topic Re: Difficulties with one way Repeated Measures ANOVA in Discussions

Difficulties with one way Repeated Measures ANOVA

kestevens — Tue, 18 Aug 2026 18:31:55 GMT

Everyone,

This seems like a very easy analysis but we seem to be having difficulty getting it to run properly. We are interested in comparing water quality at several sites along a small stream (8 sites). We obtained samples weekly for 9 weeks (one sample per site per week). The question we are interested is "are there differences in water quality among sites?" Our columns include site, week and then one column for each water quality parameter. We will analyze each water quality parameter separately. Because the data is stacked, we are using the mixed model and JMP Student Version 19.1.4. We had previously used SAS for repeated measures analysis and while JMP looks much easier we seem to be stuck. Is there an example in the data base someone could point us to that shows a similar analysis using JMP Student Version 19.1.4? Thanks in advance.

Re: Difficulties with one way Repeated Measures ANOVA

rcast15 — Tue, 18 Aug 2026 19:22:08 GMT

Hi @kestevens,

Repeated measures mixed models can be run in Fit Model using the Mixed Model personality. See the link below for an example and some documentation.

Example of Repeated Measures

Re: Difficulties with one way Repeated Measures ANOVA

statman — Tue, 18 Aug 2026 21:35:50 GMT

Minor point, you tell us the columns include site week and then water quality estimates. Your data table has "sampling Event, and Zone". I assume zone is site and sampling event is week, but you should stay consistent with terminology.

The answer to your question "are there differences in water quality among sites?" is easy, Yes there are differences.

Whether those differences are meaningful, of practical significance or statistically different are separate questions. Have you looked at any correlation of your Y's? Should they correlate? (which ones should and should not correlate?). Are there any outliers? Mahalanobis can be used.

How much of a change in each Y is of practical significance (meaningful)? What is the measurement error for each response?

I would start with graphical analysis. I have run a couple and saved the script to the data table. You might have other graphs you prefer.