Hi Team,
I am working with a professor on his Market Research course. He is moving from SPSS to JMP. He is finding challenges in doing these tests in JMP (he is also comparing the results to SPSS)
I have attached two datasets along with output from SPSS. I am also finding challenges in replicating the results for McNemar Test, Wilcoxon Rank Test and Friedman Test. (SPSS results are in PDF)
Appreciate your help in this regard
Regards
Murali
I wasn’t able to reproduce everything perfectly, but hopefully good enough for this professor.
McNemar’s test is in Fit Y by X under the Agreement Statistic option. It says that this produces Bowker’s test, which is equivalent to McNemar’s test when on a 2x2 table. The p-value is 0.0143, which doesn’t match the slides (p=0.031), but the slides indicate that the binomial distribution is being used by SPSS, so that may explain the discrepancy.
Wilcoxon’s Signed Rank test for paired samples is in the Matched Pairs platform. Its report doesn’t include all the statistics that SPSS does, unfortunately. But the p-values match.
Friedman’s test is in Fit Y by X, but the data table needs to be in tall format: one col for respondent (Block), another for ad type (X, Factor), and another for rating (Y, Response). Then in Oneway’s non-parametric submenu, Friedman is the only option. The reported chi-square approximation matches what’s in the slides.
