Logistic Regression With Continuous Variables in Spss

I'm not having any luck getting an image.

You could attach it here in the Community or send it straight to me at marcantr@us.ibm.com

------------------------------
Rick Marcantonio
Quality Assurance
IBM
------------------------------

Original Message:
Sent: Mon October 03, 2022 03:13 PM
From: Guillaume D
Subject: Logistic Binary Regression Limited to 15 Independent Variables

Hi John and Rick,

I ran a correlation. Here's a screenshot. I don't know if it helps

I can shared my database with you so you can run analyses and see. It's public data from Statistics Canada. These are constructed variables.

Thanks

------------------------------
Guillaume D

Original Message:
Sent: Mon October 03, 2022 12:02 PM
From: Jon Peck
Subject: Logistic Binary Regression Limited to 15 Independent Variables

And how many cases are there?

Original Message:
Sent: 10/3/2022 11:12:00 AM
From: Rick Marcantonio
Subject: RE: Logistic Binary Regression Limited to 15 Independent Variables

Run CORRELATIONS on those variables and see what they look like. If possible, publish the matrix. My guess is that these variables are too highly correlated (multi-collinear). There may also be missing data issues; not sure.

------------------------------
Rick Marcantonio
Quality Assurance
IBM

Original Message:
Sent: Mon October 03, 2022 11:08 AM
From: Guillaume D
Subject: Logistic Binary Regression Limited to 15 Independent Variables

Thanks Rick and John for your message.

I received no error message, except at the very beginning of the analysis, this message "Due to redundancies, degrees of freedom have been reduced for one or more variables". All 16 variables are shown in Bloc : 0, but one is missing on Block: 1.

I added 3 more dichotomous variables (total 19) at the end of the list again, they don't appear.

I added a 17th variable, this time continuous (Income) and it appears at the very bottom of the analysis but one of the 16 dichotomical variable is missing. With Jasp for instance, when I did this manipulation it says "the leading minor of order 17 is not positive definite" so no analysis if shown, but as look as I remove any dichotomical variable and keep the continuous one, the results are shown.

I listed the 16 original variables the other way around and, still, it is always the last of the list (16th) which does not appear in Block : 1, but is there in Block: 0)

So it does look like one variable is problematic, but the list ?

I use the syntax, but the tab and windows to run the regression but with the exact same problem.
LOGISTIC REGRESSION VARIABLES Lower_Class
/METHOD=ENTER
MWCM1 MwCM2 MwcM3 Mwcm4 MwCm5 MWCm6 MWcM7 MWcm8 fWCM1 fwCM2 fwcM3 fwcm4 fwCm5 fWcm6 fWCm7 fWcM8.

I have SPSS 28.0.0.0.

My data is bascially based on 4 dichotomical variables (Gender, Ethnicity, Political Status and Marital Status) from which I created 16 different variable based on the different composition (i.e. Man, Non-White, Citizen, Married, etc.) (dichotomical as well).

Thanks again for helping me with that.

------------------------------
Guillaume D

Original Message:
Sent: Mon October 03, 2022 09:54 AM
From: Jon Peck
Subject: Logistic Binary Regression Limited to 15 Independent Variables

I would guess that this is a problem with the data, especially since the same thing happens in JASP.

--

Original Message:
Sent: 10/3/2022 8:13:00 AM
From: Rick Marcantonio
Subject: RE: Logistic Binary Regression Limited to 15 Independent Variables

Hi. There is no such limitation to LOGISTIC REGRESSION; I just ran one with 30 variables and all of the variables were entered.

Did you get an error or warning message of any kind when you ran the analysis?

------------------------------
Rick Marcantonio
Quality Assurance
IBM

Original Message:
Sent: Sun October 02, 2022 11:51 PM
From: Guillaume D
Subject: Logistic Binary Regression Limited to 15 Independent Variables

Hello everyone,

For the first time I ran a binary regression using 16 variables and apparently SPSS only includes 15 in the analysis.
I changed the variables, added more and still, it remains limited to 15 independant variables.

Is the regression limited by default? If so, how can I modify it?

If it should take more, I am wondering what is the problem since I ran the same analysis using JASP (based on R) and I have the exact same problem.

Thanks for enlighting me.

Best,

------------------------------
Guillaume D
------------------------------

whickerandendee.blogspot.com

Source: https://community.ibm.com/community/user/datascience/discussion/logistic-binary-regression-limited-to-15-independent-variables