I have outcome/dependent variable = ordinal, a scale from 0 to 10, about the degree of dyspnea.
Two groups = intervention and control.
Repeated measurements (4 time points).
I heard ANOVA is old school. Should I use a generalized mixed model?
Yes. See the following page:
https://www.theanalysisfactor.com/six-differences-between-repeated-measures-anova-and-linear-mixed-models/
Ordinal logistic regression mixed model is probably a good idea. See the attached for details. Best wishes David Booth
Hi Michael, the statistical test of choice would depend on the nature and number of independent variables. For example, in your case with an ordinal outcome variable, if you have just one independent variable with 2 or more levels (dependent/matched groups) then the Friedman test will be appropriate. It is a non-parametric test and similar to the parametric repeated measures ANOVA.
Hi Michael, the statistical test you should consider during analysis I think is GEE (Generalized Estimating equations) since can deal with correlation between dependent data, while controlling the group correlation and other factors. For more, you read this Article Analysis of ordinal repeated measures data using generalized...
and
https://www.publichealth.columbia.edu/research/population-health-methods/repeated-measures-analysis
- Badges
- Science topic
- Similar topics
- Biological Science
- Anatomy
- Tissue