I have counts over time (yearly), but I can't do panel analysis because of a time-related sampling bias, i.e. my data appears trended but I think it is because of a sampling bias, not a real underlying trend. Hence, I've decided to take the historical average of the variables instead.

My yearly data is very skewed: a few very high counts vs many very low counts. It fits the assumptions for a negative binomial distribution, i.e. the variance is much greater than the mean and I've been recommended to apply a negative binomial regression.

However, once averaged over time, my data does not take the form of integers any more, although it still displays the same skewness. Can I still apply a negative binomial regression to the averaged data?

My independent variables are continuous variables, which I can average.

I find a partial answer but it doesn't clarify if averaging over time is neutral for negbin.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.