Constant Variable in Logistic Regression

[Iwase Tamiko]

Hi all,
 
I have a binary DV (Employment Status: Yes or No) and I want to run a regression.
I have been advised that in this case, the regression should be binary logistic regression.
 
After closer inspection of the data, I noticed that for all respondents who selected "No" (i.e.,
unemployed) their ENTIRE data is "missing". It's like having a TOTALLY BLANK survey
for unemployed people.
 
After running the logistic regression, I have the message "cannot be computed because
one of the variables is constant".
 
1) Is the missing data the reason why I'm getting this message?
2) Would it make sense to delete all the "No" because they are all missing anyway?
3) Would it be safe to assume that NO analysis could be done on this very important variable??
 
Thanks in advance!
 
Tamiko
 

[Jason Ferris-3]

Hi Tamiko,

Depending on your IVs of interest I would assume, as a design of the questionnaire, they have been filtered out.  That is if you say no to currently employed you are not asked do you work full time or part time etc.

 

If this is the case (i.e., the question addressing full-time status) is only completed if the question address employment is yes than you will end up with the message you are receiving.

 

So yes, the missing data is the reason you may be getting this message – but the data is missing by questionnaire design.

You should not delete the no’s if there are other questions that you ask where employment type is not a filtering question.  I.e., how many beers did you have in the last week? Not being employed will not restrict you from answering this question.

You have no reason to try and run analysis on my example (if it is similar to what you have) because only those who answered employed were asked additional questions.

 

HTH

 

Jase

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Iwase Tamiko
Sent: Friday, 31 July 2009 4:02 PM
To: [hidden email]
Subject: Constant Variable in Logistic Regression

 

Hi all,
 
I have a binary DV (Employment Status: Yes or No) and I want to run a regression.
I have been advised that in this case, the regression should be binary logistic regression.
 
After closer inspection of the data, I noticed that for all respondents who selected "No" (i.e.,
unemployed) their ENTIRE data is "missing". It's like having a TOTALLY BLANK survey
for unemployed people.
 
After running the logistic regression, I have the message "cannot be computed because
one of the variables is constant".
 
1) Is the missing data the reason why I'm getting this message?
2) Would it make sense to delete all the "No" because they are all missing anyway?
3) Would it be safe to assume that NO analysis could be done on this very important variable??
 
Thanks in advance!
 
Tamiko
 

------------------------------------------
DISCLAIMER: This message (including any attachments) is intended solely for the addressee(s) named and may contain confidential or privileged information. 
If you are not the intended recipient, please delete it and notify the sender. 
Views expressed in this message are those of the individual sender,and are not necessarily the views of the Turning Point Alcohol and Drug Centre Inc (ABN: 17 302 055 629).  

Turning Point Alcohol and Drug Centre Inc

Although this message and any attachments have been scanned for viruses by 'Trend Micro InterScan' at the time of sending, 
you are advised to rescan on receipt.

The whole or parts of this email may be subject to copyright of Turning Point Alcohol and Drug Centre Inc (ABN: 17 302 055 629), and/or third parties. 
You can only re-transmit, distribute or use the material if you are authorised to do so.

Please consider the environment before printing this email or attachments.

[Hector Maletta]

Of course it is because of that. A binary dependent variable means that some subjects have a Yes and some a No. Otherwise it is not a “variable” but a “constant” (it cannot be a “constant variable”, you know).

Now, are those IV simply “missing”, or they can only be defined for employed people? What I mean is: if you are using sex or age as predictors, and those variables are blank, you get a missing value (but there is a value, only you don’t know it); but if the “predictor” is something you can only know about employed people (like “years on the job” or “salary”) the information is not “missing”, it is simply not applicable.

Or perhaps the information is applicable and should be there but the survey just didn’t ask those questions from unemployed people, for whatever reason. In any case, you cannot perform the analysis.

 

Hector

 

---

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Iwase Tamiko
Sent: 31 July 2009 01:02
To: [hidden email]
Subject: Constant Variable in Logistic Regression

 

Hi all,
 
I have a binary DV (Employment Status: Yes or No) and I want to run a regression.
I have been advised that in this case, the regression should be binary logistic regression.
 
After closer inspection of the data, I noticed that for all respondents who selected "No" (i.e.,
unemployed) their ENTIRE data is "missing". It's like having a TOTALLY BLANK survey
for unemployed people.
 
After running the logistic regression, I have the message "cannot be computed because
one of the variables is constant".
 
1) Is the missing data the reason why I'm getting this message?
2) Would it make sense to delete all the "No" because they are all missing anyway?
3) Would it be safe to assume that NO analysis could be done on this very important variable??
 
Thanks in advance!
 
Tamiko
 

No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 8.5.392 / Virus Database: 270.13.36/2272 - Release Date: 07/30/09 18:09:00

[Jason Ferris-3]

Hi Iwase,

Depending on your IVs of interest I would assume, as a design of the questionnaire, they have been filtered out.  That is if you say no to currently employed you are not asked do you work full time or part time etc.

 

If this is the case (i.e., the question addressing full-time status) is only completed if the question address employment is yes than you will end up with the message you are receiving.

 

So yes, the missing data is the reason you may be getting this message – but the data is missing by questionnaire design.

You should not delete the no’s if there are other questions that you ask where employment type is not a filtering question.  I.e., how many beers did you have in the last week? Not being employed will not restrict you from answering this question.

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Iwase Tamiko
Sent: Friday, 31 July 2009 4:02 PM
To: [hidden email]
Subject: Constant Variable in Logistic Regression

 

Hi all,
 
I have a binary DV (Employment Status: Yes or No) and I want to run a regression.
I have been advised that in this case, the regression should be binary logistic regression.
 
After closer inspection of the data, I noticed that for all respondents who selected "No" (i.e.,
unemployed) their ENTIRE data is "missing". It's like having a TOTALLY BLANK survey
for unemployed people.
 
After running the logistic regression, I have the message "cannot be computed because
one of the variables is constant".
 
1) Is the missing data the reason why I'm getting this message?
2) Would it make sense to delete all the "No" because they are all missing anyway?
3) Would it be safe to assume that NO analysis could be done on this very important variable??
 
Thanks in advance!
 
Tamiko
 

------------------------------------------
DISCLAIMER: This message (including any attachments) is intended solely for the addressee(s) named and may contain confidential or privileged information. 
If you are not the intended recipient, please delete it and notify the sender. 
Views expressed in this message are those of the individual sender,and are not necessarily the views of the Turning Point Alcohol and Drug Centre Inc (ABN: 17 302 055 629).  

Turning Point Alcohol and Drug Centre Inc

Although this message and any attachments have been scanned for viruses by 'Trend Micro InterScan' at the time of sending, 
you are advised to rescan on receipt.

The whole or parts of this email may be subject to copyright of Turning Point Alcohol and Drug Centre Inc (ABN: 17 302 055 629), and/or third parties. 
You can only re-transmit, distribute or use the material if you are authorised to do so.

Please consider the environment before printing this email or attachments.