Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

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

st: posting question on multiple imputation

From   "Jacqueline Jodl" <>
To   <>
Subject   st: posting question on multiple imputation
Date   Thu, 27 Jun 2013 21:45:27 -0400

My question is related to Multiple Imputation Chained Equation. My data is
from the NLSY79 Children and Young Adult database which is a longitudinal
study with 9 waves beginning in 1994 to 2010. 

While I have successfully reshaped so I can move from wide to long format, I
am concerned that I have not properly accounted for missing data in my
syntax, specifically noninterview and item nonresponse, in such a way that
when I do my multiple imputation I do not impute noninterview and valid
skips (when a respondent was purposely not eligible for a question).  

This is how the National Bureau of Labor Statistics explains missing data in
my dataset:

National Longitudinal Survey of YOUTH|1979 - Children and Young Adult
Missing Data: Noninterviews and Item Nonresponse
Missing values are indicated in the data and on the codebook page for each
individual question. Following general NLS convention, a response of "don't
know" to an individual question is coded "-2," a refusal to answer an
individual question is coded "-1," and an invalid skip is coded "-3."
(Invalid skip means that the respondent should have answered the question
but didn't; this was more common in the paper-and-pencil interviews when the
respondent or interviewer might make an error following the skip pattern on
a paper instrument. The incidence of invalid skips has been significantly
reduced in the computer-assisted interviews.) Noninterviews and valid skips
(when a respondent was purposely not eligible for a question) have been
collapsed into a "-7." Thus, users must know how to distinguish between
noninterviews and valid skips for some types of research. This can be done
by using the weights to determine if a respondent is a noninterview. A
weight equal to zero indicates noninterview status. 

Can anyone advise me on the proper syntax for reshaping wide to long so when
I impute I do not impute noninterview and valid skips?


*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index