ADMS 2320 Lecture Notes - Lecture 9: F-Distribution, Null Hypothesis, Null Set
Document Summary
Take two samples of data and analyze and look at the difference. Population variance can be equal or not equal. You are not allowed to put into pool variance; treat them differently. If not stated in the question then, f-test. The question says equal or not equal; f-test not needed. F distribution has the two region test because there is no negative. Matching is determined when they are using the same thing to get results, like using the same documents or using the same person to produce under method a or b. Matched pairs questions always gives raw data to use. Case 1 no benchmark: null set to zero, allowed to pool proportion. Case 2 does have benchmark: not pool, keep p-hat 1 and p-hat 2 separate. Example 13. 10: outsell the simple design by at least 3%, so this is case 2, not allowed to pool and combine the proportions.