Logistic regression-help

[jeyni]

Hi all, I need your help.

I am estimating probability of firms default.
Have data annual data from 2002 - 2012 for each firm with 6 variables
Therefore most likely that observations are correlated. Is it still appropriate to use logistic model? should I do any other test before running a logistic regression?


Also, if I want to estimate default probability in lets say 3 years time, how can I do it?


I have little of information, therefore need some help.

Thanks

[jeyni]

Hi all, I need your help.

I am estimating probability of firms default.
Have data annual data from 2002 - 2012 for each firm with 6 variables
Therefore most likely that objectives are correlated. Is it still appropriate to use logistic model? should I do any other test before running a logistic regression?


Also, if I want to estimate default in lets say 3 years time, how can I do it?


I have little of information, therefore need some help.

Thanks

[Bruce Weaver]

I see you re-posted the same question with a different (and far less informative) subject line.  There was no need to do that--your original post did make it to the list.  But remember that this is a mailing list, not a chat room.  I.e., communication is asynchronous, and sometimes (especially on weekends), it may take a while for people to respond.  (And sometimes, people may not respond at all--nobody's getting paid, after all!)  ;-)

With regard to the first part of your question, take a look at doing logistic regression via GENLIN (logit link function, binomial error distribution) using generalized estimating equations (GEE) to account for the correlated nature of the repeated measures.  You can find some old threads on that topic in the archives of this list.  E.g.,

http://spssx-discussion.1045642.n5.nabble.com/GEE-with-Binary-Data-td1090448.html#a1090453

HTH.

jeyni wrote

Hi all, I need your help.

I am estimating probability of firms default.
Have data annual data from 2002 - 2012 for each firm with 6 variables
Therefore most likely that observations are correlated. Is it still appropriate to use logistic model? should I do any other test before running a logistic regression?


Also, if I want to estimate default probability in lets say 3 years time, how can I do it?


I have little of information, therefore need some help.

Thanks

[David Greenberg]

You need to take into account the panel structure of the data. David
Greenberg, Sociology Department, New York University

On Sun, Jun 8, 2014 at 5:39 AM, simona <[hidden email]> wrote:

> Hi all, I need your help.
>
> I am estimating probability of firms default.
> Have data annual data from 2002 - 2012 for each firm with 6 variables
> Therefore most likely that objectives are correlated. Is it still
> appropriate to use logistic model? should I do any other test before running
> a logistic regression?
>
>
> Also, if I want to estimate default in lets say 3 years time, how can I do
> it?
>
>
> I have little of information, therefore need some help.
>
> Thanks
>

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD