Cram Supplement

Main Effect vs Interaction vs Simple Effect

Main Effect: The effect of an independent variable on the dependent variables, averaged across all other variables.

i.e. Does “Caffeine”, on average, improve “Performance” regardless of “Task Difficulty”?

Interaction: The effect of an IV which depends on the level of another IV.

i.e. Does the effect of “Caffeine” on “Performance” depend on the “Task Difficulty”?

Simple Effect: The effect of one IV at one specific level of another IV. Used to break down an interaction.

i.e. What specifically happens with “Caffeine” when the task is “Hard”?

Effect Size vs Power

Effect Size: The size of the difference between groups

Power: The ability of the study to accurately assess whether or not there is an effect

As Effect Size increases, so does Power

Effect Size Calculation

(Mean difference) / SD

Pearson Correlation: Purpose, Assumptions, Non-parametric Equivalent

Purpose: Strength and direction of a relationship between variables.

Assumption: I/R, Normality, Linearity, Homoscedasticity

Non-parametric Equivalent: Spearman’s

Linear/Multiple Linear Regression: Purpose, Assumptions

Purpose: Predicts the value of a DV by 1+ IVs

Assumption: I/R, Normality, Linearity, Homoscedasticity, Independence of residuals

Independent Samples t-test: Purpose, Assumptions, Non-parametric Equivalent

Purpose: Compares means of 2 independent groups

Assumption: I/R DV, Independence, Normality, HOV

Non-parametric Equivalent: Mann-Whitney U

Paired/dependent/repeated measures t-test: Purpose, Assumptions, Non-parametric Equivalent

Purpose: Compares means from the same group at 2 times

Assumption: I/R DV, Dependent observations, Normality

Non-parametric Equivalent: Wilcoxon Signed-Rank Test

One-way ANOVA: Purpose, Assumptions, Non-parametric Equivalent

Purpose: Difference in means across 3+ independent groups

Assumption: I/R DV, Independence, Normality, HOV

Non-parametric Equivalent: Kruskal-Wallis

Repeated Measures ANOVA: Purpose, Assumptions, Non-parametric Equivalent

Purpose: Compares means across 3+ groups over 2+ times

Assumption: I/R DV, Normality, Sphericity

Non-parametric Equivalent: Friedman’s

Chi-Square: Purpose, Assumptions

Purpose: Tests nominal data

Assumption: Nominal, Independence, frequencies >5

Levene’s Test: Assumption, Criteria

Assumption: HOV

Criteria: p<.05 violates HOV

Shapiro-Wilk Test: Assumption, Criteria

Assumption: Normality

Criteria: p<.05 violates Normality

Mauchly’s Test: Assumption, Criteria

Assumption: Sphericity

Criteria: p<.05 violates Sphericity

Durbin-Watson: Assumption, Criteria

Assumption: Autocorrelation (independence of residuals)

Criteria: Ranges 0 - 4, 2 is no autocorrelation (good)

Variance Inflation Factor (VIF): Assumption, Criteria

Assumption: Multicollinearity

Criteria: VIF <5 is acceptable

Tukey’s/Bonferroni Usage

Used after a significant ANOVA to determine which specific groups differ while controlling for alpha inflation.

Tukey’s: Best for equal n and all pairwise comparisons

Bonferroni: More conservative, adjusts alpha level to control inflation

Skewness/Kurtosis: Definition and Criteria

Skewness: Distribution asymmetry

Kurtosis: Distribution “Peakedness”

Criteria (both): Values close to 0 best. Absolute values >1.96 violate normality.

Cohen’s d: Meaning and Criteria

Magnitude of effect size.

Small: 0.2

Medium: 0.5

Large: 0.8

Cronbach’s Alpha vs. Cohen’s Kappa vs. ICC

Alpha: Internal reliability of Likert scales (<.7 unacceptable)

Kappa: Inter-rater agreement for two raters (Nominal, .60+ is substantial)

ICC: Inter-rater agreement for multiple raters (I/R)

Point-Biserial Correlation

Correlation statistic for one I/R and one binary (nominal/ordinal, either way, only 2)