Lecture 2: The General Linear Model: Two-way ANOVA

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/53

encourage image

There's no tags or description

Looks like no tags are added yet.

54 Terms

1
New cards
How many categorical predictors does a two-way ANOVA have?
Two categorical predictors.
2
New cards
What example is used to illustrate two-way ANOVA in the lecture?
The effects of personality type (introvert vs. extrovert) and motivation type (praise vs. blame) on motor-task performance.
3
New cards
What happened in the praise condition in the lecture example?
Participants were told they were doing very well on the task.
4
New cards
What happened in the blame condition in the lecture example?
Participants were told they were doing very poorly on the task.
5
New cards
What is a main effect?
The individual effect of one independent variable on the dependent variable, averaged across the levels of any other independent variables.
6
New cards
How are main effects related to one-way ANOVA?
A main effect considers the effect of one independent variable while averaging across levels of other independent variables, similar to a one-way ANOVA.
7
New cards
What is an interaction?
An interaction occurs when the effect of one independent variable depends on the level of another independent variable.
8
New cards
When can an interaction occur in the lecture's examples?
When there are two factors, as in a two-way ANOVA.
9
New cards
What do parallel lines indicate in an interaction plot?
Parallel lines indicate main effects without an interaction.
10
New cards
What do non-parallel lines indicate in an interaction plot?
Non-parallel lines indicate an interaction.
11
New cards
Give an example of main effects without an interaction.
Praise could improve performance compared with blame for everyone, with no effect of personality type.
12
New cards
Give another example of main effects without an interaction.
Introverts and extroverts could differ in performance, and praise could improve performance compared with blame, with the effects being additive.
13
New cards
Give an example of an interaction between personality and motivation.
Extroverts could be more motivated by blame while introverts are more motivated by praise.
14
New cards
What does a two-factor ANOVA test?
Hypotheses about mean differences when there is more than one categorical predictor (factor).
15
New cards
How many null hypotheses are tested in a two-way ANOVA with two factors?
Three: the main effect of A, the main effect of B, and the interaction between A and B.
16
New cards
What is the null hypothesis for the main effect of factor A?
There is no overall difference between the levels of factor A.
17
New cards
What is the null hypothesis for the main effect of factor B?
There is no overall difference between the levels of factor B.
18
New cards
What is the null hypothesis for the A × B interaction?
The difference between levels of A does not depend on the level of B, and vice versa.
19
New cards
How does the GLM equation for a two-way ANOVA differ from the one-way ANOVA equation?
It extends the one-way model to include the main effect of factor B and the interaction effect AB, in addition to the existing terms.
20
New cards
In the lecture's notation, what does Yᵢⱼₖ represent?
The outcome for the kth person in the ith group for factor A and the jth group for factor B.
21
New cards
In the lecture's explanation, what does the effect of a variable mean?
The deviation associated with that variable from the grand mean, considering that variable while ignoring the other factor.
22
New cards
What does the interaction represent in the lecture's explanation?
Variation from the grand mean that the main effects cannot explain.
23
New cards
How are the degrees of freedom for main effects calculated in a two-way ANOVA?
For each factor, degrees of freedom = number of levels − 1.
24
New cards
If factor A has two levels, what is dfA?
2 − 1 = 1.
25
New cards
If factor B has two levels, what is dfB?
2 − 1 = 1.
26
New cards
How are the degrees of freedom for the A × B interaction calculated?
dfAB = dfA × dfB.
27
New cards
If dfA = 1 and dfB = 1, what is dfAB?
1 × 1 = 1.
28
New cards
What model assumptions are listed for a between-subject two-way ANOVA?
Homogeneity of variance, checked with Levene's test, and normality of residuals, checked with a Q–Q plot.
29
New cards
How can a model be extended to a three-factor ANOVA in the lecture's example?
Add a third factor, such as participants' handedness, to personality type and motivation type.
30
New cards
What design results from adding handedness to the lecture's two-factor example?
A 2 × 2 × 2 between-subjects ANOVA.
31
New cards
How many interaction terms are there in the lecture's three-factor ANOVA example?
Four interaction terms: three two-way interactions and one three-way interaction.
32
New cards
What are the advantages of within-subject ANOVAs?
Group differences cannot be due to individual differences; generally smaller samples are needed; and statistical power to detect differences increases.
33
New cards
Why can within-subject ANOVAs reduce the influence of individual differences?
The same participants take part in the different conditions, so differences between groups cannot simply be attributed to different people.
34
New cards
What is one disadvantage of within-subject ANOVAs involving certain variables?
Not all variables can be manipulated within-subjects; for example, personality type.
35
New cards
What are order or practice effects in within-subject designs?
Performance or responses may change because of the order of conditions or because participants practise the task.
36
New cards
What are carry-over effects?
One manipulation or condition affects participants' responses in a subsequent condition.
37
New cards
What is the key additional issue in a one-way within-subject ANOVA compared with a between-subject design?
The model must account for subject effects, including how much variance there is between subjects.
38
New cards
What are the two types of subject effects listed in the lecture?
Random effects and fixed effects.
39
New cards
What is a random effect in the lecture's explanation?
Participants are treated as a random factor because they are selected at random from the population, and differences between participant levels are not the main interest.
40
New cards
Why are participants treated as a random factor in the lecture?
They are selected at random from the population, and the study is not primarily interested in differences between individual participants.
41
New cards
What is a fixed effect in the lecture's explanation?
An experimental factor, such as praise type, whose levels are chosen because they are of interest rather than randomly selected.
42
New cards
What is sphericity?
An assumption for the F test in repeated-measures designs that the variances of the differences between treatment levels are equal.
43
New cards
How is sphericity similar to the homogeneity-of-variance assumption?
Both concern equality of variances, but sphericity concerns the variances of differences between treatment levels in repeated-measures designs.
44
New cards
What variances does sphericity concern?
The variances of the differences between treatment levels, rather than the variances of the raw scores.
45
New cards
When is the sphericity assumption relevant?
For designs with at least three levels per factor; with two levels, there is only one possible difference to consider.
46
New cards
How is sphericity tested?
Mauchly's test.
47
New cards
What is the null hypothesis of Mauchly's test?
The variances of the differences between conditions are equal.
48
New cards
What does p < .05 for Mauchly's test indicate in the lecture?
The sphericity assumption is not supported; the F test becomes too liberal and produces too many false positives if uncorrected.
49
New cards
What can be done if sphericity is violated?
Correct the degrees of freedom using the Greenhouse–Geisser correction or the Huynh–Feldt correction.
50
New cards
How does the Greenhouse–Geisser correction compare with the Huynh–Feldt correction in the lecture?
Greenhouse–Geisser is more conservative; Huynh–Feldt is less conservative.
51
New cards
Main effect vs. interaction: what is the difference?
A main effect is the overall effect of one factor averaged across other factors; an interaction means the effect of one factor depends on another factor.
52
New cards
What does a two-way ANOVA test?
Two main effects and one interaction: A, B, and A × B.
53
New cards
What is the key difference between between-subject and within-subject ANOVA?
Between-subject designs use different participants in conditions; within-subject designs use the same participants across conditions.
54
New cards
What is the key sphericity rule to remember?
Sphericity concerns equal variances of differences between conditions and is relevant when a factor has at least three levels.