[x] Read the data into R and create a new Q1 variable that is the response to Q1B if Q1A is missing or Q1A if Q1B is missing. Also create a new variable to indicate whether the individual was randomized to group group A or B (depending on which has missing values). Examine the data, check whether there are any strange values, clean it up if necessary (i.e. if data that should be categorical reads in as numeric, update this).
[x] Create visualizations of the responses to Q1 and Q2 by treatment
[x] Fit a proportional odds model predicting Q2 from treatment. I think we should at least adjust for whether they currently sanitize their phone, you are welcome to include other demographics (especially if they appear imbalanced in part 2). I think it probably makes sense to look at this separately by gender as well, since we know health decisions likely differ by gender, so I'd fit one full model on the whole data set, one on just females, and one on just males. There is info on fitting proportional odds models here: https://bookdown.org/Rmadillo/likert/is-there-a-significant-difference.html. To do this, I like to use the orm function from the rms package.
[x] Begin drafting the intro & methods sections of your paper