PSC 41 Lecture Notes - Lecture 16: Statistical Inference, Type I And Type Ii Errors, Null Hypothesis

• School: UC-Davis
• Department: Psychology
• Course: PSC 41
• Professor: Cross, Victoria

Document Summary

Power is the probability that a test will correctly reject a false null hypothesis (the probability of finding a difference that does exist): power = 1 (probability of type ii error). Can be used to calculate the minimum sample size: higher desired power demands a greater sample size. Sample size: increase sample size = increased power. Magnitude of the effect: smaller effect = demands more power to detect. Alpha level (type i error rate: smaller -level will demand more power. Between- or within-groups design: within-groups designs have more power. Effect size: a statistical measure that conveys information concerning the magnitude of the effect produced by the predictor variable. Effect sizes range: small effects near d = . 2 (~90% overlap, medium effects near d = . 5 (~80% overlap, large effects above d = . 8 (~70% overlap) Smaller effect sizes require larger samples to accurately detect.