Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# Re: st: RE: Interpretation of regression outputs when variables are log transformed

 From Nick Cox To statalist@hsphsun2.harvard.edu Subject Re: st: RE: Interpretation of regression outputs when variables are log transformed Date Thu, 3 Mar 2011 11:36:42 +0000

```No. From your results as presented below 1.33 is a slope or gradient
for log X w.r.t. log Y, not for X w.r.t. Y.

I think you need a basic book or to find someone who will explain this
to you one-to-one, as you appear to be asking the same question and
making no progress.

Nick

On Thu, Mar 3, 2011 at 8:37 AM, Marco Buur <marco.buur@gmail.com> wrote:
> Dear Nick
> Thanks for the tip
> I have xtreg, re cluster () model where eform() doesn't work. I run it
> as reg, cluster () and it works. I assume that now we interpret the
> results in the way as without log-log transformation?
>  So for every unit increase in Y, a 1.33 unit increase in X will be
> predicted, holding Z variable constant?
> -----------------------------------------------------
>             |               Robust
> log(X)       |   GM/Ratio   Std. Err.      t    P>|t|
> -------------+----------------------------------------
> log(Y)       |    1.33217     .145      4.44   0.000
> log(Z)       |    2.724049    .333      7.67   0.000
>
>
>
> Many thanks
> Marco
> On Wed, Mar 2, 2011 at 7:07 PM, Nick Cox <n.j.cox@durham.ac.uk> wrote:
>> This may help:
>>
>> SJ-3-4  st0054  . . . . . . . . . . Stata tip 1: The eform() option of regress
>>        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  R. Newson
>>        Q4/03   SJ 3(4):445                                      (no commands)
>>        tips for using the eform() option of regress
>>
>> http://www.stata-journal.com/sjpdf.html?articlenum=st0054
>>
>> Nick
>> n.j.cox@durham.ac.uk
>>
>> Marco Buur
>>
>> Me and my colleague are not sure that our interpretations of the
>> results are correct. Please confirm!
>>
>> 10% increase in Y will increase X for 1.1^0.223=1.021481649 which is
>> 2.1% and 10 % increase in Z will increase X for 1.1^1.05=1.10525457
>> which is 10.5%.
>>
>> Is this correct?
>>
>> -----------------------------------------------------------
>>              |               Robust
>>    Log(X)    |      Coef.   Std. Err.      z    P>|z|
>> -------------+----------------------------------------------
>>    Log(Y)    |   .223     .033     3.57   0.000
>>    Log(Z)    |   1.05     .1045    11.14  0.000

*
*   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/
```