Message-ID: <5dcdcf51b7574385b47d6a7bc047fe64@UM-MAIL3214.unimaas.nl>
Date: 2022-02-02T10:18:21Z
From: Wolfgang Viechtbauer
Subject: [R-meta] Meta-analysis of mean differences
In-Reply-To: <tencent_4048DBD9218F22FD97168DD50F0618513805@qq.com>
This would not account for the non-independence between the sampling errors of d or g values coming from the same sample at different time points.
Best,
Wolfgang
>-----Original Message-----
>From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces at r-project.org] On
>Behalf Of YA
>Sent: Wednesday, 02 February, 2022 5:34
>To: r-sig-meta-analysis
>Subject: Re: [R-meta] Meta-analysis of mean differences
>
>Hi Martin and others,
>
>What about the three level meta-analysis?
>
>Level 1: effect size between ctrl and exp group (ie, cohen's d or hedges' g),
>which are nested within
>
>Level 2: groups of the effect size (ie, pretest or posttest, or t1, t2, t3....),
>which are nested within
>
>Level 3: studies.
>
>In this way, you dont need to know the correlations between the pre and post test
>or among time points, the correlations are controlled for by involving a level 2,
>just like the control of the dependency in multilevel modeling.
>
>hope this helpful.
>
>YA
>
>------------------ Original ------------------
>From: "Martin Lobo" <mlobo4370 at hotmail.com>;
>Date: Tue, Feb 1, 2022 09:40 PM
>To: "Michael Dewey"<lists at dewey.myzen.co.uk>;"Viechtbauer, Wolfgang
>(SP)"<wolfgang.viechtbauer at maastrichtuniversity.nl>;"r-sig-meta-analysis at r-
>project.org"<r-sig-meta-analysis at r-project.org>;"Gerta
>Ruecker"<ruecker at imbi.uni-freiburg.de>;
>
>Subject: Re: [R-meta] Meta-analysis of mean differences
>
>Hello, thank you all.
>I don't quite understand how to test different sensitivity analyzes with
>different correlation coefficients. What had occurred to me was to use the values
>??after the intervention in the two groups and do a meta-regression using the
>baseline LDL level as a covariate, and thus adjust the result for it, but I don't
>know if that would be correct.
>In any case, I want to explain the case again, since out there it was interpreted
>as paired measurements and it is not.
>I have group A and Group B, in each group I have a baseline value of LDL and the
>value after the intervention, and I need to meta-analyze the impact on this
>variable. I usually do it already having the difference between baseline and
>post-treatment of the two groups with their deviations, but in this case I only
>have these data to make a difference of means. That's why I also asked how to do
>it with change scores, something like that had been explained to me by gerta but
>I didn't understand.
>Thanks a lot
>
>Lorenzo Mart?n Lobo MTSAC, FACC, FESC
>Especialista Jerarquizado en Cardiolog?a