Statistical Inference & Wilcoxon Rank Sum

Question 1

Inferential Stats

Answer 1

• Mean, median, mode etc. are descriptive; most stats is INFERENTIAL=
  o are patterns true of pop or are they sample illusions ie. chance
  o assume no dif in pop ie. sex (NULL HYPOTHESIS (H0))
  o prob of null via observed dif in desc
  o compare null prob to “criterion of truth” (STATSIG (p>.05))

Question 2

Tied Ranks (EG)

Answer 2

o 6=1, 8=2.5, 8=2.5, 9=4

o 6=1, 8=3, 8=3, 8=3, 9=4

Question 3

Null Probability (EG)

Answer 3

o Stat test=test stat (ie. F ratio, t/W/Z)
o Two-tail=non-directional/<2.5%> (common but not v STATSIG) (do they differ etc.) OR One-tail=directional/>5% (more likely to be STATSIG; used w/ previous scientific evidence) (they DO differ)?
o converted to prob via stat table

Question 4

WRS

Answer 4

• 2 dif groups; HOW MUCH DOES THEIR DATA OVERLAP? (ie. 12, 14, 16, 18=high); high=likely chance; low=unlikely chance

Question 5

WRS (EG)

Answer 5

• EG. “Women have a larger vocab than men”; impossible to use whole pop; “random” sample; RESULTS (unlikely to be the same even w/o sex dif):
  o Women= 32, 35, 40, 29, 38, 31=34.2
  o Men= 34, 16, 24, 18, 12, 22=21
• Informative or chance?? Dif of sex; women get 13.2 more AV; prob of rep in 12 if men=women in pop?

Question 6

WRS 1

Answer 6

o 1. RANK DATA (between groups!!)

o 1=12, 2=16, 3=18, 4=22, 5=24, 6=29, 7=31, 8=32, 9=34, 10=35, 11=38, 12=40

Question 7

WRS 2

Answer 7

o 2. CALC RANK SUM OF SMALLER N GROUP (take small of 2 if both =)
o Women= W=8+10+12+6+11+7=54
o Men= W=9+2+5+3+1+4=24
o W=24

Question 8

WRS 3

Answer 8

o 3. O-T/T-T?

o No prior research, so T-T.

Question 9

WRS 4

Answer 9

o 4. STATSIG IF W== TO WRS TABLE
o N1=small group (6); n2=larger group (6)=26
o 24<26, SO P>.05; STATSIG! Reject H0
o Males VS females differ more than chance.