Hi all,
I want to do a repeated measure Ancova with constant covariates. The IBM SPSS website mentions you could do it using a two-step approach in this software (https://www.ibm.com/support/pages/repeated-measures-constant-covariates-glm). Can someone please verify this is a proper way to do it?
Many thanks in advance!
When you say "constant covariate", do you mean that it is constant for each subject? I.e., the repeated measures at Level 1 are clustered within subjects at Level 2, and you are talking about a Level 2 covariate? If so, see the example here:
- https://imaging.mrc-cbu.cam.ac.uk/statswiki/FAQ/wmixed
HTH.
Thanks for the reply Bruce. I measured my DV three times (within-subject), but the covariaties (i.e., age, gender etc.) do not differ across those measurements and are therefore constant. Hence, I only want to check for them in the between-subject effects table (not within-subject effects table). This is also in line with my conceptual model. I indeed saw the post you attached. However, I do not quite understand it and I was therefore wondering if the two-step GLM approach, proposed by IBM, is okay? If yes, I really would prefer to approach it this way, because I understand it.
Hello Paul Valk As far as I understand and try to learn is that in a RM study our covariates can a) change like dependents for example in each time step compared to another time step. For example, if our covariate is the ambient temperature and we have measured DVs in 5 different times, then it is clear that we have 5 different measurements of the temperature covariate for each time. b) Covariates can remain constant from the beginning of the study (maybe even until the end of the study). For example, as you mentioned, the gender of people.
I think it is better to do the following process as recommended by IBM.
To get the classical repeated measures ANCOVA results for repeated measures with constant covariates --like those modeled in Winer (1971)-- you'll have to run two GLM models. Run the first model with the covariates, but only report the between-subjects portion of that analysis. Run the second model without the covariates, but only report the within-subjects (WS) portion of that model.
Hi Abolfazl,
Thanks a lot for your reply. I indeed agree that two types of covariates exist. However, in my study, only constant covariates exist (e.g. age/ gender). Hence, I indeed followed the approach by IBM you mentioned. However, I was just wondering if that approach is valid, as it feels weird to do run it twice and report results from each test, rather than one. Nevertheless, this might also just be because SPSS offers default options and mine is not included. As long as the approach you mentioned is valid I'm good! Thanks for the help!
Paul Valk, did you read this bit at the bottom of the IBM Support page you found?
Finally, there are two alternative ways to estimate Winer's model in SPSS. First, you can run the model using commands in the older SPSS MANOVA program. Second, you can run the model in SPSS MIXED. For a worked example of how to model Winer's (1971; p. 803) example in MIXED, see Resolution 22273. Both MANOVA and MIXED adopt Winer's convention. A constant covariate is partialled from the between-subjects effects, but it is neither partialled from nor does it interact with the WS effects.
Here is the link for Resolution 22273:
- https://www.ibm.com/support/pages/node/418471
This circumvents the need to estimate two models (using GLM), and is the preferred approach, IMO.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size