Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • How to deal with missings on partner information if the person has no partner?

    Dear all,

    I use panel / survey data and in my estimations I include variables like "age of partner" "employment status of partner" etc. How can I deal with the fact that these variables have missings if the person has no partner. I do not want these observations without partner to be dropped from the regressions.

    Should I recode those with missings due to a lack of partner to -1, for instance? I'm not sure if this would bias my estimates.

    I would be very grateful for any reply.

    Best
    Luisa

  • #2
    At most you have a different kind of missing. Recoding values to -1 is not at all the answer as ever thereafter you have to exclude such values to stop Stata taking -1 literally.

    Alternatively, you have some singleton groups and an indicator variable which is 1 for no partner and 0 otherwise might be helpful for some kinds of modelling.

    But necessarily for some analyses where partner properties are responses or predictors you can't include those observations. The question is why you would want to.
    Last edited by Nick Cox; 15 Jan 2019, 08:10.

    Comment

    Working...
    X