ECO220Y1 Lecture Notes - Lecture 1: Null Hypothesis, Sampling Error, Alternative Hypothesis
ECO220Y1 verified notes
1/26View all
Document Summary
If observed data are surprising (i. e. not very plausible) given the null hypothesis then reject h0 . Formal statistical methods calculate exactly how surprising the data are conditional on h0 being true. If sample gives conclusive evidence, then reject the null (h0) and infer the research hypothesis (h1) is true (it has been proven) If inconclusive (weak) evidence in favor of h1, then fail to reject h0. Asymmetry: cannot infer h0 is true but can reject h0 to prove h1 . Researcher hopes to prove h1: the research hyp. is more important. Summary of evidence (one number) in a sample to compare with the initial presumption. Are the data surprising, given the null hypothesis? . Sampling error cannot plausibly explain why data differ so much from the initial presumption. As it gets closer to 0 the strength of our evidence for h1 increases. If the normal value of being bigger or smaller than.