Factorial Analysis of Variance: In-Depth Notes for Psychology Statistics Exam
Basic Logic of Factorial Designs and Interaction Effects
- Importance of Factorial Designs in Psychology
- Most behaviors are influenced by multiple variables.
- Factorial research design:
- Examines the effect of two or more variables simultaneously.
- More efficient than single-variable designs.
- Effects to Consider:
- Additive effects: One variable has an impact when considered alone or with another variable.
- Interaction effects:
- Occur when the combination of variables produces a unique effect not seen when examining variables independently.
- Variables that change the relationships among others are called moderators.
Relationship Between ANOVA Types
- One-way ANOVA:
- Evaluates the impact of one grouping variable.
- Two-way ANOVA:
- Examines two grouping variables.
- Each grouping variable is considered a "way" in the analysis.
Main Effects
- Definition:
- A main effect exists when the average result for a grouping variable is significant, regardless of other variables.
- Key terms:
- Cell: Each unique combination of conditions in a factorial design.
- Example: In a $2 imes 2$ factorial design, there are four conditions (cells).
- Cell Mean: Average score of a particular condition.
- Marginal Means: Averages calculated for one grouping variable across all levels of another.
Recognizing and Interpreting Interaction Effects
- Characteristics of Interaction Effects:
- The effect of one variable is influenced by the level of another.
- To identify interaction effects, observe the patterns of cell means.
- An interaction effect is indicated when differences in means vary across different levels of the other variable.
Two-Way ANOVA: Understanding F Ratios
- Main Effects:
- Column Main Effect:
- Numerator: Between-groups variance (variation between column marginal means).
- Denominator: Within-groups variance (average variance from each cell).
- Row Main Effect:
- Similar structure as column main effect but based on row marginal means.
- Interaction Effect:
- Uses means from diagonals in the design for its estimation.
Relation of Interaction and Main Effects
- Complex Relationships:
- A study can show main effects without interactions and vice versa, making interpretation more complex when interactions exist.
Assumptions of Two-Way ANOVA
- The populations should:
- Follow a normal distribution.
- Exhibit equal variances.
- These assumptions hold true for the populations in each cell of the analysis.
Controversies and Limitations
- Dichotomizing Numeric Variables:
- Use of median splits can lead to loss of information and is often controversial.