Dear Statalist users,

I am looking for a way to find the number of observations used to estimate a certain coefficient in a regression. If the data, for example, contain 1000 observations with no missing values, might it be the case that not all of them contribute to the estimation of each coefficient? Say that some observations may not be used in estimating the coefficient of the treatment dummy in a model, while they may be contributing to the estimation of other coefficients in the model (e.g. the betas of stratum fixed effects). Is there a way for Stata to differentiate between these observations?

I greatly appreciate your help. Thank you!