Announcement

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

  • Missing observations

    A questionnaire that I am researching has 188 observations. But when I perform a linear regression, observations are 153. Does anyone know why this is?

  • #2
    Presumably one or more of your variables are missing for 35 observations. Linear regressions use listwise deletion, so if a single variable is missing for a given observation the entire observation will be excluded.

    Comment


    • #3
      Marleen:
      elaborating a bit on Ali's helpful assist, just type:
      Code:
      summarize
      and the culprits will come alive.
      Kind regards,
      Carlo
      (Stata 18.0 SE)

      Comment


      • #4
        Originally posted by Ali Atia View Post
        Presumably one or more of your variables are missing for 35 observations. Linear regressions use listwise deletion, so if a single variable is missing for a given observation the entire observation will be excluded.
        Thank you Ali! Very clear

        Comment


        • #5
          Originally posted by Carlo Lazzaro View Post
          Marleen:
          elaborating a bit on Ali's helpful assist, just type:
          Code:
          summarize
          and the culprits will come alive.
          Thank you Carlo! Very helpful

          Comment


          • #6
            To understand your data better, have a look at the -misstable- command or tag cases as valid and compare the counts of valid cases without and with listwise deletion of missing cases (the list of variables should contain all variables of your regression model, in the example price weight foreign rep78):
            Code:
            sysuse auto, clear
            
            replace weight = . in 60/65  // generate some missings
            replace price = . in 1/5     // generate other missings
            
            misstable pattern price weight foreign rep78, freq  // investigate pattern of missings
            
            mark valid                   // create variable to tag valid cases listwise
            markout valid price weight foreign rep78  // tag valid cases listwise into the variable "valid"
            
            tabstat price weight foreign rep78, s(count) c(s)  // without listwise deletion of missings
            tabstat price weight foreign rep78 if valid, s(count) c(s)  // with listwise deletion of missings

            Comment

            Working...
            X