i am currently working on a meta-analysis, and i have several studies which has outcomes of effect size and SD for different experiment protocol analysed on the same participant. for example squat jump , yo-yo , counter-movement jump and repeated sprint analysed in the same study on the same participants. how do i include such a study in a meta-analysis, do i need some form of coding?
What is your research question? Do all of these outcomes address your research question? If not, ignore the ones that don't. If two or more outcomes do address your research question, you will calculate an effect size and variance for each outcome. Assuming these outcomes were measured on the same subjects, the effect sizes and variances will be combined so that one effect size and one variance will represent that study. How you combine the effect sizes and variances depends on the information available to you. If you have individual subject data and thus can determine the correlations among outcomes, you will calculate the combined effect size and variance as in Chapter 24 of "Introduction to Meta-Analysis" by Borenstein et al. If you don't have the correlations, you simply average the effect sizes together as well as the variances. That is not ideal but is the best you can do.
Just as Gordon asked, your research question should answer that. A primary outcome (the most important to answer your question) should be chosen before searching.
As stated in the Cochrane book of systematic reviews, you should only include a population once per meta-analysis. This means that you do not combine multiple outcomes from a single study into one meta-analysis, but you can make more analyses where the study population provide different outcomes in each analysis. Including the same population multiple times in a meta-analysis will skew the outcome and is not good research practice.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis