Suggestion needed on how to test for differences between three percentages/proportions in SPSS. Non parametric chi-square tests for differences between percentages, but does not offer post hoc testing option. My data is set-up in the same format as attached link, but post hoc testing needed as I have three groups. Thanks.
https://www.youtube.com/watch?v=9dIBsa92QHY
Maybe you repeat the test pairwise after your global chi-square test suggests a significant difference and apply a bonferroni correction when interpreting the pariwise comparisons.
Hope that helps.
What actually are you trying to find? There might be other ways to reach the same goal.
You need to have a look at residuals or need to consider partioning Chi-square. Agresti's book Categorical Data Analysis talks about this.
The following might help you too
https://www.researchgate.net/post/how_can_we_perform_a_post-hoc_for_chi-square_test
Thank you for your responses. I am trying to test for differences in proportions within groups (and between groups if possible). Please see attached figure. RHR = resting heart rate, HRR = heart rate reserve. I can perform a chi-square to derive a 3 x 3 table with post hoc testing, but the results do not seem appropriate (please see attached excel output). Thanks for any insight you may be able to provide. Please let me know if more detail is needed.
Jennifer does that bar plot obtained from the table in excel sheet?
There are some zero cells in the sheet. I assume you want a post-hoc test of that table. But the figure seems different. Did you merge some rows/columns?
What results are inappropriate?
The figure is correct. The chi-square results in the excel output are inappropriate as the same participants are in the RHR+20 and RHR+30 groups.
I need to run a 3 x 1 table with post hoc testing. I would like to test whether the 3 black bars (percentages) are different from each other, and whether the 3 grey bars (percentages) are different from each other. Parametric chi square in SPSS forces a variable to be entered for "columns" and a variable to be entered for "rows." This does not make sense for me given the same participants are in both variables (i.e. RHR+20 and RHR+30 groups). I can run a non parametric chi square in SPSS (it only forces one variable to be entered as shown in the video I linked), but it does not allow for post hoc testing.
So...I think I am at a loss with SPSS, unless you have any ideas on how to run a 3 x 1 with post hoc testing. Thanks again!
I do not understand your study design and analysis.
Isn't "the nonparametric chi-square" as you called, the post-hoc test you are trying to find? Why not to do it to test if those in grey bars are equal to 33.3%?
I have not seen such a parametric/nonparametric chi-square distinction so far. Do you have some references?
By the way, there is no video link, but not important.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS