Mod. 13, Between Subjects ANOVA

Question 1

Conditions for Between Subjects ANOVA

Answer 1

Populations have equal variances (robust if this is violated)
Populations normally distributed
All groups are independent

Question 2

3 questions answered by between subjects ANOVA

Answer 2

ANSWERS these questions: main effect of first IV? Main effect of second IV? Interaction between two IVS? (focus is on the interaction effect)

Question 3

Variability due to error, between subjects design

Answer 3

any variability not due to an IV

Question 4

For within subjects ANOVA, why do we not need to calculate F observed for between groups variation?

Answer 4

We know individuals already vary from each other naturally, we are only interested in sources of variation from our manipulation

Question 5

Conditions for Repeated Measures

Answer 5

Normally distributed data
Homogeneity of variance
Sphericity: homogeneity of variance between levels (refers to the variances of the differences between all combos of all levels of the within-subjects factor:
Robust to violations of normality and homogeneity, but NOT SPHERICITY

Question 6

Spurious correlations

Answer 6

If you run 20 correlations, its likely that a correlation is due to chance: SPURIOUS CORRELATIONS (when find a correlation but doesn’t actually indicate a true association)
Outcome should always be on y axis

Question 7

4 possible outcomes of correlations

Answer 7

If there is a significant correlation between two variables, each of these possibilities considered: it might be a cause and effect relationship where x causes y, or might be that y causes x, or might be that there’s a third variable =, or relationship might be entirely due to chance

Question 8

Correlations and generalizations

Answer 8

WE CANNOT MAKE GENERALIZATIONS BEYOND OUR DATA THAT THE RELATIONSHIP WILL CONTINUE IN A LINEAR WAY BEYOND OUR DATA
We’re describing the data rather than making generalizations
Correlation is NOT significance: we need to determine whether two variables are significantly correlated at a specified level of significance
USE T TEST

Question 9

Within subjects null and alternative

Answer 9

null?: all means are same
alternative: at least 2 means are different

Question 10

Chi square null and alternative

Answer 10

null: expected and observed frequencies are the same
alternative: expected and observed are different

Question 11

between subjects null and alternative

Answer 11

null: no main effects and no interaction
alternative: significant main effects and significant interaction