About this example code

This example code can be used to estimate (a) the average main (and moderated) effect of digital components (i.e., a mobile-based prompt) on a binary proximal outcome; and (c) the main (and moderated) effects of human-delivered components (e.g., coaching sessions) on a distal outcome using simulated data from a hybrid SMART-MRT in which only non-responders were micro-randomized. This code accompanies the following manuscript in Behavior Research Methods: Design of Experiments with Sequential Randomizations on Multiple Timescales: The Hybrid Experimental Design (Nahum-Shani et al. 2023)

How can a behavioral scientist use this code?

Behavioral intervention scientists can use this code as a starting point for analyzing data from their own hybrid SMART-MRT.

What method does this code implement?

This code implements marginal causal excursion effect estimation for the proximal outcome analysis (see Qian et al., 2021) and linear regression using GEE only among non-responders for the distal outcome analysis.

Related References

Nahum-Shani, I., Dziak, J.J., Venera, H. et al. Design of experiments with sequential randomizations on multiple timescales: the hybrid experimental design. Behav Res (2023). https://doi.org/10.3758/s13428-023-02119-z

Tianchen Qian, Hyesun Yoo, Predrag Klasnja, Daniel Almirall, Susan A Murphy, Estimating time-varying causal excursion effects in mobile health with binary outcomes, Biometrika, Volume 108, Issue 3, September 2021, Pages 507–527, https://doi.org/10.1093/biomet/asaa070

Join the d3center Mailing List

Keep up to date with the latest news, events, software releases, learning modules, and resources from the d3center.