Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: =?gb2312?B?tPC4tDogc3Q6IEhvdyB0byBjaG9vc2UgYSBwcm9wZXIgbW9kZWwgaQ==?==?gb2312?B?ZiB0aGUgZGVwZW5kZW50IHZhcmlhYmxlIGlzIHdpdGhpbiBib3VuZHM/?=


From   "Cheng, Xiaoqiang" <Xiaoqiang.Cheng@econ.kuleuven.be>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: =?gb2312?B?tPC4tDogc3Q6IEhvdyB0byBjaG9vc2UgYSBwcm9wZXIgbW9kZWwgaQ==?==?gb2312?B?ZiB0aGUgZGVwZW5kZW50IHZhcmlhYmxlIGlzIHdpdGhpbiBib3VuZHM/?=
Date   Sun, 1 Oct 2006 19:46:13 +0200

Thank you very much.
It seems very helpful. 
BTW, since  "glm"  estimates likelihood, should we then interpret the coefficients of regressors as probability? 

________________________________

From: owner-statalist@hsphsun2.harvard.edu 代表 Ronnie Babigumira
Sent: 10/1/2006 (星期日) 2:14 下午
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: How to choose a proper model if the dependent variable is within bounds?



These might be helpful

http://www.stata.com/support/faqs/stat/logit.html
http://biostat.mc.vanderbilt.edu/twiki/pub/Main/BioMod/goldstein.ratios.pdf

You can also take a look at Sean Buckley's -mlbeta- ( net from http://www2.bc.edu/~bucklesj)


Cheng, Xiaoqiang wrote:
> Hi all,
>
> I want to estimate the effects of personal characteristics on the turnedown ratio of loan applicants.
>
> The dependent variable is the loan turndown ratio, lets say, varies between 0 and 1. If the applicant gets the desired loan amount she applies, the turndown ratio will be 0; and if she is totally turned down, the turndown ratio will be 1. However, if she gets some loan but not as much as she desires, the turndown ratio might vary within 0 and 1.
>
> To test the effects, I tried several ways,
>
> 1. Using "intreg" to treat the sample as censored at two bounds: 0 and 1
>
>
> 2. Simply using OLS to estimate the full sample (I want to keep the information of the observations when the turndown ratio is 1 and 0).
>
>
> 3. Estimating the sub sample with the turndown ratio equals 0 and 1,  using a probit model. (In this case, the sub sample looks like a truncated one, but do I have a better choose of model?)
>
>
> In general, I feel uncomfortable with my tests, because option 1 losses the information when the turndown ration is 1 and 0.  Option 2 looks to generate biased estimates, so does Option 3.
>
> So, what should I do then especially when I want to keep as much as information as I can?  Thank you.
>
> Best,
>
> Xiaoqiang Cheng
>
> Xiaoqiang Cheng
>
> University of Leuven
> Tel  +32 16 326853
> Fax  +32 16 326796
> Mail  Xiaoqiang Cheng
>       Center for Economic Studies
>       University of Leuven
>       Naamsestraat 69
>       Leuven,  Belgium
>       B3000
> Url www.econ.kuleuven.be/xiaoqiang.cheng
>
>
>
>
>
> Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm
>
>
> *
> *   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/




Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm

<<winmail.dat>>




© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index