## Multivariate logistic regression models

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

student25
Posts: 2
Joined: Thu Aug 20, 2015 11:04 pm

### Multivariate logistic regression models

I am building a multivariate logistic regression model with 4 exposure variables, 3 confounders, and a binary outcome variable. (all my variables are categorical, but the exposure variables are polychotomous)
I am not a student of statistics and am doing a detailed analysis for the first time. I have done the univariate analysis already, and i read that to chose the multivariate model I need to do the likelihood ratio test. the spss output does not seem to provide the likelihood ratio values in the logistic regression output, but instead in the crosstabulations with the chi squared i was able to find it.
1. I was hoping someone could help me out with how exactly i could go about with building the model and comparing the Likelihood ratios?
2. how to produce the likelihood ratios? I have been reading theoretical math books, so i am not sure if i would have to do the comparing the ratios maually, or when they say, test the model step-wise without 1 variable at a time, does spss do that for you?
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

### Re: Multivariate logistic regression models

Maybe you need odds ratios instead of loglikelihood ratios? Loglikelihood ratios are used, for instance, to compare between models.
I think it would help you to look at some youtube videos explaining the method with an example.
student25
Posts: 2
Joined: Thu Aug 20, 2015 11:04 pm

### Re: Multivariate logistic regression models

Hey thanks for the reply. Yes I need to compare models with and without certain exposure variables which is why I need to use the likelihood ratio test.
I saw a few videos that show the blockwise entry method while doing binary regression. Is that the right way to go about it?

### Who is online

Users browsing this forum: No registered users and 2 guests