Topic 6 - ANOVA

General Linear Model

Data = Model + Error

Regression

Looks at the associations between continuous variables.

ANOVA

Looks at the mean group differences

SSregression

the difference between predicted outcome scores for the individual and the mean

SSbetween

the difference between the group and grand mean

SSresidual

the difference between the predicted outcome score for the individual and their actual outcome score

SSwithin

the difference between the individuals actual outcome score and their group mean

dfbetween

a-1, where a = the number of IVs

dfwithin

a(n-1)

omnibus test

Tells us only if there is a statistically significant effect of the IV, not where it lies

between group SSwithin

random error, measurement error, individual differences

repeated measures SSwithin

random error, measurement error

multivariate tests

compare between group factors

univariate tests

compare repeated measures factors

Factorial ANOVAs

Have 1 DV and more than 1 IV

Main effect

the independent influence of one IV on the DV

Interaction effect

occur where someones score on the DV is determined by membership in more than one group

Interaction effect

The score on the DV is determined by membership in more than one group, so is a conditional effect

2-way interaction

Occurs when the outcome under one main effect is conditional on another factor

3-way interaction

Occurs when a 2-way interaction is itself conditional on another factor

Ordinal interaction

an interaction due to a different magnitude of difference in scores between two groups

Disordinal interaction

an interaction due to a different pattern of difference in scores between two groups

interpretation of disordinal interactions

cannot interpret main effects on their own and must first ivestigate significant interactions

interpretation of ordinal interactions

can interpret main effects on their own