Mathias Harrer
LightNews LightNews
banner
mathiasharrer.bsky.social
Mathias Harrer
@mathiasharrer.bsky.social
1.1K followers 440 following 14 posts
🎓 Researcher @ TU Munich & VU Amsterdam
🔍 Meta-Analysis, Statistical Software, Mental Health Research
🧑🏻‍💻 Maintainer: {dmetar}, {metapsyData}, {metapsyTools} in #rstats
🌎 mharrer.dev | metapsy.org
🦣 fediscience.org/@mathiasharrer
Posts Replies Media Videos
mathiasharrer.bsky.social
@eikofried.bsky.social @ploederl.bsky.social FYI, as promised :)
March 14, 2025 at 9:14 AM
mathiasharrer.bsky.social
Great list @floriannaudet.bsky.social! Is it possible to be added?
November 22, 2024 at 12:36 PM
mathiasharrer.bsky.social
This correlation can also have quite a big impact in meta-analysis, because some use change scores, others endpoint scores for their SMD (latter is preferable IMO).

We did a study calculating divergences between the two in PT & ADM trials depending on the "imputed" correlation. (under review).
November 20, 2024 at 8:39 AM
mathiasharrer.bsky.social
This is a common limitation. At least in psychotherapy trials, those pre-post correlations are hardly ever reported.

Edoardo Ostinelli & team extracted pre-post correlations from IPD in this paper onlinelibrary.wiley.com/doi/full/10.... (supply. 3.0.0).

Median for ADM was ~.3, ~.4 for iCBT
Combining endpoint and change data did not affect the summary standardised mean difference in pairwise and network meta‐analyses: An empirical study in depression
When studies use different scales to measure continuous outcomes, standardised mean differences (SMD) are required to meta-analyse the data. However, outcomes are often reported as endpoint or change...
onlinelibrary.wiley.com
November 20, 2024 at 8:33 AM
mathiasharrer.bsky.social
From Cohen's "The Earth is Round". The paper is famous, and I wonder why that particular part often gets neglected.
bookdown.org/MathiasHarre...
The major problem with correlations applied to research data is that they can not provide useful information on causal strength because they change with the degree of variability of the variables they relate. Causality operates on single instances, not on populations whose members vary. The effect of A on B for me can hardly depend on whether I'm in a group that varies greatly in A or another that does not vary at all. It is not an accident that causal modeling proceeds with regression and not correlation coefficients. In the same vein, I should note that standardized effect size measures, such as d and / developed in power analysis (Cohen, 1988) are, like correlations, also dependent on population variability of the dependent variable and are properly used only when that fact is kept in mind .
October 30, 2023 at 8:59 AM