Factorial Design and Analysis for Psychological Research Methodology and Statistics Course

This study investigates the impact of group membership and perceived dis-likeability on a jury's decision over a trial. The study was conducted on 60 3rd year students enrolled in the Psychological research methodology and statistics course of the University of Queensland. The results show that the accused's punishment is significantly affected by the independent variables independently as well as in conjunction.

Running head: ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
Assignment: Factorial Design and Analysis
Name of Student
Name of University
Author Note

1ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
Table of Contents
Method.............................................................................................................................................2
Results..............................................................................................................................................3
Discussion........................................................................................................................................6
Appendix..........................................................................................................................................8

2ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
Method
The study was conducted on 60 3rd year students, enrolled in the “Psychological research
methodology and statistics” course of the University of Queensland (UQ). 27% of the
participants were male, 73% were female with average age of all the participants was 20.34 years
with a standard deviation 1.47 units. The participants volunteered their participation in the study
by answering anonymously a questionnaire about plagiarism in their tutor groups.
The hypothesis under study has two parts. Whether group membership and disliking of
the one under trail by the juror would independently and in conjunction have significant effect on
their verdict or not. Due to the nature of the relevant data, the study had simulated the data by
designing a factorial experimental design with two treatment factors. Six scenarios or treatment
combinations were designed and applied randomly to respondents and their verdict as well as
perceptions of dis-likeability were recorded. The situations consisted of a female student having
been accused of plagiarism and asked to appear before a tribunal at the UQ, who had initially
denied the charges but later confessed to have unintentionally have committed the offense. The
questionnaires contained descriptions of the traits of the accused student, whether she is likeable
or not and whether she was from UQ or a transfer from Griffith University. The perceptions
about the accused’s dis-likeability was defined by a variable with three levels, namely, high,
neutral and low. There rating of the level of punishment the offender deserved was then recorded
and analysed.
The objective was to investigate how group membership and perceived dis-likeability
independently and in conjunction affects a jury’s decision over a trial. The scenarios were
manipulated to simulate the data, manipulating group membership and the emotional response

3ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
which is dis-likeability. There were thus three variables. First, the group marker representing the
situation of whether the offender is a resident or transfer student, secondly, based on the scenario
allocated to the respondent, the level of punishment that the respondent feels the offender would
deserve. The degree of punishment was defined to have seven levels. The rating 1 meant being
lenient and letting off with a warning, 2 meant applying only a mild penalty on the assignment
on which the offense was committed, 3 meant applying severe penalty, 4 meant not allowing any
credit for the assignment, 5 meant denying credit for the entire course, 6 meant academic
probation on top of denying credit for the course and 7 meant outright expulsion on top of
denying credit for the course. Note that, this served as the focal dependent variable (DV).
Thirdly, the perception of the respondents about the student based on the allocated scenario had
seven levels or response, namely, 1 which means extremely dislikeable, 2 meaning moderately
dislikeable, 3 meaning slightly dislikeable, 4 meaning neutral,5 meaning slightly likeable, 6
meaning moderately likeable and 7 meaning extremely likeable. This variable serves as the
manipulation check variable. The manipulation check was done using the variable dis-likeability
of the female as the DV and the omnibus effects were tested using the punishment level recorded
by the participants of the survey as DV. The analysis was then conducted using two-way
ANOVA.
Results
Before conducting the testing for the effects on the focal DV, manipulation check thus
consisting of a two way ANOVA using the recorded perceived dis-likeability scores from the
survey was done to verify the effectiveness of the manipulation of the dis-likeability of the
female student. The main effect of group membership, main effect of perceived likeability/ dis-
likeability and their interaction effects were considered. The SPSS code syntax is in Appendix.

4ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
The main effect of dis-likeability was found to be <0.001 and hence significant. However
the main effect of group membership was found to be insignificant at 0.05 level of significant
with 0.227 being the p-value. Interaction effects were also found to be insignificant with 0.485.
This suggests that the independent variable (IV), dis-likeability is in fact explained significantly
corresponding to the simulated values from the study. Therefore post-hoc tests were conducted to
check for whether the direction in which they ought to vary as per the underlying theory was
successfully manipulated or not. The dis-likeability variable which has three levels was therefore
to be tested via multiple pairwise tests of hypothesis. Thus Bonferroni correction was employed
to minimize errors and it was seen that significant differences exist between the levels of the
groups following the assumed order of the ordinal scale of the variable. The p-values were found
to be <0.001 for all pairwise differences. The partial Eta squared was found to be 0.71 for dis-
likeability, 0.027 for group (membership) and 0.026 for the interaction effect. The absolute
difference between high and low likeability was found to be 3.8, between high as neutral it was
1.45 and that between low and neutral, it was 2.35, all of which was seen to be significantly
different.
The focal DV, which is the ratings or scores for the level of punishment that the
participants felt the female offender deserved was then used as the DV to conduct a factorial
ANOVA to check whether the IV’s indeed have significant impact. The scenario corresponding
with the offender being a Griffith University transfer or outgroup member with high disliked
score has been seen to have the mean score 5.3 with standard deviation 1.34 and when the
offender was ingroup mean was found to be 6.2 with standard deviation 0.79. For neutral level of
disliked score, the outgroup scenario showed mean punishment level as 4.8 with standard
deviation 0.79 and for ingroup it was 3.6 with standard deviation 1.35, for low disliked, the mean

5ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
for outgroup was 5.5 with standard deviation 0.97 whereas for ingroup it was 2.3 with 0.67.The
following figure represents this graphically.
The ANOVA revealed size of the effects as explained by eta squared values was found to
be 0.39 for dis-likeability, 0.2 for group membership and 0.413 for their interaction. The effects
had p-values <0.001 and were thus significant. The analysis showed that when considering
pairwise comparisons, for high dis-likeability the test failed to reject that hypothesis that there is
significant difference based on group membership with p-value >0.05 however, for low and
neutral dis-likeability levels, the differences were found to be significant with p-values <0.001

6ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
and equal to 0.011 respectively. The post hoc test done using Bonferroni’s tests, Pairwise
comparisons were done to follow up the ANOVA. Bonferroni correction was used to minimize
errors. The syntax for the analysis in SPSS is provided in the Appendix section. Significant
difference based on group memberships was apparent due to it having only 2 groups and for dis-
likeability being considered independently, pairwise differences for high and low and high and
neutral dis-likeability were observed to be significant with <0.001 p-value. The test for
difference between neutral and low failed to reject that hypothesis that there is no difference with
p-value being >0.999.
Discussion
The study aims to investigate whether the fact that the accused belongs to same social or
communal group as the jurors and the perceptions of dis-likeability or likeability of the jurors
about the accused independently or conjunctively have any significant impact on the final verdict
about the accused’s punishment. The results was based on simulated data, where two of the
independent variables were simulated using six hypothetical scenarios and the respondent’s
response based on them. There were two manipulated variables, group affiliation (ingroup or
outgroup) and emotional response in the form of perception of dis-likeability which the study
tried to invoke through introducing three personas of the accused. The later was checked for
validity since it could not be directly manipulated and the manipulation methodology was found
to be successful. The analysis of the main focal dependent variable that is the juror’s decision
regarding punishment was then found to be significantly affected by the independent variables
independently as well as in conjunction. Comparing the difference in decisions as per the
allocated scenarios, it was seen that when ingroup, the accused was overall subjected to less
harsher punishment than outgroup, and the more liked the more lenient is the verdict. This

7ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
corresponds with what Louis, Lalonde and Esses (2010) had said, by citing prejudice as the key
reason. However for ingroup it was seen that if the accused was dislikeable, the verdict is harsher
than when accused in from outgroup. This falls in line with what Wallace and Appleby (1983)
described as the “black sheep” effect. This means that perhaps to avoid bias, a jury for a
particular trail should make sure to have homogeneous set of jurors from all backgrounds who
are known to be good judge of character. A key strength of the study is that being a factorial
design, it allows for the interactions and the various levels of interaction to be studied. However
since the data was simulated and includes human emotions as a variable, it leaves room for doubt
as human emotions are subjective to situation. A larger study using data from actual trials could
thus be considered for future research.

8ASSIGNMENT: FACTORIAL DESIGN AND ANALYSIS
Appendix
Manipulation Check
UNIANOVA Percieved_dislike BY Dislikeability Group
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/POSTHOC=Dislikeability(BONFERRONI)
/PRINT=ETASQ
/CRITERIA=ALPHA(.05)
/DESIGN=Dislikeability Group Dislikeability*Group.
ANOVA for Focal Dependent Variable
UNIANOVA Score BY Dislikeability Group
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/POSTHOC=Dislikeability Group(BONFERRONI)
/EMMEANS=TABLES(Dislikeability*Group) COMPARE(Group) ADJ(BONFERRONI)
/PRINT=ETASQ HOMOGENEITY DESCRIPTIVE
/CRITERIA=ALPHA(.05)
/DESIGN=Dislikeability Group Dislikeability*Group.
GRAPH
/BAR(GROUPED)=MEAN(Score) BY Dislikeability BY Group
/INTERVAL SD(2.0)
/TITLE='Plot comparing mean level of punishment ' 'as a function of
dis-likeability and group membership'.