Sometimes we see I2(heterogeneity) is nearly 80-90%.
Which method we should use(random effect model or fixed effect model) how we will interpret the result In meta-analysis?
Dear Sumaiya Zabin Eusuf Zai if heterogeneity is too high you should consider if you are carrying on a meta-analysis or you will present your results narratively only, providing a rationale for that based on the heterogeneity level.
The choice between the random and fixed effects model should always be about whether there is any methodological heterogeneity, regardless of the level of statistical heterogeneity. The fixed effects model is based upon the assumption that each study could just as well be one large study.
https://www.meta-analysis.com/downloads/Meta-analysis%20Fixed-effect%20vs%20Random-effects%20models.pdf
I suggest that you present your data using a random effect model. I believe that for clinical studies, any data collected by a licensed practitioner should be included in the meta-analysis. Quality assessment will document the overall quality of evidence obtained from the meta-analysis anyway. Sumaiya Zabin Eusuf Zai Please find a link to one of my published meta-analyses. Feels free to ask if you have any question.
https://rdcu.be/c4CW4
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis