Re: st: stepwise and manual drop of variables

[Richard Williams <richardwilliams.ndu@gmail.com>]

At 04:00 PM 2/10/2011, Rodrigo Briceño wrote:
> Dear Stata users. I would like to know if there is a difference when
> applying the command Stepwise(SW) and doing the elimination of
> variables manually.

When you did it manually, did you, in fact, do it 
one step at a time? When doing backwards 
selection, you should drop one variable, rerun 
the regression, drop another variable, etc. If 
you just drop all the insignificant variables at 
once that is not the same as stepwise. The fact 
that pprom made it in to your final model with a 
p-value of .0644 indicates you did something 
wrong; it should have been dropped (assuming this isn't just a typo).

I'll leave it to others to point out the problems 
with stepwise, but you should be aware of them if you aren't already.

> I recently run this two models:
>
> 1. regress logliq logsap l1liq l2liq l3liq l4liq l1sap l2sap l3sap
> l4sap liqmk pprom sem dv if isin2==169 & ano>=2009
> 2. sw regress logliq logsap l1liq l2liq l3liq l4liq l1sap l2sap l3sap
> l4sap liqmk pprom sem dv if isin2==169 & ano>=2009, pr(.05)
>
> When I drop the variables with p-value greater than 0.05 I got as
> final variables: sem (0.0315) and pprom (0.0644): p-values in
> parenthesis
> If I apply the second option my final variables are: sem (0.022), dv
> (0.027) and pprom (0.007).
>
> My final goal is to reduce the model with all the variables to a
> simple model with just significant variables (0.05 or lower p-value).
>
> Is there an explanation for the difference when applying both methods?
> in that case would be better to do the elimination of variables one by
> one?
>
> thanks for your help.
>
> --
> Rodrigo Briceño
> Economist
> rjbriceno@gmail.com
> MSN: jbric98@hotmail.com
> SKYPE: rbriceno1087
>
> *
> *   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/

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam


*
*   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/