ANOVA and Simple Contrasts

Question 1

COMPARING MEANS

Answer 1

• input data into c2
• input boxes into c1; check nominal
• ANALYZE-COMPARE MEANS-MEANS
• c2 = DL; c1 = IL = OK
• make scales
• ANALYZE-GENERAL LINEAR MODEL-UNIVARIATE
• c2 = DV; c1 = FF = OK
• make nominal
• ANALYZE-GENERAL LINEAR MODEL-UNIVARIATE-PLOTS
• c1 = HA; add = CONTINUE = OK

Question 2

T-TESTS (EXP)

Answer 2

• both used in 2-group designs
• f-ratio is STATSIG when t-test is and vice versa
• F = t^2
• ANOVA = general t-test; used whenever t is

Question 3

F-RATIO

Answer 3

• statistically useful; helps see if dif is genuine
• informally gives measure of dif between group score means (for any considered number of groups)
• group means dif + = f-ratio +
• group variability (scatter) - = f-ratio +
• group variability (scatter) + = f-ratio -
• f-ratio > CV = reliable finding

Question 4

FR: SUMMARY

Answer 4

• no way of telling genuine dif via plot
• safe judgements = variability judgement
• ANOVA applies f-ratio; compares mean variance w/underlying variability
• ANOVA makes automatic/appropriate adjustments of variability to allow statistical stability of mean difs

Question 5

ANOVA SIMPLES USES

Answer 5

• comparing reliability of performance difs (2G)
• establishing genuine difs across 3+ groups
• establishing pairwise difs when comparing single groups
• examining more complex internal comparisons (ie. is a specific score higher/lower than a specific average)

Question 6

SIMPLE CONTRASTS

Answer 6

(ie. comparing G2/3 in 4)
- ANALYSE-GENERAL LINEAR MODEL-UNIVARIATE
- paste; BEFORE method:
- /contrast (var00001) = special (0, -1, 1, 0)
- play
- now look at the results you dingus
( 0 (1/4) = excludes groups; +/- (2/3) = opposite scales of comparison)

Question 7

CONTRAST WEIGHTS

Answer 7

• for comparisons, these are calculated via exactly equivalent logic

Question 8

GOLDEN RULES

Answer 8

• contrast = IV
• contrast total = condition total
• contrasted items = opposite sign
• contrast = 0