E

I am trying to program SAS to analyze the data in the following problem using Cochran's test for related observations (a nonparametric test). The problem is located on p. 257 of Conover's Practical Nonparametric Statistcs (3rd Ed.).

The textbook question is reproduced below. I have attached a SAS file, with my attempt at programming this problem into SAS. The output gives the wrong answer, which is probably because of my entering the data incorrectly. The correct answer should be T=10.4 (which is what I get when I do the problem by hand and what the textbook answer key states).

Any help at correcting my poor data entry ability would be greatly appreciated. PROC FREQ with the /CMH is the correct command for Cochran's test, so the issue here is determining where my data entry error lies (not talking about a typo here, but rather entering the data in the wrong manner into SAS).

Question Details: (Excel spreadsheet with raw data included in zip file below)

In an attempt to compare the relative power of three statistical tests, 100 sets of artificial data were generated using a computer. On each set of data the three statistical tests were used, with α˜0.05 , and the decision to accept or reject the null hypothesis was recorded. Is there a difference in the power of the three tests when applied to populations from which the simulated data were obtained?

Test 1 Test 2 Test 3 Number of Sets of Data

Accept Accept Accept 26

Accept Accept Reject 6

Accept Reject Accept 12

Reject Accept Accept 4

Reject Reject Accept 18

Reject Accept Reject 5

Accept Reject Reject 7

Reject Reject Reject 22

The textbook question is reproduced below. I have attached a SAS file, with my attempt at programming this problem into SAS. The output gives the wrong answer, which is probably because of my entering the data incorrectly. The correct answer should be T=10.4 (which is what I get when I do the problem by hand and what the textbook answer key states).

Any help at correcting my poor data entry ability would be greatly appreciated. PROC FREQ with the /CMH is the correct command for Cochran's test, so the issue here is determining where my data entry error lies (not talking about a typo here, but rather entering the data in the wrong manner into SAS).

Question Details: (Excel spreadsheet with raw data included in zip file below)

In an attempt to compare the relative power of three statistical tests, 100 sets of artificial data were generated using a computer. On each set of data the three statistical tests were used, with α˜0.05 , and the decision to accept or reject the null hypothesis was recorded. Is there a difference in the power of the three tests when applied to populations from which the simulated data were obtained?

Test 1 Test 2 Test 3 Number of Sets of Data

Accept Accept Accept 26

Accept Accept Reject 6

Accept Reject Accept 12

Reject Accept Accept 4

Reject Reject Accept 18

Reject Accept Reject 5

Accept Reject Reject 7

Reject Reject Reject 22

Last edited by a moderator: