COMM 291 Lecture 3: C291_2010_Lectures16171819.doc
16
COMM 291 Full Course Notes
Verified Note
16 documents
Document Summary
Commerce 291 lecture notes jonathan berkowitz (copyright, 2010) Two-sample problems refer to inference for comparing responses in two groups; each group is a sample from a distinct population. Responses in one group are independent of those in the other group. We are interested in estimating 1 2: An obvious (and also good) point estimate is: 2x ) = 1 2 (i. e. the estimate is unbiased) 2x ), simply replace the population variances by the sample variances. The two-sample t-test statistic is: t = x. Although this test statistic has a t-distribution, it is only approximate. There are three choices for what to use as the degrees of freedom: the simplest choice is: df = minimum of (n1-1, n2-1) This is the safest, most conservative choice: use the sum of n1-1 and n2-1. Two-sample t-test for comparing two independent means . Ho: 1 2 = 0 this can also be written as: ho: 1 = 2.
