EC295 Lecture Notes - Lecture 3: Null Hypothesis, Statistical Hypothesis Testing, Test Statistic
Document Summary
The value of a parameter is never actually known, that is why we do estimation. Make conditional claim about a parameter, and evaluate whether the claim is likely true given random sampling (estimate can differ from claim because claim is false, or random sampling) Hypothesis testing: a process of confronting tentative beliefs with evidence and deciding whether those beliefs can be accepted or rejected. Four steps for doing hypothesis testing: formulating opposing hypotheses, deriving a "test statistic, deriving a decision rule, using sample data to compute the test statistic and confronting it with the decision rule. Test statistic: a statistic computed from a random sample, used for establishing the probable truth of null hypothesis (assuming the null hypothesis is true) Deriving a decision rule: generally, reject null hypothesis if estimate of t is unusual given random sampling, accept it otherwise. Significance level: among all possible values of t, a is the maximum proportion unusual enough to reject null hypothesis.
