Tests to justify use of quantile regression

[Karim Syed]

Hi all.
Can you suggest any formal statistical test/analysis in order to justify the application of the quantile regression model?
Something like the Hausman test to justify the choice between random and fixed effects model.
Regards

[Joao Santos Silva]

Dear Karim Syed,

You do not have to choose between quantile regression and other forms of regression; quantile regression just gives you different (more) information about the problem you are considering. So, there is no need to use a test to justify its use and you may use it to complement the results obtained with other methods.

Best wishes,

Joao

[Karim Syed]

Thanks, Joao for the reply. But given my situation, what particular statistical characteristics of my data should I consider in applying quantile regression? And, how to detect those characteristics? Should I use normality tests like Shapiro Wilk test to find if the distribution is normal or not and then decide to use quantile regression in case my distribution is not normal? Kindly comment.

[ericmelse]

Dear Karim,

Possibly, this paper offers good advice to compare and examine median values between group categories, using quantile regression:

The Stata Journal 2012 12 2 162-190 RM Conroy What hypotheses do nonparametric two-group tests actually test?

Best,
Eric

[Joao Santos Silva]

Dear Karim Syed,

If you have heteroskedasticity, then you will gain something by also doing quantile regression. I reiterate that quantile regression should not be seen as a replacement or alternative form of regression, but one of the tools that can be used.

Best wishes,

Joao