https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772772#M37772 <P>Yes, use the UNEQUALSLOPES option. See <A href="http://support.sas.com/kb/22954" target="_self">this note</A>.&nbsp;</P> Thu, 07 Oct 2021 14:54:58 GMT StatDave 2021-10-07T14:54:58Z https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772770#M37771 <P>Hi, is there a way to do&nbsp;multilevel ordered logistic regression with unequal slopes with SAS? I want to model educational attainment (no high school, high school, postsecondary) with variables such as cohort, country cohort*country, religion, language and education of the mother. My understanding is that with a normal ordered logistic model, parameters for religion, language and education of the mother might be biased.&nbsp;</P> Thu, 07 Oct 2021 14:51:27 GMT https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772770#M37771 Demographer 2021-10-07T14:51:27Z https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772772#M37772 <P>Yes, use the UNEQUALSLOPES option. See <A href="http://support.sas.com/kb/22954" target="_self">this note</A>.&nbsp;</P> Thu, 07 Oct 2021 14:54:58 GMT https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772772#M37772 StatDave 2021-10-07T14:54:58Z https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772777#M37773 It is possible to do multilevel regression with proc logistic? Thu, 07 Oct 2021 14:58:04 GMT https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772777#M37773 Demographer 2021-10-07T14:58:04Z https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772781#M37774 <P>If by "multilevel" you mean a random effects model with levels rather than meaning just multiple response levels, then no - you would need to use GLIMMIX to fit a "multilevel" model. I'm not aware that it allows separate slopes on the logits.</P> Thu, 07 Oct 2021 15:01:59 GMT https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772781#M37774 StatDave 2021-10-07T15:01:59Z https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772783#M37775 <P>Yes, random effects models.</P> Thu, 07 Oct 2021 15:05:03 GMT https://communities.sas.com/t5/Statistical-Procedures/Multilevel-ordered-logistic-regression-with-unequal-slopes/m-p/772783#M37775 Demographer 2021-10-07T15:05:03Z