I am conducting a meta-analysis, and there are some twin studies that came up in my literature search that are relevant to my research question. However, including studies with samples of twins would violate the assumption of independent observations. I am considering contacting authors to ask them to provide effect sizes after only using one member of each twin pair. Is this a reasonable approach? Are there other methods to consider?
Hi Katerina,
By twin I assume you referring to trials conducts by same authors using same population size. If that's the case even if they are relevant to your to objective of your review I suggest you choose one maybe choose the most recent one as including them both will be duplication of same trial.
By indepedent observations it (generally) means that the same individual is not represented several times in the analysis. Since a twin is an individual subject like all others, the assumption is not violated. However, it is a generally interesting discussion, as it may always be questioned whether observations are independent from one another. Siblings could also be included in your studies, which may have as much in common as twins, depending on the research topic. To adress the interdependence between group members, multilevel models may be conducted, which take this into account, e.g. studies on classroom student in which the individual pupil represent one level in a hierarchical model, classes represent another level, and schools represent a third level. If it is genetic research however (out of my research area), I suppose that monzygotic twins would NOT be regarded as independent observations.
Hi Katerina
I assume that twin study means study where in sample was pair of twin.
Here, you should consider twin as independent sample unit.
Twins may be considered as small cluster. And you can include it as cluster randomised trial.
As a researcher, I think it is relevant to tell us what your research question is. What is your analysis of?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis