Observational Study Data Analysis Plan Template (Page 2 of 3) in pdf

III. Main Analyses - Pearson’s Chi Square Test of Association Between Discrete

Variables

- Each hypothesis will have to be tested separately.

1) Group Size x Help Hypothesis: Individuals are more likely to give help than groups of two or more.

- Analyze -> Descriptive Statistics -> Crosstabs

- Rows: Always put the variable with the most groups in the Rows (In this case, it is Group Size).

- Columns: Always put the variable with fewest groups in the Columns (in this case, it is Helping

Behavior). This simply keeps the output a little more interpretable (unless you print your output in

landscape, then you should reverse the above rules).

- Statistics: Chi-Square, Phi and Cramer V

- Cells: All Counts, All Percentages, and Unstandardized Residuals

- Paste to Syntax Sheet

- Syntax for this hypothesis:

CROSSTABS

/TABLES=grpsize BY help

/FORMAT= AVALUE TABLES

/STATISTIC=CHISQ PHI

/CELLS= COUNT EXPECTED ROW COLUMN TOTAL RESID .

2) Confederate Gender x Help Hypothesis: Female Confederates are more likely to receive help than Male

Confederates.

- Once you get the syntax for one analysis you can copy and paste it and just change the variables

in the syntax:

CROSSTABS

/TABLES=congen BY help

/FORMAT= AVALUE TABLES

/STATISTIC=CHISQ PHI

/CELLS= COUNT EXPECTED ROW COLUMN TOTAL RESID .

3) Subject Gender x Help Hypothesis: Male Subjects are more likely to give help than Female Subjects.

- You will notice that your data contains a new variable called "subgenmf," In order to simplify

testing for subject gender, the Female Single and Female Group data has been combined, and the

Male Single and Male Group data has been combined. If you look at the variable view, you can

see how the new labels are applied.

- We can use this new variable to determine helping behavior of the two genders.

- Copy and paste the previous syntax and use the new variable subgenmf:

CROSSTABS

/TABLES=subgenmf BY help

/FORMAT= AVALUE TABLES

/STATISTIC=CHISQ PHI

/CELLS= COUNT EXPECTED ROW COLUMN TOTAL RESID .

4) Subject Sex x Confederate Sex Interaction Hypotheses:

For Female Confederates: Male Subjects will be more likely to help than females.

For Male Confederates: Female Subjects will be more likely to help than males.

Observational Study Data Analysis Plan Template Page 2