I'm trying to do an ANOVA for the question: does the search term used (4 different categorical) affect readability (continuous)? There are 100 rows total, with each term having 25 values. Readability is measured in 8 different ways (8 variables) but I want to assess all four search terms versus one readability variable at a time. I input the dataset but when I used this command: proc glm data= mydata; class searchterm; model words = searchterm ; means searchterm/hovtest welch; run; there was no result/ANOVA tables. It just showed my dataset in the outcome tab. I tried creating a new dataset of the average of the 25 hits for each search term (so there were only 4 rows, one for each search term), but the ANOVA test that resulted didn't have any p values. (Maybe because you need to have n>30 or assume normal distribution?) I also have a feeling this isn't the right way to use an ANOVA but I don't know how to do it using the first dataset, which I think is the correct way to input the data Screenshots for the dataset made of averages: Even trying to run a Kruskal-Wallis test for both datasets yielded the same results (no tables for the first version and no p values for the second) Command used (adjusted for each dataset): proc npar1way data=mydata wilcoxon dscf; class searchterm; var words; run; Thanks in advance!
... View more