Lecture 15 Psyc 300A
Example: Movie Preferences MenWomenMean Romantic364.5 Action745.5 Mean55
What a Factorial Design Tells You Main effect: The effect of an IV on the DV, ignoring all other factors in the study. (Compare means of different levels of IV, while ignoring [collapsing across] other IVs [ i.e., compare marginal means]) Interaction effect: When the effect of one IV on a DV differs depending on the level of a second IV. Interpret the interaction first
Group Exercise: Main Effects and Interactions Any questions from p.205 in book?
Example: Psychotherapy Outcome PrePostMarginal Mean Cognitive No Tx20 Marginal Mean 2015
Group Activity For each graph, decide whether there are main effects for each variable and an interaction.
Group Activity: Main Effects and Interactions Make graphs of the following situations: Study 1 Study 2 Study 3 Study 4 Var AVar BAxB interaction p <.05 n.s. p <.05 n.s.p <.05 n.s. p <.05
Factorial Designs: Naming Conventions The first number is the number of levels in first IV, second number is number of levels in second IV, etc. 2 x 2 2 x 3 2 x 2 x 3 Between-subjects, repeated measures (within), mixed
A 2 x 3 Interaction
Analysis of Variance (ANOVA) Test statistic for ANOVA is F Is related to t-test ANOVA is for multiple levels of IV and multiple IVs MS between F = MS within It compares the amount of variability between groups to amount within groups
ANOVA Source (or Summary) Table _______________________________________ Source df SS MS F. Between groups Within groups Total _______________________________________
Interpreting the F statistic (ANOVA) Hand calculations –Calculate F (this is F obtained ). –Compare value with F in table (Table B.3. This is F critical ). To do this need to know alpha and df. –If F obtained > F critical, a significant effect. In SPSS –Look at source (summary) table –Effects with significance values less than.05 are significant.
ANOVA (one way) Example Do preschoolers benefit from extra practice in language skills? Groups: 1=5hrs; 2=10 hrs; 3=20 hrs
Oneway ANOVA: SPSS Output
Post hoc comparisons When there are more than two conditions, a significant F-test tells you that at least two means are different, but not which ones To discover which are different, we use post hoc comparisons Some of these include Scheffe, Newman-Keuls, Duncan, Tukey tests
SPSS: Factorial ANOVA, All Between- Subjects IVs (Weight loss data) Female trainer Female trainer Male trainer Male trainer Female client Male client Female client Male client
SPSS Data File: Weight Loss Study
SPSS Weight Loss Study Plot
SPSS Output File: Weight Loss Study