Hello wonderful people, I decided that this week I would discuss meta-analyses as I think they are a very useful tool to use in research methods. They basically involve collating results from many different studies, in order to look at the bigger picture and gain better knowledge. Whilst being useful though, there are also many pitfalls and areas that require specific attention to ensure that meta-analyses are accurate.
This study is a nice example of a meta-analysis, and it looked at the results of 12 different psychological intervention studies into paediatric oncology. After all of the 12 studies results were examined, it was found that psychological interventions play an important role in a number of ways, e.g. reducing stress. But why conduct a meta-analysis? Surely if each of the 12 studies showed significant results then a meta-analysis is unnecessary. Well this is the good thing about meta-analyses, they allow you to look at a very broad range of literature, so you aren’t just taking one study as fact; you are actually looking at that study in relation to other studies. This describes excellently how a meta-analysis can be used to improve the power of small or inconclusive studies. This means that you can use more of the research available to you, as before you may have only focussed on the results of the bigger studies. So this ability to access a greater range of research enables you gain a more comprehensive understanding of the topic area. Another page also highlights a number of the advantages of meta-analyses; you are able to generalise the results to a larger population. Rather than just looking at the population studied in one piece of research, you can look at the effects that many different studies had into many different populations. So there are many positives to meta-analyses (I’m not going to list them all because I’m sure you know them/ know where to find them, and there aren’t enough hours in the day), however there are some serious issues that need to be addressed.
The two pages linked above contain a number of disadvantages. For example, if one of the original studies was conducted poorly, or had bad statistics, your meta-analysis has no way to improve this, and so you have just used some research that may be bad. One of the biggest issues with meta-analyses is of course Publication Bias. This study explains how publication bias is basically the tendency for people to only publish studies that show significant results. Being the brilliant scientific minds that we are, we of course know that not significant results have the ability to tell us just as much about the study area than significant results, however these are far more likely to not get published. This means that any meta-analysis conducted may be missing a huge portion of unpublished studies that have the potential to produce wildly different results to the ones presented.
The area of publication bias poses a huge threat to meta-analyses and highlights the point that whilst meta-analyses are important, very strict care must be taken to ensure that results are accurate, and even then, a great deal of care must be taken in interpreting the results. And this study suggests that maybe people conducting meta-analyses are not following the proper procedures, which is having a huge effect on the results found. This shows that for meta-analysis to be considered good, they need to be undertaken with extreme care, however when they are, they are very useful indeed.
tallesttales 