STAT W 1001x Study Guide - Midterm Guide: Null Hypothesis, Contingency Table, Relative Risk
11 Sep 2016
School
Department
Course
Professor
Document Summary
Chi-sq test: test of independence in contingency table. Assumptions: counted data, independence assumption (randomized condition), sample size assumption (expected cell freq > s) Degrees of freedom: (rows-1), (columns-1) always one sided. Comparing 2 independent sample means h0: sample population mean, ha: different population mean. Assumptions: independence (randomization & 10%), normal (n>40 or normal model distribution), independent gps assumption. Confidence interval (y1-y2) t*df x se(y1-y2) Assumptions: independence assumption (randomization, 10%), sample size assumption (success/failure) Ultimate goal of hypothesis testing is to obtain a p-value: if p-value is small enough ( smaller than level of sig. ) we reject the null hypothesis. Type 1 error: rejecting null hypothesis when its true (=alpha) Type 2 error: failing to reject null when it is false (=beta) Power of test is its ability to detect false null hypothesis (or true alternative hypothesis) (=1-beta) Very high probability of making a type 2 error: low power. Conditional probability rule-probability of b given a.
