Hi there,
I am I am working with data where the outcome is continuous and we have several predictor variables. My first model was using multiple linear regression but the residuals are very clearly not normally distributed.
[ATTACH]temp_23388_1626209169287_644[/ATTACH]

I considered log transforming my dependent variable, but came across these articles recommending using Poisson regression instead: https://www.stata.com/stata-news/news34-2/spotlight/ and https://blog.stata.com/2011/08/22/us...tell-a-friend/

Now, my data are such that they naturally have clusters (players on teams), so one question I have is if I should use the cluster option or the vce(robust) option. These two methods significantly change the confidence intervals of one of the coefficients in the model [in vce(robust) one of the predictors is very significant, but using cluster(team) that same predictor is not].

Lastly, I am not clear on how to interpret these coefficients- do I back transform each coefficient using the margins command?

Any help on this is greatly appreciated.
Thanks