Lecture 12 - ANOVA Part 2 Flashcards
Why is a repeated-measures design ANOVA more desirable than a between group design?
Because when looking for the effect/noise, in a repeated measures design the inter-subject variance is classed as controlled rather than unexplained, decreasing the denominator and therefore increasing the F-ratio.
For repeated-measures ANOVAs, what are the three possible sources of variance?
- variance between conditions
- variance between subjects (individual differences)
- residual (unexplained) variance
How is the F-ratio calculated for a repeated-measures ANOVA?
F= MS effect (variance between conditions) / MS noise (MS total - MS effect - MS ind diffs)
How do you run a repeated-measures ANOVA in SPSS?
Analyse|GeneralLinearModel|RepeatedMeasures
Add the number of conditions, then define the measures which are your DVs
What does ‘sphericity assumed’ mean?
That the variances between groups are about the same.
In a multi-factorial ANOVA (MANOVA), what design are factors and is there a ‘main’ effect?
Factors can all be within-subject, between-group or a ‘mixed’ design. There can be either a ‘main’ effect or a variety of ‘interactions’.
What is the difference between a main effect and an interaction?
In a main effect, one of the IVs consistently affects the DV in the same way.
With an interaction, the effect of one factor depends on the presence of another.
What F values does a MANOVA return?
One for each main effect and one for the interactions.
At what number of MANOVA levels do post-hoc tests become necessary to find where the effect lies?
At 3+ levels.
In a 2x2 MANOVA, is there family-wise error?
No, it’s a single test.
What are the pros and cons of using an ANOVA in real-world situations?
+ it seems to be a convenient way to take into account all possible interactions, as it yields a significance value and F-ratio for each one.
- interpreting MANOVAs with many levels is difficult - there is a trade-off between being realistic and able to interpret data.
