Lab 4
General Instructions
Your response should include a Quarto file (.qmd) and an HTML document that is the result of applying your Quarto file to the data we’ve provided. I haven’t provided a template for this Lab.
Data
Lab 4 returns to the dig1.csv data file we used in Lab 1. The data remain available on the 500-data web page Refer to Lab 1 for more details on the data set and supplementary files about the DIG study.
The idea here is to develop an analysis of the data in the DIG teaching data set. Choose a population (based on the available DIG data or an appropriate subset), outcome, a binary indicator of treatment/exposure group and a set of between 10 and 30 covariates, then produce a Quarto and HTML file combination which addresses Tasks 1-5 below.
Task 1.
Build and display an appropriate Table 1 comparing the treatment groups on the covariates of interest.
Task 2.
Build and describe an unadjusted analysis of the impact of the treatment on the outcome. This should yield both a point estimate and uncertainty interval.
Task 3.
Build a complete analysis using 1:1 matching including a balance assessment pre- and post-matching, and an appropriate matched-set estimate and uncertainty interval for the causal effect of treatment on outcome, in the population you have defined, accompanied by a sensitivity analysis if appropriate, or a smart stability analysis if a sensitivity analysis isn’t appropriate.
Task 4.
Build a complete analysis using propensity weighting (and regression adjustment, if you like), including a balance assessment pre- and post-matching, and an appropriate propensity-weighted estimate and uncertainty interval for the causal effect of treatment on outcome, in the population you have defined.
Task 5.
Build and describe (in complete English sentences) a comparison of the results obtained from Tasks 2, 3 and 4. Describe any concerns you have about the relative merits of your various causal effect estimates.