Moderation Flashcards
Define:
Moderation models.
When the association between two variables depends on a third variable.
Note, the moderator is often denoted as ‘w’, as mediation instead uses ‘m’ typically.
What types of questions may be answered by moderations?
- When
- Who
- Where
(e.g. When you are having a heart attack, who is more likely to receive a correct diagnosis - men or women?)
How is a moderation tested?
In the format of a multiple regression model or factorial ANOVA.
What are the β predictors of a moderation which are derived from the multiple regression model?
- Variable 1 (predictor variable, x)
- Variable 2 (moderator, w)
- Moderation Effect (V1 x V2).
What are the main effect(s) in a moderation?
They represent how much one variable/predictor is associated with the outcome.
There is a main effect for the initial predictor variable as well as the moderator.
True or False:
The moderation effect represents a combination of predictors and needs to be visually interpretted.
(Also referred to as the interaction effect).
True
This is usually interpretted using a simple slopes graph, in which you can see how the different conditions of the (categorical) moderater affect the association between the predictor variable and the outcome.
Identify:
What kind of effect is displayed in this moderation graph output?
Predictor main effect.
Identify:
What kind of effect is displayed in this moderation graph output?
Moderation main effect.
True or False:
Moderators can only be categorical variables.
False
Moderators may also be continuous variables, such as age.
When visualisng these on a simple slope graph though, data points are typical sorted into three categories based on a low (-1SD), average, or high (+1SD) value of its moderator.
List:
The THREE patterns of moderations.
- Magnification.
- Attenuation.
- Crossover.
Describe:
A magnification pattern in a moderation.
When the moderator causes a stronger effect by strengthening the association between the predictor and the outcome.
Describe:
An attenuation pattern in a moderation.
When the association between a predictor and outcome variable is weakened.
Describe:
A crossover pattern in a moderation.
When a null association between a predictor and outcome becomes significant only when taking into consideration the moderator.
This is considered the ‘exciting’ effect to find when conducting a Psychological/Behavioural study.
Why is it important to divide continuous moderators into categories of close to the mean value as well as groups one standard deviation (SD) above or below this?
TWO reasons.
- Higher confidence making generalisations closer to the average.
- Provides a standardised way of constructing and presenting interpretations.
True or False:
None, some, or all of the simple slopes of a moderation may be significant.
True
