Hi all,
I have a longitudinal study design of Parkinson's patients (about 8 years follow-up data) and I want to run a model that will examine how clinical variables (ex. motor function) influence functional ability. For this I wanted to run a generalized linear mixed model (GENLINMIXED) in SPSS and I am having some trouble setting it up.
My dependent (outcome) variable is the Schwab and england score, which has a skewed distribution therefore the need for a generalized model. As fixed effects I have months of follow-up, age, education, motor severity (etc. etc.). Random effects are the intercept.
Now my main problem is that in SPSS I can include a repeated measure - however it is difficult to find information on 1) whether I should include one and 2) what to put in. For example, I have seen a study including time as the repeated measure, and another which used age because age was repeated each visit?
Could anyone guide me on the use of the repeated measures in GLMM? Running the models with and without the 'repeated_measures' part gives very different results unfortunately.
I have pasted my syntax below, the bolded part is what I am confused about. EVENT_months is the visit date (0, 12 months, 24 months, etc) and FU_months is the actual months elapsed (which will be different for each participant).
GENLINMIXED
/DATA_STRUCTURE SUBJECTS=PATNO REPEATED_MEASURES=EVENT_months
COVARIANCE_TYPE=DIAGONAL
/FIELDS TARGET=MSEADLG TRIALS=NONE OFFSET=NONE
/TARGET_OPTIONS DISTRIBUTION=GAMMA LINK=LOG
/FIXED EFFECTS= FU_months AGE_AT_VISIT gender EDUCYRS race (other predictors here)USE_INTERCEPT=TRUE
/RANDOM USE_INTERCEPT=TRUE COVARIANCE_TYPE=VARIANCE_COMPONENTS SOLUTION=FALSE
/BUILD_OPTIONS TARGET_CATEGORY_ORDER=ASCENDING INPUTS_CATEGORY_ORDER=ASCENDING MAX_ITERATIONS=100
CONFIDENCE_LEVEL=95 DF_METHOD=RESIDUAL COVB=MODEL PCONVERGE=0.000001(ABSOLUTE) SCORING=0 SINGULAR=0.000000000001
/EMMEANS_OPTIONS SCALE=ORIGINAL PADJUST=LSD
/SAVE PREDICTED_VALUES(DemoPredict2).
Thanks.
Hello Sara Becker. I do not have my copy with me right now, but I would not be surprised if you could find some relevant material/examples in this book:
https://www.routledge.com/Multilevel-Modeling-of-Categorical-Outcomes-Using-IBM-SPSS/Heck-Thomas-Tabata/p/book/9781848729568
You may not find an example matching your scenario exactly, but surely, you will find examples showing how to handle repeated measures over time (which you can tweak as needed). The U of Calgary library probably has a copy of it.
Good luck with your work.
You could refer to chapter 12, “Repeated Measures Mixed-Modeling (20 Patients),” of Cleophas and Zwinderman’s (2016) book for guidance. Here is the full citation.
Cleophas, T. J., & Zwinderman, A. H. (2016). SPSS for starters and 2nd levelers. Springer International Publishing. https://doi.org/10.1007/978-3-319-20600-4
Good luck,
Hello Mohialdeen Alotumi. I am not familiar with that book, and I do not have institutional access to Chapter 12. Does Chapter 12 discuss both the GENLINMIXED and MIXED commands, or just the latter? Thanks.
Here is the contents of the https://tandfbis.s3.amazonaws.com/rt-media/pp/common/sample-chapters/9781848729568.pdf
Heck et al book
This site is very useful https://www.pilesofvariance.com/
Longitudinal Analysis: Modeling Within-Person Fluctuation and Change (by Lesa Hoffman)
and includes SPSS syntax:
Hello, Bruce Weaver. It discussed both. In particular, it provided data example and rendered SPSS commands for running the restructure data wizard analysis, linear mixed model analysis (for fixed effects), and generalized linear mixed model analysis (for random effects).
Best,
Thanks for clarifying, Mohialdeen Alotumi.
Kelvyn Jones, I agree that Lesa Hoffman's 'pilesofvariance' page is a great resource. But as I look at the TOC for her book, I suspect that only the final chapter (13) would have any material concerning the GENLINMIXED comand that Sara Becker is using. And no code (for any package) is provided for Chapter 13. :-(
Cheers,
Bruce
- Badges
- Science topic