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 at the end of May, and its replacement, statalist.org is already up and running.


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

RE: st: metan question,


From   "Tiago V. Pereira" <tiago.pereira@mbe.bio.br>
To   statalist@hsphsun2.harvard.edu
Subject   RE: st: metan question,
Date   Thu, 9 Sep 2010 02:47:01 -0300 (BRT)

Michael,

I believe you need further information on the data you have at hand:
either the t statistic or a correly reported P-value, say, P = 0.052.
Once you have one of these pieces of information, you can compute an
approximated value for the variance for your effect size (the difference
between "after" and "before").  See the formulations in classic
statistical books.

Denote the effect size as ES, the standard deviation as SD and the sample
size as N.

So, you end up with an N1, ES1 and SD1 for group 1 and N2, ES2 and SD2 for
group 2, the exact information required by -metan-.

If you don't have these information, I believe you have two strategies:

Assuming you have the following data for group 1:

N1 = number of subjects (identical in the "before" and "after")
m1 = mean before intervention
m2 = mean after intervention
sd1 = standard deviation before intervention
sd2 = standard deviation after intervention


You can compute the effect size "after vs before" as:


ES1 = m2-m1

And an approximate standard deviation for ES1:

SD1 = sqrt(((sd1^2)/(N1))+((sd2^2)/(N1))-(2*sd1*sd2*rho)/N1)*(sqrt(N1))

where rho is the correlation between observations from m2 and m1.

However, one does not know the correlation from summary data.  hence, your
two strategies are:

(1) set rho = 0, and ignore the correlation. This reduces the power of the
meta-analysis when rho>0.

(2) set several values of rho and check how results change. these values
can be obtained from some studies that you have raw data, or from
literature.

Let me know if that works for you.

Cheers!

Tiago






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

Michael wrote:

ello,
I notice that -metan- requires the following for continuous data:
	sample size,
	mean difference, and
	standard deviation (SD)

of the experimental group followed by those of the control group. This
could be used, for example, to pool data such as mean difference in blood
pressure before and after an intervention.

However, if instead of the sample size, mean difference and standard
deviation, I had for both groups:

	sample size,
	mean blood pressure before intervention
	standard deviation (SD)

	sample size,
	mean blood pressure after intervention
	standard deviation (SD)
would it be possible to convert into the format required by -metan-?


Best wishes,

Michael McCulloch, LAc MPH PhD
Pine Street Foundation
124 Pine Street
San Anselmo, CA 94960-2674
tel:	415-407-1357
fax: 	206-338-2391





*
*   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   |   Site index