Repeated Measures, Mixed Models

Discuss statistics related things

by bencturnbull » Thu Jun 24, 2021 4:46 pm

Hi friends!

What is the best way to set up a mixed model with repeated measures?

I have:

One, continuous, dependent variable.
Two, categorical, between subjects factors.
One, categorical, within subject factor.

Jamovi doesn't appear to have a repeated-measures option built into Mixed procedures. I found a source online suggesting using individual as a random factor (Intercept | ID), but I don't understand enough at this stage to know if that would be appropriate.

Any suggestions and guidance welcome :)
Posts: 1
Joined: Thu Jun 24, 2021 4:34 pm

by jonathon » Thu Jul 01, 2021 3:49 am

take a look here, i think there's some good examples:


User avatar
Posts: 1870
Joined: Fri Jan 27, 2017 10:04 am

by Vit » Mon Sep 13, 2021 6:02 am

I am currently working with data from a within-subject design. Thus I need to set up repeated measures in LMM.
Though I have read the GitHub page, I am unsure whether I am specifying my models correctly.

In model 1, I have a group of participants from whom we took heart rate measures at four time points. I want to know whether there are any differences in mean heart rate between the timepoints.
Is it correct to specify the model like this, or do I need to set up other random effects?
Model1: Heart rate ~ Timepoint + ( 1 | Participant ID )

In model 2, I have a group of participants from whom we measured strength measurements at four time points. At each time point, participants had five attempts to perform the strength task. I am not interested in the interaction time point*attempt.
I want to know whether there are differences in mean strength between the timepoints and whether the attempts affected the strength.

How do I specify this LMM?
Model2: Strenght ~ 1 + Time point + Attempt+( 1 | Participant ID )

Thank you for your suggestions.
Posts: 42
Joined: Fri Apr 19, 2019 11:51 am

by » Fri Sep 17, 2021 4:17 pm

your models are fine, with some caveats. Model 1 is fine like it is, but for model 2 I'd suggest including the interaction "time*attempt" also if you are not interested. This is because if the heart rate changes over time differently in the two groups, so there's an interaction, interpreting the main effects becomes more difficult. So, Estimate
Model3: Strenght ~ 1 + Time point * Attempt+( 1 | Participant ID )
if the interaction is not significant, go for Model2, otherwise, interpret Model3 results
User avatar
Posts: 272
Joined: Thu Mar 23, 2017 9:24 pm

by Vit » Mon Sep 20, 2021 6:05 am

Perfect, thank you!
Posts: 42
Joined: Fri Apr 19, 2019 11:51 am

Return to Statistics