4. Muligroup: Independent Groups Flashcards
Why use ANOVAs?
Multiple t-test ➡️ increased risk of type 1 errors
14.3% for a 3 t-test comparison
When would we use the one-way (independent) ANOVA?
When we are looking at 1 variable (e.g. social support) in different IVs (e.g. marital status).
What does the one-way ANOVA test do?
It compares the variance between groups (assumed to be due to the IV) to the variance in each group (due to chance).
What is the F-ratio?
What does a large F ratio mean?
Variance btwn groups
➗
Variance within groups
Large = more variability between groups (caused by IV)
What’s an alt test to the F-test to perform on the one-way ANOVA?
What does a NON significant result mean?
Homogeneity of variance test, brown-forsythe and Welch).
If test is non-significant, then the groups are similar. The ANOVA can therefore be trusted.
How would we report the one way ANOVA?
🔸F (anova test)
🔸p value
🔸d.f.
F(df1, df2…) = ❔, p = ❔
If the levene statistic is significant, what do we look at?
Welch or Brown-Forsythe and report adjusted F-ratio
What doesnt the ANOVA tell us?
Where the difference lies [in variance].
Need to do more tests:
🔸planned or post hoc tests
What are post- hoc tests led by? Name some post-hoc tests.
What are planned comparisons led by?
🔸are data led; didnt plan to look at data until it showed a difference
🔹tukey
🔹scheffe
🔸pllanned comparisons led by theory
Effect size ANOVAS:
🔸rather than cohen’s d, what is reported?
🔸hows it calculated?
🔸boundaries?
🔸partial eta squared (np^2)
sum of squares btwn groups ➗ total sum of squares
🔸small ➖ 0.01
🔸medium ➖ 0.06
🔸large ➖ 0.14
Kruskal- Wallis
🔸what is it?
🔸what is its output?
🔸 non para equivalent of one way independent ANOVA.
🔸is similar to Mann Whitney U
🔸 X^2 is output