ThreeWay ANOVA Analysis by SAS Interaction Plot by

Three-Way ANOVA Analysis by SAS Interaction Plot by Excel

The Design Ø Type of Crime: Burglary or Swindle Ø Culture of Defendant: American or Chinese Ø Sex of Juror: Female or Male Ø DV: Recommended length of sentence

Read in the Data Ø Data Jury. Simulation; Ø Infile 'C: UsersVatiDocumentsSim. Data ANOVA 3 B-13. dat'; Ø Input Culture $ Crime $ Sex $ Sentence;

Omnibus Analysis Ø Proc ANOVA; Ø Class Sex Culture Crime; Ø Model Sentence = Sex|Culture|Crime; Ø Means Sex|Culture|Crime; run;

Results & Follow-Up Ø The triple interaction is significant Ø Conduct simple interaction analysis Ø Test Culture x Crime for each sex Ø I asked you to do the simple interaction analysis even if the triple interaction was not significant

Simple Interaction Effects Ø Proc Sort; By Sex; Ø Proc ANOVA; Ø Class Culture Crime; Ø Model Sentence = Culture|Crime; Ø By Sex; run;

Results & Follow-Up • For each simple interaction that is significant, test the simple, simple main effects. • Effect of type of crime at given levels of culture and sex. • I asked you to do these tests regardless of whether or not the simple interaction was significant

Simple, Simple Main Effects Ø Proc Sort; By Sex Culture; Ø Proc ANOVA; Class Crime; Ø Model Sentence = Crime; Ø By Sex Culture; run;

The Comparisons Made Ø Compare length of sentence for burglars versus swindlers when ØJuror is is female, defendant is American female, defendant is Chinese male, defendant is American male, defendant is Chinese

Prepare an Interaction Plot • Enter the appropriate means into Excel (2010) • These are for the female jurors

Insert, Charts, Line

Select “Line With Markers”

Tweak the Vertical Axis

Gee Whiz

Add Axis Titles

Enter Text in Title Box Then do the same for the Vertical Axis

Enter Chart Title

Other Tweaks • Resize the chart • Change fonts • Remove or change grid lines • Et cetera • Then copy the chart to Word, Power. Point, or other application.

Prepare the Second Plot • We need a plot for the male jurors too. • Enter the appropriate data into Excel. • Tweak the plot so it has the same dimensions as the previous plot. • If you do not, Excel may scale it in a way that distorts the differences between the two plots.