Message-ID: <704e977a-ba10-4fed-bee9-8b4fcc37844e@Spark>
Date: 2022-01-18T07:56:00Z
From: David Pedrosa
Subject: [R-meta] Question on effect sizes
Dear list members,
I currently have a comprehension question where I would like to ask for an assessment of the list. We are doing a meta-analysis where there are different expressions of outcomes that I am trying to combine using effect sizes.For the pre-post controlled tests, it looks something like this:
+------------+-------------------+-------------+-------------------+-------------+
| Study | Pre | | Post | |
| # | Mean | SD | Mean | SD |
| ===== = | ===========| =====================| =======|
| 1 | Mean_x_y1 | SD_x_y1 | Mean_x_y2 | SD_x_y2 |
| 2 | Mean_x1 | SD_x1 | Mean_x_x2 | SD_x2 |
| 2 | Mean_y1 | SD_y1 | Mean_y2 | SD_y2 |
+-------+----------------------+-------------------+---------------+-------------+
My questions would be, if it resonable to assume that the pooled SD of the second study can be somehow estimated (unfortunately there is neither pretest data available nor a correlation between x and y)?
And the other question would be a simple one which I could not find a definite answer for: How do I deal with studies indicating a mean change score, so how do I standardize Mean_change_x_y and SD_change_x_y in the scenario above when I don?t have a baseline score?
Best wishes,
[[alternative HTML version deleted]]