Tools for Those Who Summarize the Evidence Base
Hi there, I was introduced to this network several years ago, so I'm not sure if it's still active, but will put a question out there anyway, as I appreciate any suggestions or help.
My colleague and I are planning to conduct a meta-analysis of correlations using the Hunter and Schmidt method, correcting for attenuation. We have found relevant correlations in some studies, but they are reported at different levels of analysis (We are interested in the individual level). For example, while the full study recruited 850 participants (and reliability coefficients are reported at the individual level, for n = 850), the correlations between the constructs of interest are reported at the group-level, such as n = 98 (at the team level for 98 teams) and again at n = 18 (organisational unit level for 18 units) – using the same variables. We're wondering whether since the reliability coefficients are reported at the individual level, there is a way to use this to correct for group-level correlations. Or should these studies just be excluded? Hunter and Schmidt (2004) mention this issue on p. 99, but do not provide a way around it.