Interpretation of Results, Effect Sizes and 3-Way ANOVAs

Question 1

INTERPRETATION OF MAIN EFFECTS (ME)

Answer 1

(using plots)

• DAY/NIGHT ME: mean day is STATSIG to mean night
• FOGGY ME: mean clear is STATSIG to mean foggy
• (DAY/NIGHT) FOGGINESS: D/N STATSIG different in fog than clear

Question 2

GRAPH INTERACTIONS

Answer 2

• parallel; non-interacting

- non-parallel; interacting

Question 3

INDEPENDENCE OF SOURCES OF VARIATION

Answer 3

• because 2 MEs of a 2-way interaction are orthogonal, evidence for each is independent from the other; allows for various combinations of effects resulting in different interactions

Question 4

MAIN EFFECTS (MEs)

Answer 4

1. D/N = distance means affected by day/night
2. F-ME = distance means affected by foggy/clear
3. 2WI = mean dif in day VS night affected by whether foggy VS clear

Question 5

BE CLEAR!

Answer 5

• “Distance estimates are affected by both fogginess and time of day at which the test is done” = WRONG!
• could describe any outcome

Question 6

KEYWORDS

Answer 6

MAIN EFFECTS = tell you something about the differences in performances that occur when an IV is manipulated (ie. effect of day VS night)
INTERACTIONS = tell you about differences between differences (or more generally how one IV can be influenced by another)

Question 7

EFFECT SIZES

Answer 7

• STATSIG indicates genuine dif; STATSIG more bothered about whether a difference is there rather than how large the effect it
• HOWEVER effect size still influences SIG alongside sample size
• quantify effect size rather than just STATSIG status

Question 8

EFFECT SIZES (EXAMPLE)

Answer 8

• input data in c2
• input levels in c1
• scales
• ANALYZE-GENERAL LINEAR MODEL-UNIVARIATE
• options; estimates of effects size
• assign variables = OK

Question 9

ES: COHEN’S GUIDELINES

Answer 9

• no effect < .01
• small effect > .01
• medium effect > .06
• large effect > .14

Question 10

3WA: EXAMPLE

Answer 10

• 3WA is the limit

- ie. age (young/old); gender (male/female); IQ (low/high); DV = average salary (k)

Question 11

3WA

Answer 11

• input data in c4
• input levels in c1/3 and title them
• ANALYZE-GENERAL LINEAR MODEL-UNIVARIATE
• assign variables = OK

Question 12

SUMMARY

Answer 12

• 3WA works the same as OWA/2WA but with an extra column; extra info = extra output
• for STATSIG, plot data to interpret results