Module 3

Question 1

t or f: if the h0 is very unlikely, you can conclude that the difference/magnitude is large

Answer 1

false

Question 2

Bayesian Stats

Answer 2

• alt to NHST
• argues that NHST logic is flawed
• in the data you calculate a Bayes factor (ratio of likelihood of alt hypothesis relative to likelihood of null hypothesis)
• 1=equal, <1=null more likely, >1=alt more likely (3=moderate/baseline, 10=strong)

Question 3

raw effect size

Answer 3

• unstandardized regression coefficients are raw effect sizes
• the size of the difference between two means and treat it as an indication of magnitude
• works well if variable of interest is on a meaningful metric

Question 4

standard effect sizes indicators in relation to t tests

Answer 4

• cohen’s d
• pearson’s r

Question 5

cohen’s d

Answer 5

• effect size you can apply to a t test
• expresses magnitude as a standard difference between means
• percentage of the SD
• aka ds

Question 6

cohens d for independent sample t test formula

Answer 6

ds=(x1-x2)/pooledS

S=standard deviation around first and second mean

Question 7

pooled S (standard deviation) formula for use in cohens d formula

Answer 7

pooled S= √ (n1-1)s1^2+(n2-1)s2^2/(n1+n2-2)

Question 8

t or f: ds (cohen’s d) has a minimum of 0 and a max of 10

Answer 8

false, a min of 0 (no difference) and no upper boundary
ex. 0.5=dif between means is half size of DV’s SD, 1=difference is just as big, 2=mean difference is twice as big as DV’s

Question 9

ds size guidelines

Answer 9

• 0.2=small
• 0.5=med
  -0.8=large

Question 10

d(av) formula

Answer 10

• cohen’s d for repeated measures
  d(av)=D_/Avg.S

Question 11

formula for Avg.S in d(av) formula

Answer 11

Avg.S=(S1+S2)/2

Question 12

S

Answer 12

standard deviation

Question 13

d(rm) formula

Answer 13

• repeated measures effect size indicator that considers magnitude of correlation between observation sets
  d(rm)= (M diff/√ s1^2+s2^2 x r x S1 x S2) x √ 2(1-r)

Question 14

t or f: d(av) and d(rm) are both equally similar to d(s) exepct when r is low and the difference between standard deviations are large

Answer 14

flase, above is only true for d(av) not d(rm)

Question 15

which d is considered overly conservative when r is large

Answer 15

d(rm)

Question 16

because cohen’s d is a positively biased estimate of pop effect size (especially for small samples), what can be used to correct for this bias

Answer 16

Hedge’s g

Question 17

pearson’s r coefficient

Answer 17

• used to quantify effect size
• determines point biserial correlation

Question 18

r

Answer 18

• used to calculate strength and direction of correlation

Question 19

point biserial correlation

Answer 19

• expressing relationship between dichotomous variable (membership in 1/2 groups) and cont variable (DV)

Question 20

r^2 (pearson’s r)

Answer 20

• proportion of variance in DV accounted for by group membership

Question 21

r ranges

Answer 21

• -1.00 = strong negative
• 0.00 = no relation
• 1.00 = strong positive

Question 22

cohen’s guidelines for r

Answer 22

• 0.10 = small
• 0.30 = med
• 0.50 = large

Question 23

it is necessary to make assumptions about _____ size when doing power calculations

Answer 23

effect

Question 24

t or f: large effect sizes do not directly imply practical significance

Answer 24

• true
• durability of effect may be relevant
• cost/benefit analysis
• metric is hard to interpret

Question 25

when are small effect sizes impressive

Answer 25

- w minimal manipulation to IV - when its difficult to interpret DV

Question 26

existence of an effect ______: a) challenges reigning theory b) differentiates between competing theories c) indicates strength of argument d) demonstrates new or disrupted phenomenon e) all of the above f) a), b), and d)

Answer 26

f)

Question 27

confidence intervals

Answer 27

- mean of your estimate plus and minus the variation in that estimate - calculated using standard error - typically specified 95% (less common=99/90%) - percent chance that the interval you calculate contains the population