dfbeta cutoff point problem

[Kevin Cooper]

Hi everyone,

I need to calculate dfbetas of an independent variable for my work. I basically write "dfbeta" command after running the regression. It gives me dfbetas of the independent variable but not for all observation. I noticed that it does not give dfbetas of the observations if dfbetas >1 which is probably the cutoff point. But, I need to have dfbeta for all observation.

Could you please help me to increase the cutoff point or see dfbeta of all observation (maybe the way of manually calculating on STATA)?

Thank you very much in advance.
Kevin

[Richard Williams]

Welcome to Statalist.

Read the list FAQ, especially pt. #12, on asking Qs effectively. Showing your actual code and output using code tags give us more of a chance of answering your Q.

I am not aware of limitations like you are claiming, but we can tell better once we see exactly what you are doing.

See pp. 4-6 of the following for a discussion of dfbeta:

https://www3.nd.edu/~rwilliam/stats2/l24.pdf

When it does not give a dfbeta, is the case perhaps being excluded from the analysis because it has missing data?

[Kevin Cooper]

Hi Richard,

Thanks for your warm welcome and really sorry about the way I asked the question. I will be very careful for the next time.

However, as you said, the problem was due to exclusion of other variables. Many thanks for your help.

Best wishes.