Installing "MI" - Stata

[Mariana Silveira]

Hello,

I'm sorry for asking this kind of "very basic/easy" question, but I would like to know how to install the MI package of Stata (created by Ken Scheve).
I had used the link bellow for dowlonding the program, but I didn't arrive to install: the software doesn't recognize the comands for installing.

http://gking.harvard.edu/files/gking/files/mi.zip

Does anybody could give some tips about using MI? Thank you a lot !

Mariana

[Nick Cox]

Opening the .zip file directly shows README and four files to be installed.

You need to install those manually

The README file does tell you how to install! Here it is copied below.

!!! But watch out: daniel klein also has a command called misum on SSC.That's completely different from the one in this package.

By the way, since this package was produced, Stata has introduced an immensely more powerful mi suite.

Contents: Stata procedures for multiple imputation analysis.
Files: miest.ado
miest.hlp
misum.ado
misum.hlp
Version: 2.1
Date: April 29, 2003
Author: Kenneth Scheve
Department of Political Science
Yale University
[email protected]

DESCRIPTION

"miest" combines the results of statistical analyses across multiply imputed
datasets and reports the resulting multiple imputation estimates. "misum"
implements the same procedures but for calculating multiple imputation
estimates of the mean.

Multiple imputation is a method of statistical inference for incomplete
multivariate data. The method involves imputing m values for each
missing cell in your data matrix and creating m "completed" datasets.
Across these completed datasets, the observed values are the same, but the
missing values are filled in with different imputations that reflect
uncertainty about the missing data. There are a number of alternative methods
for producing the imputations. One that is particularly simple to use is
Amelia: A Program for Missing Data by James Honaker, Anne Joseph, Gary King,
Kenneth Scheve, and Naunihal Singh. It is freely available as a standalone
windows program and as a Gauss program at http://GKing.Harvard.edu/. After
creating the multiply imputed datasets, you then apply whatever statistical
method you would have used if there had been no missing values to each of
the m data sets, and use a simple procedure, implemented by "miest", to combine
the results. The multiple imputation point estimate for the parameters of
the statistical model is simply the average of the estimates across the m
analyses. The variance of this point estimate is the average of the estimated
variances from within each completed dataset, plus the sample variance in
the point estimates across the datasets (multiplied by a factor that corrects
for bias because the number of imputed datasets m is less than infinity). For
more information about multiple imputation, see "Analyzing Incomplete
Political Science Data: An Alternative Algorithm for Multiple Imputation"
American Political Science Review Vol. 95 No. 1 (March 2001):49-69
by Gary King, James Honaker, Anne Joseph and Kenneth Scheve.

Once you have installed the procedures, consult the on-line Stata help
files for a description of the command syntax for "miest" and "misum".

INSTALLATION

Copy miest.ado, miest.hlp, misum.ado, and misum.hlp to your personal ado
directory. To get the name of this directory, type

adopath

at the Stata command prompt. You should see a message that looks like this:

. adopath
[1] (UPDATES) "C:\STATA\ADO\UPDATES/"
[2] (BASE) "C:\STATA\ADO\BASE/"
[3] (SITE) "C:\STATA\ADO\SITE/"
[4] "."
[5] (PERSONAL) "C:\ADO\PERSONAL/"
[6] (STBPLUS) "C:\ADO\STBPLUS/"
[7] (OLDPLACE) "C:\ADO/"

The fifth directory is your personal ado directory; in this
example, it is C:\ADO\PERSONAL. If this directory does not exist on your
computer, you will need to create the directory from a DOS prompt or by
using Windows Explorer.

ADDITIONAL INFORMATION

If you find a bug or have a comment about the procedure, please send mail to
the author at [email protected]. Updates of this zip file will be
posted at http://GKing.Harvard.Edu.

Thanks to Nick Cox, James Honaker, Anne Joseph, Gary King, and Michael Tomz
for very helpful comments and suggestions.

[Mariana Silveira]

Thank you!

Acctually, I had followed the indicated steps on the "read-me" document. Even doing this, it doesn't work. But that's ok, I'll try to understand what's happening here. Thank you a lot by the way.

[Nick Cox]

"doesn't work" unfortunately means nothing we can reply to without details.