Wk 10 - Moderation with ANOVA Flashcards
What is ANOVA used for? (x1)
Analysing difference in means
What kind of variables are used in ANOVA? (x2, x2)
Categorical IVs
*Measured or manipulated
Continuous DV
*(use logistic regression instead for categorical)
Explain the key concept in ANOVA
Partitioning variance into:
* Accounted for by known factors * And not
How is variance partitioned in ANOVA? (x2, x2)
Known factors = ‘systematic variance’
*Between-groups
Unknown factors = random/unmeasured/error variance
*Within-groups
What does between-groups variance describe in ANOVA? (x1)
Differences between Group Means and Grand Mean
What does within-groups variance describe in ANOVA? (x1)
Differences between individual scores and Group Means
What does the F ratio tell us in ANOVA? (x1)
Whether systematic variance is greater than error variance
What do omnibus tests tell us in 2-way ANOVA? (x3)
Whether IVs affect DVs:
Main effect of single IVs in isolation
Interaction effects - combined multiple factors
What don’t omnibus tests tell us in 2-way ANOVA? (x1)
Which means/conditions differ from one another
How do we assess main effects in 2-way ANOVA? (x2)
Compare marginal means:
*Means of each level of 1 factor, ignoring the levels of the other
How do we assess presence of an interaction effect in 2-way ANOVA? (x2)
Evaluate effect of 1 factor at each level of another
*Is change in scores the same (stable) across the different levels?
How do we follow-up significant Main Effects in 2-way ANOVA? (x1, x3)
If 2 levels - which mean is higher?
If 3+ levels - main effect comparisons
*Pairwise t-tests
*Linear contrasts (can group multiple conditions)
What questions are asked by follow-up tests of significant interactions in 2-way ANOVA? (x2)
How does pattern of means for 1 factor change across different levels of another?
How does one factor moderate effect of another?
What follow-up tests are needed for a significant interaction in 2-way ANOVA? (x2)
Simple effects
*Separate 1-way ANOVAs for effect of 1 factor at each level of the other
What follow-ups are needed for significant simple effects in 2-way ANOVA?
2 levels of primary factor - compare means
3+ levels of primary factor -
simple comparisons
When should you plot the slopes/means for ANOVA? (x1)
Why? (x1)
Before any analyses
Easier to grasp interactions (than matrics of means)
What does an interaction represent in ANOVA? (x1)
A moderating effect of 1 variable on another
How do you identify main effects from ANOVA graphs/plots? (x2)
Line factor - compare average value of lines
x-axis factor - compare average values between points at each x-value
Are main effects required to produce an interaction in ANOVA, or vice versa? (x1)
No
Describe the general 3 stages of analysis plan for 2-way ANOVA (x3, x2, x3)
Assess omnibus effects
* Main Effects of each factor * Interaction Effect of the two factors
Interpret significant Main Effects
*May require follow-up simple comparisons
Interpret the Interaction, if significant
* Follow-up simple effects analysis * Simple effects analyses may require follow-up simple comparisons!
What considerations affect the follow-up tests conducted in 2-way ANOVA?
How were factors manipulated?
* Both factors manipulated between participants? * Both factors manipulated within? * 1 factor manipulated between, 1 within?
Describe how Ps are allocated in a 2-way between-Ps ANOVA (x2)
Random allocation to 1 level of 1 factor
*Each cell summarises data from different individuals
What simple effects test do we use to follow up interactions in 2-way between Ps ANOVA? (x1, plus explain x3)
Series of 1-way ANOVAs for each level of primary factor
* But don't split data file into levels of the factor * Because assume MSerror/random variance is equal across all cells * And doesn't need recalculating for single factor
Describe how Ps are allocated in a 2-way within-Ps ANOVA (x2)
All Ps complete all levels of all factors
*Cells summarise data from same Ps
What assumption is important in 2-way within-Ps ANOVA? (x1, plus describe x3)
Sphericity:
Variances of levels within a factor are equal
Covariances between pairs of levels within a factor are equal
Violation of this assumption increases Type I error rate (False Alarm)
How is the assumption of spericity evaluated for 2-way within-Ps ANOVA? (x1)
Which is interpreted such that…(x2)
And it is best to…(x1)
Because… (x1)
Mauchley’s Test (a chi-square)
Significance indicates violation
And need for correction to df to power down the test
Apply Greenhouse-Geisser correction universally
Mauchley’s is prone to Type 2 (misses)
What follow-ups are used in 2-way within-Ps ANOVA?
Main effects - same as between-Ps (pairwise comparisons/linear contrasts)
Interactions - 1-way ANOVAs using MSerror only for primary factor in question
How are Ps allocated for 2-way mixed factor ANOVA?
One factor, Ps complete all levels
Other factor, allocation to 1 level
How are main effects followed up in 2-way mixed factor ANOVA? x2)
Depending on whether the factor is between- or within-Ps,
Follow steps as for an entirely between- or within-
How are interactions followed up in 2-way mixed factor ANOVA? (x2)
Simple effects comprised of between- or within-Ps ANOVAs
depending of which is the primary factor