Statalist


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

Re: st: Re: Calculate beta values for mim results


From   Frank Gallo <fjgallo@mac.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Re: Calculate beta values for mim results
Date   Mon, 04 May 2009 19:32:19 -0400

<>

Hi Maarten,

Would you consider 1.2E-15 close enough? Thank you.

Best,
Frank


On May 4, 2009, at 5:12 AM, Maarten buis wrote:


Looks ok, one way to test this is that the constant should be very
very small (actually 0, but usually in the order of 1E-16.

Hope this helps,
Maarten

-----------------------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://home.fsw.vu.nl/m.buis/
-----------------------------------------


--- On Sun, 3/5/09, Frank Gallo <fjgallo@mac.com> wrote:

From: Frank Gallo <fjgallo@mac.com>
Subject: Re: st: Re: Calculate beta values for mim results
To: statalist@hsphsun2.harvard.edu
Date: Sunday, 3 May, 2009, 11:40 PM
<>
<>

Hi Maarten,

I just wanted to check-in with you on my below message,
which might have been lost in the shuffle of trying to
correct my posting dilemma. Did I carryout the below syntax
correctly based on your suggestion? Thank you.

Best,
Frank


On May 2, 2009, at 10:04 AM, fjgallo@mac.com wrote:

Hi Maarten,

First, I opened the data file in which I saved the stacked
imputed data sets: m(5), each set n = 3,300. Next, I ran the
below syntax (for my below example, I only included a few
predictor variables and the criterion variable pforce). I
obtained coefficients for the run. Did I do this correctly?
Please remember that I am a Stata beginner. Thank you.

Best Frank

----------------------------------------------------------
mim, storebv: regress pforce assign time day

foreach var of varlist pforce assign time day {
gen double z`var' =.
}

forvalues i = 1/5 {
foreach var of varlist pforce assign time day {
sum `var' if _mj ==`i'
replace z `var' = (`var' - r(mean)) / r(sd) if _mj
== `i'
}
}

mim: regress zpforce zassign ztime zday

-------------------------------------------------------------





On May 1, 2009, at 12:04 PM, Maarten buis wrote:


This is one way of getting standardized coefficients after
multiple imputation:

*------------------- begin example
----------------------------
sysuse auto, clear
ice rep78 price mpg, m(5) clear

foreach var of varlist rep78 price mpg {
	gen double z`var' = .
}

forvalues i = 1/5 {
	foreach var of varlist rep78 price mpg {
		sum `var' if _mj == `i'
		replace z`var' = (`var' - r(mean)) / r(sd) if _mj
== `i'
	}
}

mim : reg zprice zrep78 zmpg
*--------------------- end example
---------------------------------

Hope this helps,
Maarten

-----------------------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://home.fsw.vu.nl/m.buis/
-----------------------------------------


--- On Fri, 1/5/09, fjgallo@mac.com <fjgallo@mac.com>
wrote:

From: fjgallo@mac.com <fjgallo@mac.com>
Subject: Re: st: Re: Calculate beta values for mim
results
To: statalist@hsphsun2.harvard.edu
Date: Friday, 1 May, 2009, 4:16 PM
Thank you. Your thoughts were helpful, and I also hope
that
an experienced person with the -mim- command might
offer
some insight.

Best,
Frank

On May 1, 2009, at 11:01 AM, Martin Weiss wrote:

<>

You can find a general recipe for recovering the beta
coefficients here:

http://www.stata.com/statalist/archive/2009-03/msg00154.html

As I have never used -mim-,  I am highly reluctant to
endorse any calculation in connection with it. AFAIK,
P.
Royston, its author, does not post to the list
regularly, so
he is unlikely to comment, either. But there are
probably
more experienced listers who can step in...

Note that my  -mata- code was way too complicated as
you
have now revealed that you are a beginner. Sorry about
that...

HTH
Martin
_______________________

*
*   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