Hi Brian
That should be fine. There are two variants:
(a) either treat the ordinal as a numeric predictor, which assumes a linear relationship between the categories in the Logit OR
(b) dummy code two of the three categories as 0/1 leaving one (the reference category) blank.
Suggest you run (b) first and compare with (a) as a more restricted form of the model. You can either use -2LL or BIC/AIC as a parsimony adjusted assessment.
Good luck!
Gary
_______________
Gary Bennett
Logit Research
Tel: +44 (0)1732 465662
Fax: +44 (0)1732 465662
Mob: +44 (0)7795 342575
http://www.logitresearch.com/
Logit Research Limited, Unit 3, 7 Blighs Walk, Sevenoaks, Kent. TN13 1DB. Company Reg No. 5086604. VAT Reg No. 826 5147 25
-----Original Message-----
From: A UK-based worldwide e-mail broadcast system mailing list [mailto:[log in to unmask]] On Behalf Of Dr Brian McMillan
Sent: 18 March 2012 15:31
To: [log in to unmask]
Subject: Logistic regression with ordinal and dichotomous predictors?
Hello all,
Is it possible to run a logistic regression with one ordinal predictor variable (1,2,3) and one dichotomous predictor variable (0,1)? The outcome is dichotomous (0,1).
Many thanks in advance,
Brian
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|