nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Chi Square Test with Tukey adjustment for multiple comparisons

Claudia92

Hello,

I'd like to compare various categorical variables across 3 different groups of participants that were previously matched for age and BMI. I've done contingency tables (3x3, sometimes 3x4) and observed significant overall differences across my 3 groups (p-value for chi-square test <0.0001). However, since I have more than 2 groups, I'm unable to identify where exactly is the difference (group 1 vs 2, 1 vs 3 or 2 vs 3). Someone told me I could use a Tukey adjustment for multiple comparisons but I don't see this option anywhere... Am I using the correct analysis/pathway?

Thank you for your help!

Accepted answers

Mark_Bailey

The Nominal Logistic and Ordinal Logistic platforms in JMP do not provide multiple comparisons. You would need to perform a lot of manual calculation after JMP.

The Generalized Linear Model platform in JMP provides Custom Tests. You can specify more than one test and an additional omnibus test is added over the set of all tests.

The Generalized Regression platform in JMP Pro does provide a variety of multiple comparison methods, including Tukey.

All comments

Mark_Bailey

The Contingency platform offers dome alternatives to a Tukey multiple comparisons of the mean. The Analysis of Means for Proportions command in the platform menu (red triangle) compares the mean proportion of the response for each group to the mean proportion. The Correspondence Analysis command produces a rigorous visualization of the associations between levels of the response and predictor variables..

Claudia92

Thank you for your reply.

Is there any other way I could perform a chi-square test (other than with the contingency platform) for which the Tukey adjustment is available/offered?

Thanks again!

Mark_Bailey

The Nominal Logistic and Ordinal Logistic platforms in JMP do not provide multiple comparisons. You would need to perform a lot of manual calculation after JMP.

The Generalized Linear Model platform in JMP provides Custom Tests. You can specify more than one test and an additional omnibus test is added over the set of all tests.

The Generalized Regression platform in JMP Pro does provide a variety of multiple comparison methods, including Tukey.

Claudia92

Great, thanks!