Binary Logistic Regression - Split File/Reference Category

[analyze28]

Hi all,

I'm hoping someone might have a clue as to what is happening here.  To explain - I am running a Logistic Regression analysis using a split file.  I am using Binary Logistic as my DV is categorical.

Ok, so the file is split by imputation and by age group as I am interested in whether differences occur between the age groups.  Running the analysis by age group is fine (aside from an error on the 20's group stating that "due to redundancies, degrees of freedom have been reduced."  I've checked multicollinearity and it's not that.  Removing some variables that are low in number seems to resolve this).

Now when I alter the reference category for my variable Alcohol Frequency to be medium consumption, for some reason this comes up fine for one age group.  For the other two age group, however, the reference category is light consumption.  Now a reminder - I am running this as a split file so the same syntax applies to all three.  The other variables are fine and remain consistent across the three age groups.  It is just this variable that changes.  I ran it with light, medium and heavy as the reference group to see if it still occurred, and it does.  The youngest runs with the specified reference group but the other two do not.  I can't work out why this is happening.  Any suggestions?

[Kelly Vander Ley]

I will be out of the office on August 2nd. If you need immediate assistance please call the main office number 503/223-8248 or 800/788-1887 and an office assistant will ensure that I get the message. 

 

Kelly

 

 

[lori.andersen]

In a logistic regression SPSS picks the categorical IV value with the largest number of cases as the reference. However, you can change this by using the options in SPSS. You can tell SPSS which category to use as the reference.

On Wed, Aug 1, 2012 at 11:30 PM, analyze28 [via SPSSX Discussion] <[hidden email]> wrote:

Hi all,

I'm hoping someone might have a clue as to what is happening here.  To explain - I am running a Logistic Regression analysis using a split file.  I am using Binary Logistic as my DV is categorical.

Ok, so the file is split by imputation and by age group as I am interested in whether differences occur between the age groups.  Running the analysis by age group is fine (aside from an error on the 20's group stating that "due to redundancies, degrees of freedom have been reduced."  I've checked multicollinearity and it's not that.  Removing some variables that are low in number seems to resolve this).

Now when I alter the reference category for my variable Alcohol Frequency to be medium consumption, for some reason this comes up fine for one age group.  For the other two age group, however, the reference category is light consumption.  Now a reminder - I am running this as a split file so the same syntax applies to all three.  The other variables are fine and remain consistent across the three age groups.  It is just this variable that changes.  I ran it with light, medium and heavy as the reference group to see if it still occurred, and it does.  The youngest runs with the specified reference group but the other two do not.  I can't work out why this is happening.  Any suggestions?

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Binary-Logistic-Regression-Split-File-Reference-Category-tp5714559.html

To start a new topic under SPSSX Discussion, email [hidden email]
To unsubscribe from SPSSX Discussion, click here.
NAML

--
Lori Andersen
Ph.D. student, Educational Policy, Planning & Leadership
College of William & Mary
Williamsburg, VA

[Maguin, Eugene]

This statement is not correct. Although the logistic regression documentation for 19 says NOTHING about the reference category for the DV, my understanding is that the lowest valued category is the customary reference category. There is no option to change/define the reference category for the DV. For IVs, yes, there is an option to define the reference category. Also the algorithms documentation does not define the DV reference category.

 

The DV reference category can be user-defined for Nomreg and Genlin and Genlinmixed.  

 

Gene Maguin

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of lori.andersen
Sent: Thursday, August 02, 2012 2:16 AM
To: [hidden email]
Subject: Re: Binary Logistic Regression - Split File/Reference Category

 

In a logistic regression SPSS picks the categorical IV value with the largest number of cases as the reference. However, you can change this by using the options in SPSS. You can tell SPSS which category to use as the reference.

On Wed, Aug 1, 2012 at 11:30 PM, analyze28 [via SPSSX Discussion] <[hidden email]> wrote:

Hi all,

I'm hoping someone might have a clue as to what is happening here.  To explain - I am running a Logistic Regression analysis using a split file.  I am using Binary Logistic as my DV is categorical.

Ok, so the file is split by imputation and by age group as I am interested in whether differences occur between the age groups.  Running the analysis by age group is fine (aside from an error on the 20's group stating that "due to redundancies, degrees of freedom have been reduced."  I've checked multicollinearity and it's not that.  Removing some variables that are low in number seems to resolve this).

Now when I alter the reference category for my variable Alcohol Frequency to be medium consumption, for some reason this comes up fine for one age group.  For the other two age group, however, the reference category is light consumption.  Now a reminder - I am running this as a split file so the same syntax applies to all three.  The other variables are fine and remain consistent across the three age groups.  It is just this variable that changes.  I ran it with light, medium and heavy as the reference group to see if it still occurred, and it does.  The youngest runs with the specified reference group but the other two do not.  I can't work out why this is happening.  Any suggestions?

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Binary-Logistic-Regression-Split-File-Reference-Category-tp5714559.html

To start a new topic under SPSSX Discussion, email [hidden email]
To unsubscribe from SPSSX Discussion, click here.
NAML

--
Lori Andersen
Ph.D. student, Educational Policy, Planning & Leadership
College of William & Mary
Williamsburg, VA

---

View this message in context: Re: Binary Logistic Regression - Split File/Reference Category
Sent from the SPSSX Discussion mailing list archive at Nabble.com.