Statalist


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

Re: AW: st: imputing continuous values when respondents select categories, e.g., income category


From   Richard Williams <Richard.A.Williams.5@ND.edu>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>, "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   Re: AW: st: imputing continuous values when respondents select categories, e.g., income category
Date   Fri, 24 Apr 2009 17:09:40 -0500

At 03:29 PM 4/24/2009, Alan Acock wrote:
Thanks for recommending intreg. I was not familiar with this command. When I run the example in the reference manual, the estimated income is not always within the limits of the interval the respondent selected. Using their example and predicting est_income, here are the first 10 observations:
  +---------------------------+
     | wage1   wage2    est_wage |
     |---------------------------|
  1. |     .       5    4.266946 |
  2. |     5      10    7.502522 |
  3. |     5      10    10.19239 |
  4. |    10      15    8.924339 |
  5. |     .       5    8.116896 |
     |---------------------------|
  6. |     .       5     10.2202 |
  7. |     .       5    10.35355 |
  8. |     5      10   -2.894233 |
  9. |     5      10    13.32243 |
 10. |     5      10    9.792462 |

Some estimated values are well outside of the interval. Is it recommended to simply replace these with the interval boundaries?

There is nothing that says the estimated wages have to lie within the interval. The est_wage is just E(Y | X). Just as in regular regression, the estimates will sometimes be higher than the observed value, sometimes lower. My guess is that replacing them with the interval boundaries would be a terrible idea.

The intreg approach might be good if income was your dependent variable, but by the way you are talking I'm guessing you really want it as an independent variable.

Sometimes people just use the midpoint of the interval; other times they will just break income up into dummy variables. I believe there are other more advanced methods but I am not that familiar with them.


-------------------------------------------
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/



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