Hi,
I am working on a meta-analysis about the impact of parent-child relationship. I have a few longitudinal research papers reporting the data for two or more waves. I don't think using data for all timelines is a good idea because it might violate independency. I am thinking of (1) choosing a wave with the most consistent age with other studies I coded or (2) choosing Time 1 data for every longitudinal study as there might be something(intervention/retest effect/...) that affected response from Time 2. Is there any guideline for longitudinal data in meta-analysis? Any suggestion?
Thanks!
Hi Jisoo Youn,
An option would be to do the longitudinal meta-analysis within the general linear mixed model framework (i.e. regression type models often fitted to longitudinal data).
I'd recommend this article: Musekiwa, A., Manda, S.O., Mwambi, H.G. and Chen, D.G., 2016. Meta-analysis of effect sizes reported at multiple time points using general linear mixed model. PloS one, 11(10), p.e0164898.
Your idea to perform separate univariate meta-analysis for 1 time point for each study each is an approach sometimes used... the problem with performing separate univariate meta-analysis is that it ignores correlation between the effect sizes at different points... Would be a shame to not make use of longitudinal data when you have it.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis