PSYC 305 Lecture Notes - Lecture 7: Type I And Type Ii Errors, Null Hypothesis, Statistical Hypothesis Testing

• School: McGill University
• Department: Psychology
• Course: PSYC 305
• Professor: Heungsun Hwang

Document Summary

To test whether k > 2 populations significantly differ. H1 = not all are the same we do not know which ones are different: what else can we do aside from calculating the fratio, we want to know where the differences are happening. Post-hoc comparisons: used if >3 means were compared. We only look at these once we have finished the experiment. 2 kinds: 1) sheffe and 2) tukey"s shd: you want to choose what kind of test you want before you gather the results by determining the parameters and implications. We use this when the groups have different sample sizes. More robust to assumption violations less sensitive to divergence from normality/assumptions/equal variance in the population. Most conservative test where it is very unlikely to reject h0. It will reject h0 less often than tukey"s hsd. Uses f ratio to test for a significant difference between any 2 means (e. g. h0: 1 = 2, or 2 = 3, or.