Lecture Notes – Moderation & Interaction

Fill-in-the-blank flashcards covering key definitions, concepts, statistical procedures, assumptions and practical tips related to moderation and interaction analysis in psychology research.

Moderation occurs when the relationship between two variables changes as a function of a third variable called the __.

moderator

Statistically, moderation is tested as an __ effect in a regression or ANOVA model.

interaction

A variable that alters the strength or direction of the relationship between a predictor (IV) and an outcome (DV) is known as a __ variable.

moderator

In the regression model Y = b0 + b1X + b2M + b3XM + e, the interaction is carried by the term.

XM (product)

The individual-specific value used to create an interaction term is obtained by multiplying that person’s X score by their score.

M (moderator)

The effect of X on Y at a specific value of M is called the __ effect of X on Y.

simple

An interaction where increasing the moderator increases the effect of the predictor is called an __ interaction.

enhancing

An interaction where the moderator decreases the predictor’s effect is termed __.

buffering

If increasing the moderator reverses the predictor’s effect, the interaction is __.

antagonistic

Transforming a predictor by subtracting its mean from every score is called __.

centering (grand-mean centering)

Centering makes zero on the predictor correspond to its in the sample.

mean

Centering does NOT change the estimate or significance test of the __ effect.

interaction

In moderated multiple regression, b1 is a conditional effect of X on Y at the value for M.

0

With continuous moderators, “high” and “low” values are commonly defined as ± standard deviation(s) from the mean.

1

For a categorical moderator with k levels, __ coding uses k-1 dummy variables referenced to a base group.

dummy

The Durbin–Watson statistic is used to test the assumption of independence of __.

residuals

Homoscedasticity means the error variances are the same for all combinations of __ and moderator.

independent variable (predictor)

Multicollinearity can be detected with the statistics __ and VIF.

Tolerance

Outliers can be evaluated using the __ distance measure.

Mahalanobis

Residuals should be approximately distributed to satisfy regression assumptions.

normally

Moderator analyses are most powerful when either the IV or the moderator is an __ manipulation.

experimental

If X is manipulated, the moderator M should ideally be measured __ X is manipulated.

before (prior to)

When X and M are highly correlated, the interaction term XM may suffer from __ issues.

collinearity

A complete moderation occurs when the causal effect of X on Y becomes for some value of M.

zero

The Johnson-Neyman technique identifies the range of M values where the effect of X on Y is __ significant.

statistically

Never turn continuous predictors into __ variables in moderated regression.

categorical

Always include __ effects of the individual predictors when testing an interaction.

main

The combined effect of two predictors that exceeds the sum of their individual effects is termed a(n) __.

interaction

In a 2×2 factorial ANOVA, variance is partitioned into main effect A, main effect B, interaction A×B, and __ error.

residual

A moderated multiple regression is simply a multiple regression that includes __ terms as well as main effects.

interaction

Simple slopes analysis compares the slopes of X predicting Y at different levels of the __.

moderator

Three-way interactions require visualisation through __ for proper interpretation.

plots

Moderation can __ (weaken, amplify or reverse) a causal effect.

alter

One advantage of moderated regression with experimental manipulation is the ability to claim __ effects.

causal

A disadvantage of moderation analysis without manipulation is that results remain largely __.

correlational (cross-sectional)

If the interaction term is significant, we conclude that __ is present.

moderation

Scatterplots can be used to visually check the assumption of __ between X and Y within each level of a dichotomous moderator.

linearity

In dummy coding with Female = 1 and Male = 0, Female is compared against the __ group.

reference (control, Male)

When testing 3-way interactions, the regression equation must include all possible __ among the three predictors.

two-way and three-way products

The R² value represents the proportion of variance in the DV explained by the predictors, moderators and their __.

interactions