according to the following criteria: education

characteristics, labour characteristics, individual characteristics, urban
characteristics, and regional characteristics.

Education: BA, AA, jc, univ, stotal

Labour: lwage, exper
Individual: female, black, hispanic
Urban: smcity, medcity, submed, lgcity, sublg, vlgcity, subvlg
Regional: ne, nc, south

3.2. There are some dummy variables in the dataset. You can see the
variable “female” and the variables “hispanic” and “black”. Why are there
two variables for “hispanic” and “black” and only one for “female”?

Because there are more than 2 alternatives, where gender generally only has male and female

3.3. How many subjects are recorded as “female” in the sample?

51,95%
4. Have a look at Table 4 and Figure 1 below:

4.1. Describe the distribution of the variable 𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒 in terms of its

symmetry with the information provided in Table 4.

𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒?
4.2. What does the Kurtosis value in Table 4 tell you about the variable

a kurtosis value of 3.3 indicates that the distribution of the variable "experience" has
moderately heavy tails and might be slightly more peaked than a normal distribution, but it is
overall mesokurtic.

4.3. What is the range of the variable 𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒?

166-3 = 163
4.4. Use Figure 1 to answer this question: Keeping in mind that category

𝑒𝑥𝑝𝑒𝑟𝑖𝑒𝑛𝑐𝑒 is different among genders?

1 is female, can you conclude that the distribution of the variable

Yes, males tend to have a higher density when experience increases

5.1. Select the best model. Explain the statistics that you have chosen to
select between these models. Justify your answer.

Model 3 because it has the most significant variables. The additional

variables that are added are meaningful as seen by the adjusted r
squared.

5.2. The survey was held in a specific year; all variables were measured
at the same time. Given that the data are collected at the same point in
time, what might be a principal problem?

Causality or directionality.

5.3. Interpret the regression coefficient of female in model 3.

-0.23363083 implies that being female decreases the predicted value of lwage. Each unit
increase in female decreases lwage by 0.23363083 units. This is significant at the 99.9% level

5.4. Wage is taken as the dependent variable in the regression equation.

Is that reasonable or not?

yes

6.1. The regression equation to be estimated is 𝑙𝑛(𝑤𝑎𝑔𝑒) = 𝛽􏰀 + 𝛽􏰁 𝑒𝑑𝑢𝑐

+ 𝜀. Which variables were used in each of the two graphs?
6.2. Which is the official graph for detecting heteroscedasticity?

The official graph for detecting heteroscedasticity in regression analysis is the scatterplot of
residuals versus fitted values or the scatterplot of squared residuals versus fitted values.

𝑤𝑎𝑔𝑒 and 𝑒𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛.

6.3. Look at the first graph (Figure A). Explain the relationship between

Higher education results in a higher wage

6.4. Explain the heteroscedasticity using the variables 𝑤𝑎𝑔𝑒 and

𝑒𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛.

Tutorial 6

1. What is a moderator? What is the moderator in Rhee and

Cheng (2002)?

A variable that affects the relationship between two other variables.

Organizational slack, RR, locational advantage
Example:

2. Knowing what a moderator is, what is a mediator? Is it the

same as moderator?

A mediator explains the relationship between two variables, which is not

the same as the moderator which affects the relationship.

Example:
 3. Table 1 in Rhee and Cheng (2002) presents correlations
among variables. How should the correlation matrix be
examined to detect possible multicollinearity?

Looking for correlations >.5 or <-.5 (conservative) (liberal .7)

4. In Table 2 of the paper, why have the authors analyzed two

different specifications of the same model?

They split the variable foreign market uncertainty into cultural and
operational to prevent multicollinearity issues

5. Looking at Table 3, which control variables have been used

in the model? Do all control variables have a significant
effect in the regression model?

First define DV and main IV then moderators, the rest is control

Firm size, age and market size

6. Hypothesis 1 states: “Foreign market uncertainty will have a

positive effect on firms’ involvement in incremental
international expansion”. Based on Table 2, can we conclude
that there is enough evidence to support Hypothesis 1?
Yes, b=0.361**, 0.598*** both positive and significant

7. Explain the set-up of Table 3.

3 hypotheses, tested based on cultural and operational uncertainty,

separated interaction with moderator in each model

8. The following statement is taken from the paper by Barkema

& Drogendijk (2007, p. 1141): ‘if the interaction effect is
included, the main effect should be interpreted as a
conditional effect rather than as an independent effect on
the dependent variable’. What do the authors mean by that?

