1
$\begingroup$

After much labor, I've yet to come up with the answer to:

Is it valid to run a ordered logistic regression with a two-level categorical dependent variable?

I'm running a logistic regression on a categorical dependent survey response variable ("Most People Can Be Trusted," vs. "You Cannot Trust Most People), against a variety of categorical predictors (age with decile factor levels, household income strata, etc.).

$\endgroup$
1
$\begingroup$

That is a very common usage of logistic regression. Assign one alternative to 0 the other to 1 and compute a logistic regression (the simple two-case, not the complicated "ordered" for more then two alternatives). Usually this is interpreted as modelling the chance of one or the other alternative.

Every two-alternative categorial variable can be viewed as ordered. For example you could consider the two states man and women as being ordered by being increasingly feminine or decreasing male-ish. Things only become difficult if there are more than two alternatives.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Would the end result of doing a OLR AND a LR with the standard generalized linear model be fairly similar, then? $\endgroup$ – MattCrow Feb 18 '17 at 22:49
  • $\begingroup$ To my undestanding, LR and OLP with just two levels are the same. I don't know, whether the implementation in whatever software you use is the same. The more important step towards answering your question is, that every two-alternative categorial can be treated like ordinal. $\endgroup$ – Bernhard Feb 18 '17 at 22:51

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.