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.

# st: Re: Question about scaling intercept

 From Christopher Baum To "BOONYAWAT K." Subject st: Re: Question about scaling intercept Date Wed, 28 Nov 2012 19:16:22 +0000

```<>
The easiest way to do this is with analytic weights. Using analytic weights keeps the summary statistics in the usual realm, which manual scaling does not.

. sysuse auto,clear
(1978 Automobile Data)

. g pp = price/mpg

. g hh = headroom/mpg

. g ll = length/mpg

. g mm = 1/mpg

. reg pp hh ll mm, nocons

Source |       SS       df       MS              Number of obs =      74
-------------+------------------------------           F(  3,    71) =  135.81
Model |  10400518.4     3  3466839.45           Prob > F      =  0.0000
Residual |  1812466.22    71  25527.6932           R-squared     =  0.8516
-------------+------------------------------           Adj R-squared =  0.8453
Total |  12212984.6    74  165040.332           Root MSE      =  159.77

------------------------------------------------------------------------------
pp |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
hh |  -722.7482   503.9827    -1.43   0.156    -1727.661    282.1648
ll |   82.69822   19.34002     4.28   0.000     44.13532    121.2611
mm |  -6918.923   3358.944    -2.06   0.043    -13616.47   -221.3797
------------------------------------------------------------------------------

. reg price headroom length [aw=1/mpg^2]
(sum of wgt is   1.9839e-01)

Source |       SS       df       MS              Number of obs =      74
-------------+------------------------------           F(  2,    71) =    9.41
Model |   179144244     2  89572121.9           Prob > F      =  0.0002
Residual |   676064553    71  9522035.95           R-squared     =  0.2095
-------------+------------------------------           Adj R-squared =  0.1872
Total |   855208796    73    11715189           Root MSE      =  3085.8

------------------------------------------------------------------------------
price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
headroom |   -722.748   503.9827    -1.43   0.156    -1727.661     282.165
length |   82.69822   19.34002     4.28   0.000     44.13532    121.2611
_cons |  -6918.924   3358.944    -2.06   0.043    -13616.47   -221.3807
------------------------------------------------------------------------------

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html

On Nov 28, 2012, at 1:50 PM, BOONYAWAT K. <karuntarat.boonyawat@durham.ac.uk>
wrote:

> Dear Prof. Kit,
>
> I wonder if you could kindly give me some suggestions about how to scale intercept.
>
> I would like to scale all variables including intercept with "1/asset" in the following equation.
>
> yit =  a+ β1Xit  +eit
>
>
> Do I need to scale all variables manually before running the equation? I am not sure how to scale intercept manually (because Stata will calculate the intercept automatically).
>
> or is it the same as I run the equation with -regress-  , nocon ?
>
> yit /(1/asset)=  a(1/asset) + β1X/(1/asset)it  +eit
>
> Thank you very much for your kindly attention.
>
> Best regards,
> Karuntarat
>
>
>
>

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