Factorial designs 6.2&6.3 Flashcards
Factorial designs
Used to test relationships between more than one factor
•Used to test for interactions between factors
•Interactions illustrate complicated relationships
•The effect of one variable on the DV is not constant, it depends on another variable
•Factorial designs have more than one factor
Different questions warrant different designs
•Single factor design
- Do people mimic facial expressions?
•Factorial design
-Do people mimic facial expressions of people they like and that they dislike?
Factor levels and conditions
Each factor has levels and together they create conditions
Conditions: ways the levels of different factors can combine
Main effects
- There is a significant difference between the means of the levels of a factor
- There can be a main effect for each factor
Interactions
•The levels from the different factors interact with each other and lead to
significant differences in the DV
•The outcome of the level of one factor depends upon its relationship to the level of
another factor.
Possible outcomes in factorial designs
- A main effect is possible for each factor
* An interaction is possible for each combination of factors
Main effects are based on marginal means
Marginal means – means for the levels of one factor when
collapsing over the levels of the other factor(s)
“qualified by the interaction.”
If a factor has a main effect and an interaction, the interaction is
used to explain the main effect.
The main effect is “qualified by the interaction.”
- Always interpret main effects in the context of their interactions
Common parametric stat tests
- Assumptions of parametric tests
- Interval or ratio data
- Normally distributed data
- Homogeneity of variance
- Independence of samples
- linearity
- Two conditions only
- T-test
- More than 2 conditions
- ANOVA: Analysis of Variance
ANOVA
Analysis of Variance
•ANOVAs do not tell you the direction of the effects
One -way ANOVA
- When testing a design that has a single factor with 3+ levels
- One-way repeated measures ANOVA (one-way within groups ANOVA)
- One-way between groups ANOVA
- H0: mean 1 = mean 2 = mean 3
•n-way or factorial ANOVA (e.g., 2-way ANOVA)
- When testing n-number of factors
- Factorial repeated measures ANOVA (e.g., 2-way factorial within-groups ANOVA)
- Factorial between-groups ANOVA
- Mixed-factorial ANOVA
- H0: the mean of the different conditions do not differ
Family-wise error rate
the probability of making one or more errors
Controlling for multiple comparisons
- 1st run an omnibus test (ANOVA is an omnibus test)
- Don’t run multiple t-tests
- Family-wise error rate
•2nd run Post-hoc (follow up) tests
Post hoc comparisons
- Follow up significant main effects and interactions with post hoc tests
- Follow up on main effects for factors with 3 or more levels
- Main effect of emotion (sad, happy, neutral)
- Multiple t-tests: Sad vs. happy; sad vs neutral; happy vs. neutral
- Follow up on interactions
- “Simple effects” that characterize the interaction
- Significant emotion (happy, sad) x task difficulty (easy, difficult) interaction
- For the happy level, is there a significant difference between easy and difficult tasks?
- For the sad level, is there a significant differences between easy and difficult tasks?
•If you are running multiple post hoc analyses, you should control for
multiple comparisons using other methods (like Bonferroni
corrections)
