Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.

# Re: st: Confidence interval in natural units after GLM with log link

 From Shehzad Ali To "statalist@hsphsun2.harvard.edu" Subject Re: st: Confidence interval in natural units after GLM with log link Date Thu, 3 Nov 2011 17:59:27 +0000 (GMT)

```Thanks, Maarten. Very helpful.

In my particular example, I am interested in the difference between costs in two treatment arms. Hence, based on your suggestion, I suppose -margins, dydx(treat) - should do the trick?

However, a related question is: do the CI and p-value on 'average marginal effects' from -margins, dydx(treat) - and 'marginal effects at means' -margins, dydx(treat) at means - differ from the coefficient from linear regression? When I implemented a simple OLS, I noticed that the p-value and CI were slightly different from the marginal effects (coefficients were the same). Hence, if I used -margins, dydx(treat) - after glm, would the marginal effect represent the linear coefficient?

Thank you

>________________________________
>From: Maarten Buis <maartenlbuis@gmail.com>
>To: statalist@hsphsun2.harvard.edu
>Sent: Wednesday, 2 November 2011, 18:52
>Subject: Re: st: Confidence interval in natural units after GLM with log link
>
>On Wed, Nov 2, 2011 at 6:50 PM, Shehzad Ali <drshehzad_ali@yahoo.com> wrote:
>> I am using GLM model with log link and gamma family for cost data. Covariates include treatment variable (1 for new treatment, 0 for old). What would be the best way to obtain cost difference in natural scale, the confidence interval around it and the p-value? I understand one option is to predict costs for the two arms and take the difference and estimate the confidence interval by bootstrap method. Is there another way to do it (parametrically or non-parametrically)?
>
>This is what -margins- is for, see -help margins-
>
>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/
>
>
>

*
*   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/
```