Although Meta Analysis has outgrown as one of the most popularly used data analysis method, yet there are some issues related with the same. In this blog of mine, I would be dealing with these issues which are being faced during undertaking dissertation data analysis using the Meta-analysis concept.
So, here an explanation of the serious issues related with Meta-analysis:-
- Choosing between different models of error – An important decision you will make when conducting a meta-analysis involves whether a fixed-effect or random-effect model of error should be used to calculate the variability in effect size estimates averaged across studies. In many cases, it may be most appropriate to treat studies as randomly sampled from a population of all studies. In a random-effect model, study-level variance is assumed to be present as an additional source of random influence. Calculating random-effect estimates of the mean effect size, confidence intervals, homogeneity statistics and moderator analyses is complex and involves a two-stage process where first the meta-analysts must estimate the between-study variance and then this estimate is added to the variance of each study before the set of studies is combined statistically.
- Combining slopes form multiple regressions – Meta-analyses of regression coefficients are difficult to conduct. Measures differ across studies and regression models are diverse in terms of which additional variables are included in them.
- Considering multiple moderators simultaneously or sequentially – Homogeneity statistics can become unreliable and difficult to interpret when the meta-analysts wish to test more than one moderator of effect sizes at a time.
After knowing about the issues related to meta-analyses, I am sure you would have definitely become more alert and aware in using this means of analyzing the huge chunks of data that you’ve collected for your research document.