BIO 301D Lecture Notes - Lecture 22: Statistical Significance, Statistical Hypothesis Testing, Scientific Method
Document Summary
Experiments are manipulations of nature: usually done to destroy unwanted correlations with third variables. Two types of experiments: 1) you do not know the third variables, which may cause the problem. So, randomly assigned to treatment and control groups. Preexisting variables are not eliminated but they become balanced. If we could randomly assign color and leave everything else alone. Randomization needs to apply at the step of assigning to treatment vs. control. Statistical test is a test of a null model: usually null model is of sampling error. That sampling error could account for the observed differences. If the differences could be explained by sampling error, then we don"t care about them- they are not real: when you do your test, you get a p value (must range between 0-1) If p < . 05, we consider the differences significant . Whatever the p value is how often we will make a mistake in rejecting the null model.
