Hello everyone!!
I am entering the world of simultaneous quantile regressions, and a problem that I cannot get my head around are the values of tau: is there any good rule of thumb to decide them?
Reading on manuals, it sees to me that the answer is "it depends on your research question", many use 0.25 0.5 and 0.9 but in articles I've seen also 0.05 0.5 0.95 or 0.1 0.5 0.9 or 0.1 0.2 0.3 etc etc. I am interested in the extremes of the distribution. I do not have clear research question- or theory-related reasons to set say 0.1 0.5 0.9 instead of 0.05 0.5 0.95. Help?
Another somewhat related questions about sqreg: what is the command to get robust st errors?
Thank you so much!!!!
Andrea
I am entering the world of simultaneous quantile regressions, and a problem that I cannot get my head around are the values of tau: is there any good rule of thumb to decide them?
Reading on manuals, it sees to me that the answer is "it depends on your research question", many use 0.25 0.5 and 0.9 but in articles I've seen also 0.05 0.5 0.95 or 0.1 0.5 0.9 or 0.1 0.2 0.3 etc etc. I am interested in the extremes of the distribution. I do not have clear research question- or theory-related reasons to set say 0.1 0.5 0.9 instead of 0.05 0.5 0.95. Help?
Another somewhat related questions about sqreg: what is the command to get robust st errors?
Thank you so much!!!!
Andrea
