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: IV and matching estimator


From   Ayman Farahat <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: st: IV and matching estimator
Date   Wed, 9 Jan 2013 09:49:46 -0800 (PST)

Hello Dimitry 

What i want to do is the following . 
I want to use an IV (2sls in a way). 
In first stage, i want to estimate the number of impression seen based on number or impression seen by  similar users that day and perhaps lagged number of impressions of same user. This will give me first stage estimate of the number of impressions. 
In the second stage, i will use estimate from first stage in a probit/logit. 
Does that make sense?



----- Original Message -----
From: Dimitriy V. Masterov <[email protected]>
To: Statalist <[email protected]>
Cc: 
Sent: Wednesday, January 9, 2013 9:35 AM
Subject: Re: st: IV and matching estimator

Ayman,

If I understand your problem, you have selection on unobservables.
People who are enthusiastic consumers are likely to both see lots of
advertisements as they are browsing and are also likely to buy or
convert. Presumably, this creates a positive correlation between the
error term and the impressions variable, so you get biased results
with OLS.

Unfortunately, propensity score matching will only deal with the
selection on observables problem, so I don't think this sort of
approach will work.

DVM

On Tue, Jan 8, 2013 at 7:23 PM, Ayman Farahat <[email protected]> wrote:
>  am interested in evaluating the marginal effect of seeing an advertising impression on conversion (binary).
> However, because of activity bias , the number of impression seen is correlated with the unobserved error and are therefore exogenous.
>
> I was thinking about using an IV to get around this problem. The instruments that I have are number of impressions of similar users on that day.
> That seems very much like a matching estimator.
>
> Is this possible? Is there a way of doing this in Stata?
> Best
> Ayman
>
>
> *
> *   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/
*
*   For searches and help try:
*  http://www.stata.com/help.cgi?searchhttp://www.stata.com/support/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


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