st: marginal effects after xtlogit

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

st: marginal effects after xtlogit

Shuaizhang Feng
Dear list:

I just found that after using xtlogit (re), the command "margins, dydx(*)" produce identical average marginal effects as the estimated coefficents. This is for both discrete and continuous explanatory variables. Using the same command following logit seems OK. Is this some kind of error on stat's side?

Thank you for any response.

Shuaizhang


     
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

st: AW: marginal effects after xtlogit

Martin Weiss-5

<>

Note that the default prediction after -xtlogit, re- is the linear
prediction, as seen in -help xtlogit postestimation-. For -logit-, it is the
probability. I guess this is the source of the confusion: -margins- uses the
default -predict- option after the respective estimation command, unless
instructed otherwise...




HTH
Martin

-----Ursprüngliche Nachricht-----
Von: [hidden email]
[mailto:[hidden email]] Im Auftrag von Shuaizhang Feng
Gesendet: Montag, 30. November 2009 22:13
An: [hidden email]
Betreff: st: marginal effects after xtlogit

Dear list:

I just found that after using xtlogit (re), the command "margins, dydx(*)"
produce identical average marginal effects as the estimated coefficents.
This is for both discrete and continuous explanatory variables. Using the
same command following logit seems OK. Is this some kind of error on stat's
side?

Thank you for any response.

Shuaizhang


     
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

Re: st: AW: marginal effects after xtlogit

Shuaizhang Feng
Thanks for the clarification.

But what should I do after running xtlogit , in order to get results on average marginal effect that are comparable to "logit" and then "margins, dydx(*)"? I have looked at the help files but could not figure this out.


--- On Mon, 11/30/09, Martin Weiss <[hidden email]> wrote:

> From: Martin Weiss <[hidden email]>
> Subject: st: AW: marginal effects after xtlogit
> To: [hidden email]
> Date: Monday, November 30, 2009, 4:54 PM
>
> <>
>
> Note that the default prediction after -xtlogit, re- is the
> linear
> prediction, as seen in -help xtlogit postestimation-. For
> -logit-, it is the
> probability. I guess this is the source of the confusion:
> -margins- uses the
> default -predict- option after the respective estimation
> command, unless
> instructed otherwise...
>
>
>
>
> HTH
> Martin
>
> -----Ursprüngliche Nachricht-----
> Von: [hidden email]
> [mailto:[hidden email]]
> Im Auftrag von Shuaizhang Feng
> Gesendet: Montag, 30. November 2009 22:13
> An: [hidden email]
> Betreff: st: marginal effects after xtlogit
>
> Dear list:
>
> I just found that after using xtlogit (re), the command
> "margins, dydx(*)"
> produce identical average marginal effects as the estimated
> coefficents.
> This is for both discrete and continuous explanatory
> variables. Using the
> same command following logit seems OK. Is this some kind of
> error on stat's
> side?
>
> Thank you for any response.
>
> Shuaizhang
>
>
>      
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>


     

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

Re: st: AW: marginal effects after xtlogit

Maarten buis
--- On Mon, 30/11/09, Shuaizhang Feng wrote:
> But what should I do after running xtlogit , in order to
> get results on average marginal effect that are comparable
> to "logit" and then "margins, dydx(*)"? I have looked at the
> help files but could not figure this out.

The closest you can get is to specify the -predict(pu0)- option.
This will fix the group specific intercept at 0 (= the mean), but
otherwise averages the marginal effects over the other explanatory
variables. So this is not quite an average marginal effect, but
a mixture between the marginal effect at average values for the
explanatory variables (the group specific intercepts) and
average marginal effects (over the other variables). Purists would
probably prefer to use the BLUPs for the intercepts and also
average over those.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


     

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

Re: st: marginal effects after xtlogit

ammariaymen
This post has NOT been accepted by the mailing list yet.
In reply to this post by Shuaizhang Feng
The default predict option after xtlogit is -xb- , which stands for linear prediction. -margins- uses the default prediction option. If you don't like that you can use -pu0-. which gives the probability of the specified outcome (outcome()) assuming that the random effect is zero.