lauken13/mrpkit

Errors and warnings for missing data

Closed this issue · 3 comments

  • In SurveyMap$fit: warning if missingness in outcome variable, then let model fitting drop rows with missingness

  • In SurveyMap$validate: error if any missingness in the mapped variables (ignoring any other variables)

Assigning myself to do this

Quick question: do we want an error if NAs in the adjustment variables in sample survey data, the population survey data, or both? I'm thinking both but wanted to double check.

Also, right now there's nothing preventing the user from never calling validate(). So we need to force the call of validate() in order to make sure this error will happen.

I ended up combining this with validate improvements in PR #90.