Re: st: Multivariate imputation using chained equations

[Stas Kolenikov <skolenik@gmail.com>]

What version of Stata do you have? Do you use the official -mi impute
chained- or Patrick Royston's -ice-?

Btw, I would use something like

recode <varlist> (-1 = .a) (-2 = .b) (-3 = .c) (-4 = .d) (-5 = .e) (
-6 = .f ) ( -7 = .g )

to retain the information about missing values. That way, you could
also use -ice, conditional- or -mi impute chained (regress if inlist(
thisvar, .a, .b, .c) ) thisvar-

-- Stas Kolenikov, PhD, PStat (SSC)
-- Senior Survey Statistician, Abt SRBI
-- Opinions stated in this email are mine only, and do not reflect the
position of my employer
-- http://stas.kolenikov.name



On Mon, Jul 8, 2013 at 11:45 AM, Jacqueline Jodl
<jmj2138@tc.columbia.edu> wrote:
> I am attempting multivariate imputation using chained equations.
>
> My longitudinal dataset (NLSY79) codes the missing data that I wish to
> impute as -1, -2, and -3 (don't know, refusal, and invalid skip,
> respectively). The data that I do NOT want to impute is coded at -4, -5, and
> -7 (noninterviews and valid skips).
>
> I have included the following in my syntax to account for the observations
> that I wish to impute:
>
> ds*
>      local allvars = r(varlist)
>      foreach var of local allvars {
>         replace `var' = . if `var' == -1 | `var' == -2 | `var' == -3
>      }
>
> Is there a similar approach to including syntax that excludes the data
> points coded as -4, -5, and -7?
>
> Thanks,
> Jacqueline Jodl
> Teachers College
> Columbia University
> New York, New York
>
> *
> *   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?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/