I have a rank-ordered logit model in which the various independent variables are pertinent for some cases but not others. For example, cases 1, 2, 3, & 4 might have variables A & B that are relevant whereas cases 5 & 6 might have variables B, C, & D that are relevant. My question is, prior to running the regression, do I need to code the data in any special way to indicate that only certain variables are applicable to certain cases, or will the regression extract the correct coefficients to use on the applicable cases. I appreciate any specific or conceptual guidance you may have on this.

Thank you!