Principal Stratification Tool

  1. Start → Calculating Bounds for Principal Causal Effects: Interactive Analysis Tool
  2. Design Requirements
  3. Identify the Principal Strata for your Study
  4. Use Substantive Assumptions to Restrict the Number of Strata of Interest for these Analyses
  5. Determining Proportions of Subgroups using Observed Quantities
  6. Solving for Proportions that are not Directly Observed
  7. Solving for Means under Treatment and Control
  8. Applying the Exclusion Restriction
  9. Solving for the Remaining Means
  10. Calculating the Bounds and Estimating Ranges of Impact
  11. Next Steps

In order to use this tool, your experimental dataset must contain the following items:

  1. An indicator of treatment or control status: Participants in your study must have been randomly assigned to a treatment or control condition
  2. A binary outcome of interest: You must be interested in looking at principal causal effects for a dichotomous outcome. While this framework can be used with continuous outcomes as well, this tool is only equipped to handle binary outcomes. You must have outcome data for members of participants in both of your experimental groups.
  3. Indicators of post-randomization behaviors for members of both experimental groups: In order to determine the membership in principal strata, you need to be able to observe the post randomization behavior you are interested in among both treatment and control assigned participants.
Continue