Chi-square is significant. I know that there is an effect of experimental manipulation. But how can I check which groups between A, B, C, E and D differ? I would like to know the difference between group
Please give advice.
I believe this is when you would run standardized residuals which would be the observed minus the expected outcomes divided by the squared root of the expected values. If your alpha is set at 0.05 and the standardized residual you have is greater than 1.96 then you find which group is contributing to the statistically different chi square. However, if you are testing within the same sample with multiple tests you would have to account for family-wise error by dividing the alpha level by the number of tests (i.e. alpha = 0.05 with 5 groups would have a new alpha of 0.01 and the criterion for comparison would be 2.58). Im new to this statistic myself, but hope this helps.
For 2 x C crosstabulations, there is a Tukey like SAS macro:
http://www2.sas.com/proceedings/sugi31/204-31.pdf
For other cases, you can do this using the SAS GENMOD and MULTTEST procedures, see attachment.
In R: https://www.youtube.com/watch?v=KzEROLH3afg
https://www.rdocumentation.org/packages/fifer/versions/1.1/topics/chisq.post.hoc
https://stats.stackexchange.com/questions/85664/r-procedure-for-comparing-multiple-categorical-variables-similar-to-anova-fol
https://cran.r-project.org/web/packages/coin/coin.pdf
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running