LOGIT - Logit Transformed Responses CALCULATION

[ruegdeg]

Dear all,

i perform LOGIT regression in PASW software and Im wondering about the results. This table shows the probabilities that were calculated from inputs. In this table  are related estimates. This picture shows the Logit Transformed Responses graph. But i have no idea how to calculate values showed in this picture and why there are less points like probability value. I really appreciate any help.
Kind regards,
R.

[MaxJasper]

It appears that you only have a single scale variable and trying to fit a logit model?!!! Is there any categorical var? any freq (weight) var? Can you specify the data and then required stat you need?

 

Max.

 

From: ruegdeg [via SPSSX Discussion] [mailto:[hidden email]]
Sent: Friday, November 11, 2011 3:31
To: MaxJasper
Subject: LOGIT - Logit Transformed Responses CALCULATION

 

Dear all,

i perform LOGIT regression in PASW software and Im wondering about the results. This table shows the probabilities that were calculated from inputs. In this table  are related estimates. This picture shows the Logit Transformed Responses graph. But i have no idea how to calculate values showed in this picture and why there are less points like probability value. I really appreciate any help.
Kind regards,
R.

---

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

http://spssx-discussion.1045642.n5.nabble.com/LOGIT-Logit-Transformed-Responses-CALCULATION-tp4983976p4983976.html

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

[Bruce Weaver]

Max, you seem to be suggesting it is incorrect, or at least inappropriate to have a logit model with only one continuous (or scaled) explanatory variable?  Why is that?  I don't see a problem, provided the relationship is linear in the logit scale.  

Cheers,
Bruce

MaxJasper wrote

It appears that you only have a single scale variable and trying to fit a
logit model?!!! Is there any categorical var? any freq (weight) var? Can you
specify the data and then required stat you need?

 

Max.

 

From: ruegdeg [via SPSSX Discussion]
[mailto:[hidden email]]
Sent: Friday, November 11, 2011 3:31
To: MaxJasper
Subject: LOGIT - Logit Transformed Responses CALCULATION

 

Dear all,

i perform LOGIT regression in PASW software and Im wondering about the
results. This table  <http://www.scandinavia.sk/data/spss/table1.jpg> shows
the probabilities that were calculated from inputs. In this table
<http://www.scandinavia.sk/data/spss/table2.jpg>  are related estimates.
This picture  <http://www.scandinavia.sk/data/spss/picture1.jpg> shows the
Logit Transformed Responses graph. But i have no idea how to calculate
values showed in this picture and why there are less points like probability
value. I really appreciate any help.
Kind regards,
R.



  _____  

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

http://spssx-discussion.1045642.n5.nabble.com/LOGIT-Logit-Transformed-Respon
ses-CALCULATION-tp4983976p4983976.html

To start a new topic under SPSSX Discussion, email
[hidden email] 
To unsubscribe from SPSSX Discussion, click here
<http://spssx-discussion.1045642.n5.nabble.com/template/NamlServlet.jtp?macr
o=unsubscribe_by_code&node=1068821&code=bWF4amFzcGVyQHNoYXcuY2F8MTA2ODgyMXwz
MTg4NTI5ODI=> .
See how NAML generates this email
<http://spssx-discussion.1045642.n5.nabble.com/template/NamlServlet.jtp?macr
o=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namesp
aces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web
.template.InstantMailNamespace&breadcrumbs=instant+emails%21nabble%3Aemail.n
aml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail
.naml>

[MaxJasper]

I was surprised that with only one scale var the fit should be fairly straightforward…but the var00003 vs logit does not look so linear and that should indicate perhaps a 2^nd degree or another transformation be used or a discontinuous function.

 

This is a reverse problem: from solution we should guess data and analysis needed!

 

Max.

 

From: Bruce Weaver [via SPSSX Discussion] [mailto:[hidden email]]
Sent: Friday, November 11, 2011 12:38
To: MaxJasper
Subject: RE: LOGIT - Logit Transformed Responses CALCULATION

 

Max, you seem to be suggesting it is incorrect, or at least inappropriate to have a logit model with only one continuous (or scaled) explanatory variable?  Why is that?  I don't see a problem, provided the relationship is linear in the logit scale.  

Cheers,
Bruce

MaxJasper wrote:

It appears that you only have a single scale variable and trying to fit a
logit model?!!! Is there any categorical var? any freq (weight) var? Can you
specify the data and then required stat you need?

 

Max.

 

From: ruegdeg [via SPSSX Discussion]
[mailto:[hidden email]]
Sent: Friday, November 11, 2011 3:31
To: MaxJasper
Subject: LOGIT - Logit Transformed Responses CALCULATION

 

Dear all,

i perform LOGIT regression in PASW software and Im wondering about the
results. This table  <http://www.scandinavia.sk/data/spss/table1.jpg> shows
the probabilities that were calculated from inputs. In this table
<http://www.scandinavia.sk/data/spss/table2.jpg>  are related estimates.
This picture  <http://www.scandinavia.sk/data/spss/picture1.jpg> shows the
Logit Transformed Responses graph. But i have no idea how to calculate
values showed in this picture and why there are less points like probability
value. I really appreciate any help.
Kind regards,
R.



  _____  

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

http://spssx-discussion.1045642.n5.nabble.com/LOGIT-Logit-Transformed-Respon
ses-CALCULATION-tp4983976p4983976.html

To start a new topic under SPSSX Discussion, email
[hidden email] 
To unsubscribe from SPSSX Discussion, click here
<http://spssx-discussion.1045642.n5.nabble.com/template/NamlServlet.jtp?macr
o=unsubscribe_by_code&node=1068821&code=bWF4amFzcGVyQHNoYXcuY2F8MTA2ODgyMXwz
MTg4NTI5ODI=> .
See how NAML generates this email
<http://spssx-discussion.1045642.n5.nabble.com/template/NamlServlet.jtp?macr
o=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namesp
aces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web
.template.InstantMailNamespace&breadcrumbs=instant+emails%21nabble%3Aemail.n
aml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail
.naml>

--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

 

---

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

http://spssx-discussion.1045642.n5.nabble.com/LOGIT-Logit-Transformed-Responses-CALCULATION-tp4983976p4985345.html

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

[ruegdeg]

Dear Max, Bruce, all,

thanks for all replies.

I didnt, post the data because i throught that its not important. I know that some 2,3 order or other calculation will be necessary, but this is exactly the point for detection.
But I was wondering about the calculation. For example in the picture the value 0.5 of VAR0003 has an logit value  something below -3.5. But how is that calculated?? If i assume that logit is equal to ln(p/(1-p)) then for the value 0.5 based on constant and beta value the logit should be -2.48+(-0.388)*0.5 and that is -2.67 and not -3.5. any suggestions?