Getting around listwise deletion

[Sabrina Solanki]

Hi,

How can I run a regression without stata dropping missing variables. Basically, I want to get around the list wise deletion default. For example, in SAS, there is a skip missing command. Do we have something similar in stata?

Thanks!
Sabrina

[Joe Canner]

Sabrina,

How would you even do a regression that included observations where one or more of the variables in your model were missing? SAS does the same thing; I am not familiar with the "skip missing command", can you elaborate?

Regards,
Joe

[Sabrina Solanki]

Hi Joe,

Basically, I do not want to lose an observation because it is missing on one of my predictor variables.

Thanks.
Sabrina

[Clyde Schechter]

The most direct way to do that is to run the regression under -sem- with full-information maximum likelihood estimation (-method(mlmv)-), which will give unbiased results if you can credibly claim that the missing values in your data are missing at random.

I do not know what the SAS skip missing command is, so I cannot tell you if anything in Stata is comparable. Perhaps somebody who is a recent SAS user on this Forum will chime in on that. My understanding was that SAS, like Stata, deletes listwise when missing observations are encountered, and I was not aware that there was any way around it other than using structural equations modeling with full-information maximum likelihood estimation in SAS either.

(Of course, multiple imputation of missing observaitons is another alternative if the data are missing at random.)

[Sabrina Solanki]

Thank you!