non- dichotomous variable in logistic regression

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

scifi
Posts: 3
Joined: Mon Jan 28, 2013 10:01 am

non- dichotomous variable in logistic regression

Postby scifi » Mon Jan 28, 2013 10:19 am

Hi all,

I have used SPSS before but this time I am really stuck in sth that must be really simple:

I am asked to do a logistic regression . I have one categorical variable "town", "0" and "1" , and I can also define dummy variables for "agegroup" that has several categories.
The problem is that my dependent variable is not dichotomous.

THe dependent variable " is the answer to the question whether a person has skin cancer", and I am also given Y, the number of cases with the disease , and n, the total number of the population size for each subgroup.

So, I want to link the probability p= a person has skin cancer, to this logistic regression model.

How do I do that? I can denote p=Y/n, and the logistic regression is ln(p/1-p)= (coefficients*categorical variables),but how do I make SPSS understand that?

In other words, what is my dependent variable here??

Thank you sooooo much for any help.
scifi
Posts: 3
Joined: Mon Jan 28, 2013 10:01 am

Re: non- dichotomous variable in logistic regression

Postby scifi » Sat Feb 02, 2013 1:25 am

Ok, I figured it out, so I am posting here the answer for other forum members to see:

Instead of going to ANALYSE -> Regression -> Binary Logistic (where you really don;t understand what you are doing wrong and you bang your head against the wall ) , you go to
ANALYSE -> GENERALIZED LINEAR MODELS -> GENERALIZED LINEAR MODELS (twice) and in the window that appears under "Events/ Response data" you choose "binary logistic"

Then at the next tab , you just have to define your dependent variable and the "trial variable", this is the number of events (here skin cancer cases) :wink:

Who is online

Users browsing this forum: No registered users and 1 guest

cron