About this code
This code is used to estimate and compare change in the log-odds of success in a binary longitudinal outcome for the adaptive interventions embedded in a SMART.
How can a behavioral scientist use this code?
Behavioral intervention scientists can use this code to make inferences about the relative causal effects of one adaptive intervention versus another using data from a SMART with a longitudinal binary outcome.
What method does this code implement?
This code implements a weighted least squares regression approach that is similar to longitudinal regression analyses using GEE with a logit link. In a SMART, the weights (which are potentially impacted by treatment) are known, by design.
Related References
Nahum-Shani, I., Dziak, J. J., Venera, H., Spring, B., & Dempsey W. (2023). Design of Experiments with Sequential Randomizations at Multiple Time Scales: The Hybrid Experimental Design. Behavior Research Methods, doi:10.3758/s13428-023-02119-z.