Meta analysis: a question

S

Hi guys! I have been set the task to perform a meta analysis (especially looking at the effect size of each study)on 50 studies. Does anyone know how long this will take?

A

As far as I know (not got masses of experience but have undertaken a meta-analysis course), the actual analysis is relatively quick. The lengthy part is identifying and capturing all the relevant literature and wading through it to ensure that studies meets the criteria you require. However, it sounds as though somebody has already done that part for you?

S

Hi Ann. Thanks for your reply. No, I am doing this by myself at the moment. I havefound the relevant articles but I am not sure how to calculate the effect size and which studies I can actually use. It takes me quite a lotas it is my second week of the PhD, have never done something like that before and I am unfamiliar yet with the subject. I was given a week period to finish it; I find it impossible and I am very stressed to be honest. Any advice?

W

This seems too short to do a meta-analysis of 50 studies...surely that would take months not weeks??? What field are you in?

S

I am in the department of Psychology; my project deals with financial decision making.

You think 50 is not a lot? What experience of meta analysis do you have? Any advice would be helpful as I feel completely lost doing this project.

A

You need a friendly statistician to help you out! Most Unis will have some kind of facility for getting staff/postgrads statistical advice, ask the postgrad office or maybe your supervisor?

Meta-analyses are complex, especially the literature search. If you want it to be publishable, then you will need to ensure that your search is as thorough as possible and that you have dealt with (or attempted to deal with!) such things as file-drawer bias/publication bias i.e. negative results tend not to be published and hence, if your search only reveals published studies (which tend to be positive) this can bias your results. Ways to check for unpublished studies are to search the Index of Theses and to contact researchers in the field to ask if they have unpublished work.

A

It is generally advised to have more than one person checking the studies that are found to see if they meet the required criteria - this is very important for studies that may be borderline.

I don't think it's realistic for someone to expect you to do this on your own with no guidance, if you have never done one before or had any training.

A

Re: the 50 studies

With a meta-analysis, it's not (usually) a question of pre-determining a set number of studies to analyse. If it is the type of meta-analysis also known as a 'systematic review' then you will need to look at ALL the studies that have been performed that investigate the topic that you are interested in. Ask your supervisor to clarify this for you.

S

Yes, I definitely need help here! It seems to be much harder than I thought! I just felt insecure to ask my supervisor so early on and show the level I am in Thank you for your help Ann!

A

You're welcome, good luck with your analysis!

S

Hi Scamp,
Meta-analyses are very difficult to perform, often hampered by study heterogeneity and publication bias. Have you used the Cochrane Reviewer's Handbook, free from Wiley Interscience? It talks at length about how to perform a meta-analysis and the issues concerning such a process. You want the reviewers handbook, and it's completely free! Hope this helps.

S

Thank you vey much Stevo69. That was really helpful! This work seems really difficult. And it is only my third week. I feel completely out of my depth. Not very happy

5215