14

A (non-statistician) colleague has been encountering meta-analyses in papers he reviews for medical journals and is looking for a good introductory level treatment so he can educate himself. Any recommendations? Favorites? Books, monographs, nontechnical survey articles would all be fine.

(Yes, he's familiar with the Wikipedia entry and other stuff readily accessible by a Google search, such as Jerry Dallal's nice little article.)

Shane
  • 11,961
  • 17
  • 71
  • 89
whuber
  • 281,159
  • 54
  • 637
  • 1,101
  • Thank you @shane for creating the meta-analysis tag: I wanted to but couldn't. – whuber Aug 20 '10 at 20:38
  • Thank you @propofol and @Jeromy Anglim for your excellent contributions: after skimming DeCoster and looking over the Amazon reviews of the books, I have passed them all along to my colleague. It is a real shame one cannot checkmark multiple answers; in this case I literally flipped a coin: tails, it went to propofol. – whuber Aug 24 '10 at 22:29
  • 1
    http://stats.stackexchange.com/questions/how-to-ask "avoid asking for opinions or open-ended discussion" see, I can find you are doing similar thing or not following the rules. – ilhan May 20 '14 at 20:08
  • 1
    @ilhan: therefore you (and presumably everyone else, in the name of "fairness") should be allowed to break them? – Nick Stauner May 20 '14 at 20:29
  • @NickStauner, no, but he can go back and close his question. Or you can close. Since this an "opinion question". – ilhan May 20 '14 at 20:34
  • "Answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise," is false, so it is not off-topic, and I have voted to invalidate your flag. – Nick Stauner May 20 '14 at 20:37
  • 4
    @Nick Thanks for jumping in. However, you--and all other readers who chance across these comments--might like to know ilhan is not really concerned about this issue but is only using it to widen a personal attack being carried out at http://meta.stats.stackexchange.com/q/2064 and in comments at http://stats.stackexchange.com/q/99376. Because it has become evident that I--as the moderator who voted to close his question--lack any power of persuasion in this matter, I am standing aside and trusting that time and gentle advice from this community will soothe whatever feelings may have been hurt. – whuber May 20 '14 at 20:45

4 Answers4

11

I have two suggestions:

  1. Systematic Reviews in Health Care: Meta-Analysis in Context (Amazon link)
  2. Introduction to Meta-Analysis (Statistics in Practice) (Amazon link)

Both books are very good, including introductory information as well as detailed information about how to actually perform meta-analyses.

pmgjones
  • 5,543
  • 8
  • 36
  • 36
  • I liked Systematic Reviews in Health Care, though I found it - as I often do with books with a clinical focus, that it was both needlessly critical of observational studies, and a bit blind to the weaknesses of poorly run clinical trials. It was helpful, but it requires some adaptation to be useful outside the RCT world. – Fomite Oct 16 '11 at 09:02
10

I'll add an independent recommendation for Jeromy's blog post, and second the suggestions of James DeCoster's notes and the Borenstein textbook (propofols' no. 2).

At risk of indulging in self-promotion, I recently published a methods paper entitled Getting Started with Meta-analysis. It's aimed at ecologists and evolutionary biologists, so the examples are taken from these fields, but I hope it will be useful for those working in other areas.

Freya Harrison
  • 3,212
  • 4
  • 25
  • 31
7

I wrote a post a while back on getting started with meta analysis with: (a) tips on getting started, (b) links to online introductory texts, and (c) links to free software for meta analysis.

Specifically, you might want to read James DeCoster's notes.

Jeromy Anglim
  • 42,044
  • 23
  • 146
  • 250
2

Fredric M. Wolf's little green Sage book is worth the $18 or so. "Pleasantly mathematical" but not too technical, not too dogmatic either (it's a fiercely contested field, you probably know), good for a person with what I'd call intermediate-level stats/research experience.

rolando2
  • 11,645
  • 1
  • 39
  • 60