I have performed a Chi-square analysis to see if there are differences between the type of food (6 types of food) consumed between 2 groups. Chi-square gives me a p less than 005, but I do not know how to know which levels of the variable type of food (that is, what type of food) is consumed significantly or less by one of the groups. How would this be calculated?
Did your stats package output a frequency table as well as the chi-square results?
One approach would be a post-hoc testing to compare the counts between groups for each food. You could use an exact binomial test or a chi-square goodness-of-fit test, with the null hypothesis that the counts for one group follows the proportion 0.50.
A different approach would be if your software will give you the standardized residuals. These are interpreted as a z-score. So that a standardized residual > 1.96 or
If you are familiar with R software, the "chisq.post.hoc()" function embedded in the R package called "fifer" could assist you to run a posthoc multiple comparisons as Salvatore suggested.
Thank you very much! I use SPSS, and I think that I could use de Z-score to calculate p values and adjust them with Bonferroni corrections.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude