PS296 Lecture Notes - Lecture 17: Null Hypothesis, Type I And Type Ii Errors, Sample Size Determination
WEEK 9 – LECTURE 17
Factors Affecting Power:
1) Significance level (alpha)
-increasing alpha decreased beta, thereby increasing power
2) Difference between the H0 and H1 (mu0 – muA)
-increasing the distance decreases the overlap of the curves, thereby increasing power (increasing effect
size)
3) Sample size (n) and variance (s2)
-as n increases the variance decreases (less spread), which increases power
4) Type of test, and one vs two tailed test
-rejection region is larger for one tailed test, so generally power will be greater
Recall:
Power = 1 – beta which is 1 – the probability of making a type II error
-power represents the probability of finding a signifcant difference when the population means are, in
fact, different
Delta (δ):
-represents the power of an experiment based on the effect size (dhat) and the sample size (n)
δ = dhat[f(n)]²
