EEB225H1 Lecture Notes - Lecture 19: Biostatistics, Sampling Distribution, Null Hypothesis
18 Dec 2019
School
Department
Course
Professor
Document Summary
Simulation: used for hypothesis testing, mimics sampling from population under null, frequency values obtained for test statistic is approximative null distribution for hypothesis test. Randomization (aka resampling): used for hypothesis testing on measures of association, mixes real data randomly, done without replacement. Bootstrapping: method for estimation and confidence intervals, approximates sampling distribution of an estimate (using resampling) Bootstrap standard error: standard deviation of bootstrap replicate estimates obtained from resampling the data. Assumptions: random sample, large enough that frequency distribution. Useful when assumptions of standard statistical methods cannot be met is a good approximation to the population. Simulations provide frequency distributions of test statistics. Randomization provides frequency distributions of test statistics under the null hypothesis but using given data.