They mean that the main / conditional effect varies depending on the level
of interaction with other variables

BONUS QUESTION 1

Scholars suggest that CEOs’ transformational leadership (TFL) behaviors

will have a positive effect on firm performance, and this effect will be
weaker in large firms, and stronger in firms with founder CEOs.

1. a) Write down the regression equation to test the suggested

relationships. You can use 'b1' ‘b2’ ‘b3’ etc. to represent the
regression coefficients. You can use short names for variables, just
explain them in your answer.

Firm Performance=b0+b1×TFL+b2×Firm Size+b3×Founder CEO+b4×(TFL×Firm Size)+b5

×(TFL×Founder CEO)+ε

2. b) What would be the marginal effect of CEO’s TFL behaviors on

firm performance? Show using the regression coefficients (e.g., b1,
etc.) that you used in the previous question.

b1 + b4xFirm Size + b5xFounder CEO

9. Based on the results presented in Table 3, evaluate whether

hypotheses 2-4 are confirmed, in Rhee & Cheng (2002).

Hypothesis 2: no bc the results are not significant (otherwise yes)

Hypothesis 3: partially, bc the result in model 5 is significant but in 2 it is not
Hypothesis 4: partially, bc the result in model 6 is significant but in 3 it is not

10. Explain the relationship between foreign market

uncertainty and incremental expansion when locational
advantages are considered (see Model 6 in in Table 3).

B= -0.029* meaning that there is a slightly negative relationship when locational advantages
are considered.
BONUS QUESTION 2

Theory suggests that the effect of CEO TFL behaviors on firm performance
will realize through the commitment of employees in firm’s vision, such
that CEO’s TFL behaviors make employees feel more committed to the
vision of the firm and therefore the firm achieves higher performance.
What is the role of employee commitment in this relationship?

Employee commitment is a mediator that links CEO TFL behaviors to firm performance
 Tutorial 7
All questions are based on Yiu, Lau & Bruton (2007)

1. What is the sample of the study?

Final sample 274 firms

2. Why and how did the authors use two waves of surveys in
the study?
1. Through cooperation with the state statistics bureau of china,
both waves asked about different things to the same CEO’s
2. Why: method used to test for causal relationships
3. List down the independent, dependent, mediating and
moderating variables in the study.
1. Dependent: International venturing activities
2. Independent: Technological capabilities, management
capabilities, home country network ties.
3. Mediating: Corporate entrepreneurship
4. Moderating: home industry competition, export intensity
4. Explain in detail how the first independent variable
(technological capabilities) is measured.
1. The three-year average R&D expenditures to capital
investment
2. Number of rewards for innovativeness by government or trade
associations, number of government contracts, or sponsored
research, and number of collaboration programs with
universities or research institutes over the past 5-year period.
5. What is not shown in the conceptual model/framework of the
paper?
1. Control variables
6. What is the difference between exploratory and
confirmatory factor analysis? Which one is used in the
study?
1. Exploratory, EFA is used to explore and uncover the
underlying structure of a set of variables, while CFA is used to
confirm or reject a specific factor structure based on prior
hypotheses or theories
7. See Table 1 on page 530. What is the Cronbach`s alpha on
this table used for? How do you interpret the Cronbach`s
alpha for the DV? What is the threshold for that value in
general?
1. It is used to determine the internal consistency or reliability of
a scale or test. Threshold is usually 0.7. (range: 0-1)
2. 0.91 indicates that the items in de DV are measuring the
same construct consistently.
 8. What is the difference between orthogonal (unrelated) and
oblique rotation?
1. the choice between orthogonal and oblique rotation depends
on the underlying assumptions about the relationships
between factors and the goals of the analysis. Orthogonal
rotation is preferred when factors are assumed to be
unrelated, whereas oblique rotation is more appropriate when
factors are expected to be correlated.

9. Consider the following matrix of factor loadings from a

factor analysis:

What can you conclude from this matrix about:

(a) the number of variables, 6

(b) the number of factors, 2

(c) the relationship between variables and factors, and

First three factors are loaded to the second factor, the second three
are loaded to the first

(d) give a graphical representation of the factors. In your graph, clearly

indicate all factors and variables.
BONUS QUESTION

A team of researchers examines the implications of CEO foreignness for a

firm’s corporate social performance (CSP), and how firm’s profitability is
associated with this relationship.

The researchers further investigate the relationship between firms’

corporate social performance (CSP) and corporate reputation. To measure
a firm’s corporate reputation, they survey firm’s various stakeholders
(customers, employees, suppliers, investors, etc.).

Using the responses from stakeholder surveys, their finalized total of 20

items was factor analyzed into six dimensions, as shown in the table
below.

A: Exploratory
B: moderator