Wilcoxon Matched Pairs

Question 1

Group Difference

Answer 1

o Subjects can be related UNRELATED/BETWEEN (ie. is female vocab better (DV=vocab; IV=gender) or RELATED/WITHIN (ie. are wives vocab better than their husbands (DV=vocab; IV=marriage???)
o Most common relation is same person under 2 CONDITIONS.
o Differences are:
 Due to CONDITION (REPEATED MEASURES/MATCHED PAIRS): ie. DV=blood pressure; IV=drug treatment

Question 2

WMP (EG)

Answer 2

• Eg. HP=Alcohol consumption increases reaction times (RT); DES=Simple 2 choice reaction time task before (RT1) & after (RT2) consumption of 6u alcohol (7pp;2c)
• RT1=330, 285, 420, 281, 263, 381, 550; RT2=298, 687, 952, 305, 263, 409, 590

Question 3

WMP

Answer 3

o Wouldn’t psychologically expect exact same RT in 2c (aka RT2-RT1=0 is NOTPROB); this would happen if alcohol had no affect (NULL)
o If alcohol slows RT=RT2-RT1=+!
o WMP finds stat showing H0 extent; table shows STATSIG prob

Question 4

WMP 1

Answer 4

o	1. CALC DIF
o	RT2(!!!)-RT1= -32, 402, 532, 24, 0, 28, 40

Question 5

WMP 2

Answer 5

o 2. REMOVE 0 (AND REDUCE N (IE. 7 TO 6)

Question 6

WMP 3

Answer 6

o 3. RANK DIF

o -32=3, 402=5, 532=6, 24=1, 28=2, 40=4

Question 7

WMP 4

Answer 7

o 4. CALC SUM OF + RANKS

o T+=18

Question 8

WMP 5

Answer 8

o 5. CALC SUM OF – RANKS

o T-=3

Question 9

WMP 6

Answer 9

o 6. T IS SMALLER

o T=3

Question 10

WMP 7

Answer 10

o 7. STATSIG IF T=/< CV

o 3>1; accept null; 6 units doesn’t significantly affect reaction time, so NOT STATSIG

Question 11

Variables

Answer 11

• IV can be categorical/nominal (ie. men vs women); DV can be continuous/scale (ie. RT)

Question 12

Categorical Responses

Answer 12

• Responses can also be categorical if:
  o a continuous variable is categorised (ie. large vocab=>20; small=<20)
  o a true categorical response is used (ie. yes/no)
• What is truly being measured is the frequency of categorical responses.

Question 13

Frequency Pre-Requisites

Answer 13

o Show data as COUNTS (ie. 10 toddlers, 10 kids) not percentages (ie. 50% toddlers, 50% kids); convert if otherwise:
 P(kid) x sample size=frequency of kids