Dummy variable puzzle

[Firangiz Aghayeva]

Dear Statalist Users,

I create a dummy variable that is equal to 1 if a fund manager had a prior fund with strong performance (above benchmark value) and 0 if otherwise. The question is related to the 0 values of this dummy variable. Should I also define funds with no performance data as 0, or should they be missing?

Looking forward to hearing from you.

Kind regards,
Firangiz Aghayeva

[Clyde Schechter]

In general any variable for which you have no data should be entered as a missing value in your data set. Only if there is some special circumstance in how the data were originally created, from which it would necessarily follow that the absence of performance data implies that the performance was not strong, would you be justified in coding it as 0. Off hand I can't think of any circumstance like that which would apply to your situation, but as I do not work in finance, I would defer to the judgment of somebody who does about that. But the key point is that if you don't actually know the performance was not strong, then you can't justify coding it as 0.

[Richard Williams]

Based on the info given, I agree with Clyde. Depending on how badly you want those missing cases you can try multiple imputation or FIML or some other method. See

https://www3.nd.edu/~rwilliam/xsoc73994/MD01.pdf

https://www3.nd.edu/~rwilliam/xsoc73994/MD02.pdf

[Firangiz Aghayeva]

Dear Clyde and Richard,

Noted. Many thanks for your prompt answers. I highly appreciate it.

Kind regards,
Firangiz