Since the median is often a better central descriptor of skewed distributions, I'd like to know if there is an equivalent method to a linear model for explaining variability of one variable through a qualitative variable, that uses the median as coefficient for categories instead of the mean ?

(Kind of like the Kruskal-Wallis test is the equivalent of the ANOVA for the median)

  • 1
    $\begingroup$ You mean like quantile regression? $\endgroup$ – Repmat Jun 19 '18 at 11:43
  • $\begingroup$ Oh. I was not introduced this entire field of regression in my studies. This seems to fit my description ! Thanks for pointing it out, I'll see what I can do with it. I'm surprised I didn't find (or maybe notice) it when I looked for stuff like "median linear model" and such $\endgroup$ – RoB Jun 19 '18 at 12:28
  • $\begingroup$ Well my number 1 google result for "median linear model" is quantile regression on wikipedia. The method also goes by the name "least absolute deviation" $\endgroup$ – Repmat Jun 19 '18 at 18:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.