I have an ordinal dependent/response variable with 9 levels. I have 26 independent/predictor variables that include nominal, ordinal, interval, and ratio data types. Should I use stepwise ordinal logistic regression?

  • 1
    $\begingroup$ Welcome to the site! Please check our Help...Tour and have a look at previous threads on this site. If that doesn't answer your questions, please come back here and say more. Also, the question of which regression to use (if any) is different from that of whether to use a stepwise approach to variable selection. Both have been addressed extensively on this site. Cheers ~ $\endgroup$
    – rolando2
    Mar 28 '17 at 14:36
  • 2
    $\begingroup$ The answer is almost certainly no. Youll find, if you search our site, that these selection methods are always reccomended against when they come up. To discuss exactly what you should do, we would need a more detailed description of your analysis and its goals. $\endgroup$ Mar 28 '17 at 15:22