Bookmark and Share

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


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: predict after logit


From   Tim Wade <[email protected]>
To   [email protected]
Subject   Re: st: predict after logit
Date   Thu, 12 Jul 2012 15:39:17 -0400

Without having margins, you can always use a bootstrap, need to check
the syntax for -bootstrap- in version 10.

sysuse auto.dta

xtile hiprice=price, nq(2)


program bootphat, rclass
logistic  foreign hiprice mpg
tempvar phat
predict `phat'
qui summ(`phat')
return scalar pm=r(mean)
qui summ(`phat') if hiprice==1
return scalar plo=r(mean)
qui summ(`phat') if hiprice==2
return scalar phi=r(mean)
end



bootstrap pmean=r(pm)  phi=r(phi) plo=r(plo), reps(1000): bootphat





On Thu, Jul 12, 2012 at 2:08 PM, Sara Borelli <[email protected]> wrote:
> thank you Tim for the suggestion,
> I cannot find the margins options...may it be because this is a new
> feature of STATA ?
> I am using stata 10
> thanks
> Sara
>
>
> On Wed, Jul 11, 2012 at 5:27 PM, Tim Wade <[email protected]> wrote:
>> apologies,  the first segment of code  should read:
>>
>> sysuse auto.dta
>> xtile hiprice=price, nq(2)
>> logistic foreign hiprice mpg
>> margins, by(hiprice)
>>
>>
>> On Wed, Jul 11, 2012 at 5:26 PM, Tim Wade <[email protected]> wrote:
>>> If I understand what you are asking, I think margins does this:
>>>
>>> margins, by(hiprice)
>>> sysuse auto.dta
>>> xtile hiprice=price, nq(2)
>>> logistic foreign hiprice mpg
>>> margins, by(hiprice)
>>>
>>>
>>> ------------------------------------------------------------------------------
>>>              |            Delta-method
>>>              |     Margin   Std. Err.      z    P>|z|     [95% Conf. Interval]
>>> -------------+----------------------------------------------------------------
>>>      hiprice |
>>>           1  |   .2162162   .0557663     3.88   0.000     .1069163    .3255161
>>>           2  |   .3783784   .0681487     5.55   0.000     .2448094    .5119474
>>> ------------------------------------------------------------------------------
>>>
>>>
>>> note this margin estimates are the same as the means of phat:
>>>
>>> logistic foreign hiprice mpg
>>> predict phat
>>> tabulate hiprice, summ(phat)
>>>
>>>
>>> 2 quantiles |       Summary of Pr(foreign)
>>>    of price |        Mean   Std. Dev.       Freq.
>>> ------------+------------------------------------
>>>           1 |   .21621622   .23645819          37
>>>           2 |   .37837838   .25520846          37
>>> ------------+------------------------------------
>>>       Total |    .2972973   .25759862          74
>>>
>>> Tim
>>>
>>>
>>> On Wed, Jul 11, 2012 at 1:01 PM, Sara Borelli <[email protected]> wrote:
>>>> Hi all,
>>>> I am running a logit model to obtain a predicted probability for each
>>>> individual (phat_i)  using predict, pr command .
>>>>
>>>>  I have to calculate the average of  this prediction within some
>>>> categories of a variable x:
>>>>  mean_phat_i=mean( phat_i)  by(x)
>>>> and obtain standard errors and confidence interval for this mean.
>>>>
>>>> I know I can use adjust for this purpose, but I also need to calculate
>>>> standard errors/confidence intervals  for mean_phat_i
>>>>
>>>> Thanks in advance for any insight
>>>>
>>>> Sara
>>>> *
>>>> *   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/
*
*   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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index