Power (pwr)

Question 1

alpha convention

Answer 1

.05

Question 2

power convention

Answer 2

.8

Question 3

power (definition)

Answer 3

probability of correctly rejecting a false null hypothesis - when we find an effect that exists

Question 4

factors that affect power

Answer 4

effect size
sample size
significance level

Question 5

effect of sample size on power

Answer 5

power increases as sample size increases

Question 6

effect of effect size on power

Answer 6

power increases as parameter values move further into H1 (alt hyp) values and away from H0 (null hyp) values
power increases as effect size increases

Question 7

effect of significance level on power

Answer 7

power increases as significance level increases

Question 8

why should we not increase the significance level to increase power?

Answer 8

leads to more type 1 errors

Question 9

effect size (definition)

Answer 9

‘detectability’ of alternative hypothesis - comapres distance between null and alt hypothesis to variability in data

Question 10

pwr.t.test code

Answer 10

n = sample size
d = effect size
sig.level = significance level (alpha). default = .05
power = power level
type = type of t-test to perform, e.g. 2-sample (two.sample), 1-sample (one.sample) or dependent sample t-test (paired). default = 2sample
alternative = if alt hyp is two-sided (two.sided) or one-sided (less or greater). default = twosided

Question 11

code for finding power for linear regression

Answer 11

pwr.f2.test

Question 12

pwr.f2.test code

Answer 12

u = numerator degrees of freedom (predictors in model = k)
v = denominator degrees of freedom (n - k - 1)
f2 = effect size
u and v come from study deisgn -

Question 13

code for finding power for correlations

Answer 13

pwr.r.test

Question 14

pwr.r.test code

Answer 14

n = sample size
r = effect size
sig.level = significance level
power = power