Main and Interaction Effects in Factorial Designs

Key Concepts of Main and Interaction Effects

  • Main Effects

    • Definition: The effect of an independent variable on a dependent variable averaged over all other variables.
    • Example: When conducting a t-test, comparing two levels of a single independent variable measures a main effect.

  • Interaction Effects

    • Definition: Occur when the effect of one independent variable on a dependent variable varies depending on the level of another independent variable.
    • Example: When considering team cohesion interventions, the relationship between intervention outcomes can vary based on whether additional social support is provided.

Understanding Interaction Effects through Examples

  • Scenario Example:
    • Intervention Analysis: Pre and post intervention measurements for team leaders.
    • Factor A: Time (pre and post intervention)
    • Factor B: Social Support (provided vs not provided)
    • Results:
    • Group with No Support → Improvement in team cohesion,
    • Group with Support → Greater improvement in team cohesion, demonstrating an interaction effect.

Factorial Designs and Their Benefits

  • Use of Factorial Designs: Allows for exploration of interaction effects between multiple independent variables.
  • ANOVA Example:
    • Two-way ANOVA with two by three design (two types of toys with different cartoon conditions):
    • Main Effects of Toy Type: Assessing general aggression levels with antisocial vs neutral toys.
    • Main Effects of Cartoon: Assessing aggression regardless of toy type.
    • Interaction of Toy and Cartoon: Significant interaction might indicate the effect of cartoon on aggression depends on the type of toy.

Application in Experiments

  • Hypothetical Lecture Example:
    • Lecture styles (cat memes vs dry) may influence sleep duration among students based on their hangover status, raising questions about main effects vs interaction effects.
  • Importance of Graphing Interactions:
    • Graphs clarify interactions and help visualize varying effects among different groups.

Graphical Interpretation of Interaction Effects

  • Seeking Interactions:
    • Parallel lines in a graph indicate no interaction effect, while non-parallel lines suggest a possible interaction.
  • Two Graph Examples:
    • Example 1 (No interaction): Women’s selection of date attractiveness remains higher than men's regardless of alcohol consumption → Parallel lines.
    • Example 2 (Interaction): Variations in selection based on alcohol levels for men and women at different consumption levels suggest an interaction → Non-parallel lines.

Statistical Confirmation

  • Main effects can be confirmed through ANOVA, rather than solely reliance on graphical data.
  • Graphs serve as tools for interpretation and hypothesis generation.