Factorial Design & ANOVA
PSYC 71, Week 7 Lecture
Date: February 19th, Winter 2025
Week 7:
Due: Stats HW #2 by Sunday at 11:59pm (available on Canvas)
Week 8:
Exam 2: First half of lecture (12:00 - 12:50pm)
Due: Stats HW #3 by Sunday at 11:59pm (released on Canvas one week prior)
Week 9:
Due: Poster (PDF version submitted through Canvas by Sunday at 11:59pm)
Topics:
Factorial design & notation
Between, within, and mixed designs
Main effects & interactions
Eyeballing effects
Factorial ANOVA model
Decomposing cell means into main effects and interactions
Tests of effects in the two-factor model
ANOVA output table for the two-factor model
Focus on analyzing multiple independent variables.
Example Questions:
Which is more delicious: hot food or cold food?
Which is more delicious: hamburgers or ice cream?
Answer: It depends on various combinations of food types.
Investigates the interaction effect of one independent variable with another.
Factorial Design: Each level of one IV combined with each level of the others for all combinations.
Investigates the effect of two or more independent variables (factors) on the dependent variable.
Factors: Price and style affecting wine rating.
Example of levels: Cheap/Expensive and Sweet/Dry can result in different ratings.
Written like a multiplication expression: # x # x #
# of IVs = number of factors
# of levels = specific values of each factor
# of conditions = product of the factors
Examples include:
2x2 = 2 IVs, each with 2 levels = 4 conditions
2x2x2 = 3 IVs, each with 2 levels = 8 conditions
3x3 = 2 IVs, each with 3 levels = 9 conditions
Scenario: Researchers study the impact of stereotype threat on intelligence tests with 4 conditions.
Notation for Design:
a) 2x2 factorial
b) 1x4 factorial
c) 3x3 factorial
d) None of the above
Definitions and examples of these designs will follow.
Between-Subjects Design: Each subject in just one condition.
Within-Subjects Design: Each subject in ALL conditions.
Mixed Design: Some factors between subjects, some within; subjects in more than one but not all conditions.
Different conditions assessed by different subjects:
Sweet, Cheap / Sweet, Expensive / Dry, Cheap / Dry, Expensive
Subjects participate in all conditions;
Order does not have to be counterbalanced, could be randomized.
Subjects experience more than one condition but not all; example shown with price and style combinations.
More variety in how participants experience combinations of factors.
Scenario: Robert's study on age and emotion's effect on memory with conditional manipulation:
Design type:
a) 2x2 between-subjects factorial
b) 2x2 within-subjects factorial
c) 2x2 mixed design factorial
d) None of the above
Main effect - on average, are the levels of one factor different from each other?
Simple effect - at a specific level of one factor, are the levels of the other factor different from each other?
interaction - is the pattern/effect at one level of a factor different from the pattern/effect at another level of that factor?
Main Effect: Average differences between levels of a factor.
Simple Effect: Differences at a specific level of one factor.
Interaction: Differences in patterns across levels of factors.
Example calculations shown using price and style, with marginal means detailed.
Reiteration of the main effects of style across different types of wine ratings.
Sample simple effects calculated for each type of wine helping illustrate interactions.
The dependence of effect of one factor on the level of another demonstrated through wine rating differences.
Explanation of parallel lines indicating no interaction between factors.
Evaluation of potential statistical effects from provided data table.
Evaluation of effects shown in a figure based on quality ratings.
Reiteration of evaluating effects from another figure representation.
Eyeballing Effects: Worksheet available on Canvas to aid practice.
Introduction to models related to factorial ANOVA.
Equations demonstrating the general model structure for one factor.
Description of how to formulate a two-factor linear model with examples.
Steps to analyze variance of component scores relative to residual variance after group mean decomposition.
Concept of treatment offsets within routes and time when examining deviations.
Detailed calculations illustrating how offsets derived from observed cell means reflect interaction laws.
Confirming calculations around the grand mean of given data.
Overview of tests regarding main effects and interactions in factorial ANOVA.
Structure and specifics of how to populate an ANOVA table for two-factor designs.
Task to complete Annotated ANOVA table for the Times to Campus data, prompting calculation and estimation.