Does not that mean that this,
> Turning of the effect of X:
> 0.99%*1000000=9900
>
> Turning on the effect of X:
> 0.29%*1000000=2900
>
> Then, the way I have understood this:
> Discrete change, reduction induced by x=9900-2900=7000?
Will be wrong? Here I have applied the on and off effects on the total sample.
In the model n=1000000, and there is 500000 members of the program (x=1), thus 600000 are not member.
I the example above I argue that x causes a reduction in the 1000000 sample of 7000, due to the dicrete change of .7%.
Alex
-----Opprinnelig melding-----
Fra: [email protected] [mailto:[email protected]] P� vegne av Maarten buis
Sendt: 11. januar 2008 16:11
Til: [email protected]
Emne: Re: SV: st: From probit to dprobit to interpretation
The separate probabilities need to be applied to their group, but the discrete change needs to be applied to the total sample.
--- [email protected] wrote:
> Thanks Maarten! That is very helpful. I guess what have been confusing
> for me then is how to apply the predicted -0.7% discrete change (the
> difference between turning on and off the effect), on the full sample
> as I have done below, or only on those 500000 that are signed up to
> the membership program. The difference offcourse making a huge impact
> on the result.
>
> Best wishes,
> Alexander
>
> -----Opprinnelig melding-----
> Fra: [email protected]
> [mailto:[email protected]] P� vegne av Maarten buis
> Sendt: 11. januar 2008 14:16
> Til: stata list
> Emne: RE: st: From probit to dprobit to interpretation
>
> What you say is correct and there is no contradiction between all
> these statements. From a probit model you can derive predicted
> proportions, and with predicted proportions you can derive predicted
> counts in your sample (and if you know the size of your population the
> predicted counts in your population).
>
> Hope this helps,
> Maarten
>
> --- [email protected] wrote:
> I have estimated a probit model where n=1000 000 customers with only
> 1 independent dummy variable (x) (for the sake of clarity), and get
> the following estimated coefficients:
>
> y_pred=-2.33-0.431*x (x being significant)
>
> No the way I understand this is that these coefficients, except for
> the signs and significance level, is hard to interpret. Thus, I can
> derive it as a probability model, and then again calculate
> probabilities from any table with standard cumulative normal
> distribution values. Turning on and off x will give me the discrete
> change, thus
>
> Turning off the effect of X thus gives me:
> y_pred=-2.33-(0.431*0) and
> Pr(z<2.33)=0.99%
>
> Tuning on the effect
> y_pred=-2.33-(0.431*1)=-2.761 and
> Pr(z<2.761)=0.29%
>
> The difference between these probabilities is the discrete change, and
> this change can be directly estimated using a dprobit model in Stata?
> Discrete change=0.99-0.29=-0.7%
>
> Most textbooks stops here, and I think that so far I am on the right
> track - but I want to interpret this probability in terms of what this
> x induced effect means in terms of my sample...
>
> In this particular model my sample is 1000000, and x=1 is a membership
> program of which there are 500000 members. Would it be correct to
> assume that the discrete change estimated above in terms of customers
> could be interpreted as following:
>
> Turning of the effect of X:
> 0.99%*1000000=9900
>
> Turning on the effect of X:
> 0.29%*1000000=2900
>
> Then, the way I have understood this:
> Discrete change, reduction induced by x=9900-2900=7000?
>
>
>
> -----------------------------------------
> Maarten L. Buis
> Department of Social Research Methodology Vrije Universiteit Amsterdam
> Boelelaan 1081
> 1081 HV Amsterdam
> The Netherlands
>
> visiting address:
> Buitenveldertselaan 3 (Metropolitan), room Z434
>
> +31 20 5986715
>
> http://home.fsw.vu.nl/m.buis/
> -----------------------------------------
>
>
> ___________________________________________________________
> Support the World Aids Awareness campaign this month with Yahoo! For
> Good http://uk.promotions.yahoo.com/forgood/
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology Vrije Universiteit Amsterdam Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
__________________________________________________________
Sent from Yahoo! Mail - a smarter inbox http://uk.mail.yahoo.com
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
