rmcelreath / rethinking

Statistical Rethinking course and book package
2.17k stars 606 forks source link

Can we have an imputation example where the outcome has missing values? #379

Open KonstantinosChristopoulos opened 1 year ago

KonstantinosChristopoulos commented 1 year ago

Greetings,

How does one assign a distribution to a variable that has been already assigned one---since it is the outcome we model. All the examples in the book concern NAs in predictors, except for the dog example which has no imputations. I would love for example to see the d) scenario were we know the (MNAR) mechanism of NA. Could one impute random homeworks when H<5 from a distribution that produces values <5 to fill in the NAs before each iteration and use the full vector in the model?

Thank you!