Statalist


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

st: Imputing values from bracketed responses


From   diebold@email.unc.edu
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   st: Imputing values from bracketed responses
Date   Thu, 11 Jun 2009 09:39:56 -0400

Hello,

I have a data set where respondents were asked for continuous values for variables such as income. If individuals were unsure of a specific value they were provided with intervals (0-10,000, 10,001-20,000, 20,001-30,000...) and asked to select which interval best approximated their income. I would like to replace these bracketed responses with continuous values.

My plan is to regress income on a set of covariates and obtain predicted values from the regression line. I will then match respondents with continuous values to those with bracketed responses using these predicted values and impute the continuous value for the matched individual with a bracketed response.

First, is this a reasonable way to get continuous values from bracketed responses?

Second, what matching program would you recommend?

Third, after the match, what command should I use to tell Stata to impute the continuous value of the matched respondent for the individual with a bracketed response?

The diagram below is an example of what I am trying to accomplish. Where "income" are continuous value data, "incomeb" are bracketed responses, and "matchid" indicate the matched respondent for those with bracketed reponses.

id income incomeb ystar matchid new income 1 12 . 12 . 12
2         15           .          14.5         .         15
3         35           .          38.25        .         35
4         42           .          44.5         .         42
5          .         10-15        10.75        1         12
6          .         30-40        37           3         35

In this instance, individuals are matched based on the predicted income "ystar" and I need to know a command that will allow me to put the income for respondent "1" for matched respondent "5" and the income for respondent "3" into matched respondent "6". The "new income" variable is the result that I am hoping for in the end.

Thank you very much.
Jeff D


*
*   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