xtqreg warning

[David Barker]

Running xtqreg on a sample of over 5,000 with 124 clusters and a lot of fixed effect variables I received this message:

WARNING: .57481275% of the fitted values of the scale function are not positive

Should I worry about this? What problems does this indicate?

[Joao Santos Silva]

Dear David Baker,

That is a small proportion of the observations, so I would not worry too much. The scale function should be positive, but it is natural that some of the estimated values are negative.

Best wishes,

Joao

[David Barker]

Thank you!