5

First post here; first I would like to say that I have no background in stats whatsoever and not so much in math either (enriched high school math, and that's it).

I started an MA and we have to do a meta-analysis; we've also been reading a lot of publications with effect sizes.

I checked Wikipedia's page for effect sizes; some parts are helpful, and for others I feel the gap when I'm reading, and it almost looks like gibberish. I've found some lectures on YouTube which I think will be useful. I was wondering if some of you would have suggestions for a good point to start learning about this process?

EDIT: I think I need to make my question more precise as I read the comments and when I skim the answers in the other questions, it still seems too high level sometimes. The effects sizes books seemed like it could help though, thanks for that!

I do understand mostly what a meta-analysis is (I think!) but it's how people go from the data to the numbers in their effect sizes that I am wondering about. Like how do effect sizes, chi-square work etc. When I read the definitions on these formulas, realize I would also need a course on distribution, regression, multivariate as I have no idea what these mean and it doesn't ring much of a bell. What I think helped the most with the chi-square is when I read something about comparing product reviews on Amazon where one product had 5-stars but one review and another product had 4.5 stars but 350 reviews.

If there are resources with worked cases for the type of formulas used in a meta-analysis, to understand what they're for and how they're used and with realistic examples so I can refer to something I know, I think that would greatly help. In fact, I need beginner resources in stats generally so I can build on that.

curious
  • 517
  • 1
  • 6
  • 12
  • 1
    Not sure what you mean by effect size. Usually I recommend (1) BMJ review article of meta-analysis (2) If OR is an issue similarly BM (http://www.bmj.com//content/320/7247/1468.1) (3) and software specific walkthrough example. (For stata: http://www.stata-journal.com/article.html?article=sbe24_2) – charles Feb 23 '14 at 00:49
  • 1
    possible duplicate of [Looking for good introductory treatment of meta-analysis](http://stats.stackexchange.com/questions/1963/looking-for-good-introductory-treatment-of-meta-analysis)...though I may have misread the extent to which this question concerns meta-analysis specifically. – Nick Stauner Feb 23 '14 at 01:52
  • 3
    [The Essential Guide to Effect Sizes: Statistical Power, Meta-Analysis, and the Interpretation of Research Results](http://www.amazon.com/Essential-Guide-Effect-Sizes-Interpretation/dp/0521142466) by Ellis is a good introduction book. The first half introduces the idea of effect sizes and how to report them, the second half concerns its application in meta-analysis. It wouldn't make you ready to immediately work on advanced cases, but will get you reasonably literate in the subject to dive deeper. – Penguin_Knight Feb 23 '14 at 02:28

2 Answers2

3

I highly recommend the following excellent papers on effect sizes in the meta-analysis context:

  • Paper "Calculating and reporting effect sizes to facilitate cumulative science: A practical primer for t-tests and ANOVAs" by Daniël Lakens (PDF) - doi:10.3389/fpsyg.2013.00863;

  • Paper "How to select, calculate, and interpret effect sizes" by Joseph Durlak (PDF) - doi:10.1093/jpepsy/jsp004.

Aleksandr Blekh
  • 7,867
  • 2
  • 27
  • 93
0

I'm just delving into Meta-analysis myself, and I've found this book very helpful:

Handbook of Meta-analysis in Ecology and Evolution Edited by Julia Koricheva, Jessica Gurevitch & Kerrie Mengersen

Paperback | 2013 | ISBN: 9780691137292 
Hardcover | 2013 |  ISBN: 9780691137285 
eBook | ISBN: 9781400846184 |

Thought it is intended for Ecology and Evolution, I think it provides a nice overview of meta-analysis philosophy, how it differs from systematic reviews and vote counting, and the direct steps you need to take to successfully carry out a meta-analysis. A large portion of the book is dedicated to choosing an effect size metric and the statistical implications of that.

emudrak
  • 278
  • 2
  • 11