Quantitative Data Analysis 2: Final Exam Responses
(Grade: 8.9)
Q01. Planned contrast test (9/10 points)
• H1: Dutch students who do their BSc at UVA have higher intentions to do their MSc at UVA
than any other of the three groups. -- > one-sided hypothesis!
• Levene's test
o H0: equal variances assumed - p-value = 0.033 - p-value < alpha = 0.05, therefore we
reject H0 - we cannot assume equal variances.
• Contrast test
o H1: Dutch students who do their BSc at UVA have higher intentions to do their MSc
at UVA than any other of the three groups.
o one-sided hypothesis
o p-value (do not assume equal variances) = 0.000 (but this is two sided)
o one-sided p-value = 0.000/2 = 0.000 (because in line with hypothesis)
sample mean for Dutch, BSc at UvA is higher than all the other means, so in
line with hypothesis
o p-value < alpha = 0.05, therefore we reject H0 (support H1)
• Based on the p-value and an alpha = 5%, there is sufficient statitsical evidence that the
Dutch students who do their BSc at UVA have higher intentions to do their MSc at UVA than
any other of the three groups.
Q02. Interaction in Factorial ANOVA (9/10 points)
• Interaction effect?
• the graph indicates that there is a potential interaction effect because the two lines are not
parallel (they have different slopes)
• based on the spss output, the interaction effect (BScUVA* Nationality) has a p-value = 0.003
o H0: there is no interaction effect between BScUvA and Nationality.
o p-value < alpha = 0.05, therefore we reject H0
o Based on the p-value and alpha =5%, there is sufficient statistical evidence that
there is a significant interaction effect between BScUvA and Nationality.
• Graph
o We can see this interaction effect as the two lines have different slopes.
o When BsC UvA is yes, Intention MSc UvA is higher for the Dutch than for foreign
students.
o When BsC UvA is no, Intention MSc UvA is higher for the foreign than for Dutch
students.
o extreme interaction effect -- > direction of the effect changes
• Hypothesis
, o There is a significant interaction between BScUvA and Nationality, such
when students do their Bachelor program at Uva, the intention to do the
Masters at UvA is higher when the student is Dutch than when they are
foreign
and when students do not do their Bachelor program at Uva, the intention
to do the Masters at UvA is higher when the student is foreign than when
they are Dutch
o The difference in doing your masters at UvA between people who did do their
bachelor at UvA and people who did not doing theit bachelor at UvA is larger for
dutch people than for foreign people.
Q03. Regression coefficients (10/10 points)
• Unstandardised B Coefficients = Std. Error * t
o AttitudeAmsterdam: Unst. B = 0.077 * 3.264 = 0.251328
o NonEU: Unst. B = 0.282 * -2.121 = -0.598122
• Interpretation
o As PV increases by 1 unit, OV increases by Unst. B
o AttitudeAmsterdam: As AttitudeAmsterdam increases by 1 point (7-point likert
scale), IntentionMSc increases by 0.251328 points (7-point likert scale)
o NonEU: When the student is NonEU, IntentionMSc decreases by -0.598122 points
(7-point likert scale) compared to when they are Dutch.
• Effect Strength
o Use standardised beta to compare PVs effect sizes -- > look at absolute values of
standardised beta -- > higher value = stronger effect
o Standardised beta does not have an issue of scaling because it used standard
deviations (instead of units)
o The effect of PV AttitudeAmsterdam on OV IntentionMSc is not the strongest effect.
(standardised b = 0.219)
o The effect of PV UncertaintyAvoidance on OV IntentionMSc is the strongest effect.
(standardised b = -0.444)
o The effect of PV Education on OV IntentionMSc is the weakest effect. (standardised
b = -0.025)
o AttitudeAmsterdam has a stronger effect compared to Education, but a weaker
effect compared to UncertaintyAvoidance.
(Grade: 8.9)
Q01. Planned contrast test (9/10 points)
• H1: Dutch students who do their BSc at UVA have higher intentions to do their MSc at UVA
than any other of the three groups. -- > one-sided hypothesis!
• Levene's test
o H0: equal variances assumed - p-value = 0.033 - p-value < alpha = 0.05, therefore we
reject H0 - we cannot assume equal variances.
• Contrast test
o H1: Dutch students who do their BSc at UVA have higher intentions to do their MSc
at UVA than any other of the three groups.
o one-sided hypothesis
o p-value (do not assume equal variances) = 0.000 (but this is two sided)
o one-sided p-value = 0.000/2 = 0.000 (because in line with hypothesis)
sample mean for Dutch, BSc at UvA is higher than all the other means, so in
line with hypothesis
o p-value < alpha = 0.05, therefore we reject H0 (support H1)
• Based on the p-value and an alpha = 5%, there is sufficient statitsical evidence that the
Dutch students who do their BSc at UVA have higher intentions to do their MSc at UVA than
any other of the three groups.
Q02. Interaction in Factorial ANOVA (9/10 points)
• Interaction effect?
• the graph indicates that there is a potential interaction effect because the two lines are not
parallel (they have different slopes)
• based on the spss output, the interaction effect (BScUVA* Nationality) has a p-value = 0.003
o H0: there is no interaction effect between BScUvA and Nationality.
o p-value < alpha = 0.05, therefore we reject H0
o Based on the p-value and alpha =5%, there is sufficient statistical evidence that
there is a significant interaction effect between BScUvA and Nationality.
• Graph
o We can see this interaction effect as the two lines have different slopes.
o When BsC UvA is yes, Intention MSc UvA is higher for the Dutch than for foreign
students.
o When BsC UvA is no, Intention MSc UvA is higher for the foreign than for Dutch
students.
o extreme interaction effect -- > direction of the effect changes
• Hypothesis
, o There is a significant interaction between BScUvA and Nationality, such
when students do their Bachelor program at Uva, the intention to do the
Masters at UvA is higher when the student is Dutch than when they are
foreign
and when students do not do their Bachelor program at Uva, the intention
to do the Masters at UvA is higher when the student is foreign than when
they are Dutch
o The difference in doing your masters at UvA between people who did do their
bachelor at UvA and people who did not doing theit bachelor at UvA is larger for
dutch people than for foreign people.
Q03. Regression coefficients (10/10 points)
• Unstandardised B Coefficients = Std. Error * t
o AttitudeAmsterdam: Unst. B = 0.077 * 3.264 = 0.251328
o NonEU: Unst. B = 0.282 * -2.121 = -0.598122
• Interpretation
o As PV increases by 1 unit, OV increases by Unst. B
o AttitudeAmsterdam: As AttitudeAmsterdam increases by 1 point (7-point likert
scale), IntentionMSc increases by 0.251328 points (7-point likert scale)
o NonEU: When the student is NonEU, IntentionMSc decreases by -0.598122 points
(7-point likert scale) compared to when they are Dutch.
• Effect Strength
o Use standardised beta to compare PVs effect sizes -- > look at absolute values of
standardised beta -- > higher value = stronger effect
o Standardised beta does not have an issue of scaling because it used standard
deviations (instead of units)
o The effect of PV AttitudeAmsterdam on OV IntentionMSc is not the strongest effect.
(standardised b = 0.219)
o The effect of PV UncertaintyAvoidance on OV IntentionMSc is the strongest effect.
(standardised b = -0.444)
o The effect of PV Education on OV IntentionMSc is the weakest effect. (standardised
b = -0.025)
o AttitudeAmsterdam has a stronger effect compared to Education, but a weaker
effect compared to UncertaintyAvoidance.
