1447: Meta-Analysis

Explain xkcd: It's 'cause you're dumb.
Revision as of 07:12, 14 November 2014 by PLL (talk | contribs) (Added references section (not sure if correctly placed for this Wiki's conventions))
Jump to: navigation, search
Meta-Analysis
Life goal #29 is to get enough of them rejected that I can publish a comparative analysis of the rejection letters.
Title text: Life goal #29 is to get enough of them rejected that I can publish a comparative analysis of the rejection letters.

Explanation

In the scientific literature, meta-analyses are studies which compare multiple studies on a single topic, with the aim of giving a balanced overview of the known results.

This comic explores the idea of iterating the process, going from meta-analyses to meta-meta-analyses (which actually exist, though not necessarily by that name) and hence to a meta-meta-analysis.

Of the papers cited as meta-meta-analyses, at least A.R. Lemeshow (2005)[1] is real; the others presumably are too, but since their authors’ names are less distinctive, they are difficult to identify without full bibliographic data.

Transcript

Ambox notice.png This transcript is incomplete. Please help editing it! Thanks.
  1. A.R. Lemeshow et al, Searching one or two databases was insufficient for meta-analysis of observational studies., J. Clim. Epidemiol, 2005


comment.png add a comment! ⋅ comment.png add a topic (use sparingly)! ⋅ Icons-mini-action refresh blue.gif refresh comments!

Discussion

What is this "Medline, Embase and Cochrane" ? Thanks. Dams. 108.162.254.30 (talk) (please sign your comments with ~~~~)

Databases of Medical Publications 108.162.254.30 08:19, 14 November 2014 (UTC)
Medline[1], Embase[2] and Cochrane[3] are medical research databases. You can find there studies on various drug uses or treatment plans. A useful information source if you want to compare studies on use of Allopurinol for chronic gout or else. Jkotek (talk) 08:28, 14 November 2014 (UTC)
Seems to me that should be in the explanation, no? -- Brettpeirce (talk) 15:03, 14 November 2014 (UTC)

Would it be helpful to include a breakdown of the terms?

  • Meta-Analysis = "We searched M, E, & C for [keyword]"
  • Meta-Meta-Analysis = "We searched M, E, & C for 'We searched M, E, & C for [keyword]'"
  • Meta-Meta-Meta-Analysis = "We searched M, E, & C for 'We searched M, E, & C for 'We searched M, E, & C for [keyword]""

Smperron (talk) 08:46, 14 November 2014 (UTC)

Your paper is rejected. //"TOO META"--Theme (talk) 08:49, 14 November 2014 (UTC)

IMHO a better would be this:

  • Meta-Analysis = "We searched M, E, & C for [keyword] and compared results between each other"
  • Meta-Meta-Analysis = "We analyzed how others 'search the M, E, & C for for [keyword] and compare the results'"
  • Meta-Meta-Meta-Analysis = "We analyzed how other 'analyze how others search the M, E, & C for for [keyword] and compare the results'"

too meta Jkotek (talk) 15:22, 14 November 2014 (UTC)

I don't understand this explanation! Why isn't there an "Explain Explain XKCD" site? --RenniePet (talk) 23:43, 15 November 2014 (UTC)

I don't understand this comment! We need a Explain XKCD comment comments section!--108.162.231.199 09:51, 19 November 2014 (UTC)

It seems that Randall actually enjoys rejections. 173.245.50.139 03:18, 18 November 2014 (UTC)

The term "too meta" means that something is too self referential. Not "so abstract that they can't be easily interpreted." 108.162.254.88 (talk) (please sign your comments with ~~~~)

Actually, if we stick to the greek root, "meta" simply signifies "beyond", or "after". So, by adding level after level of analysis, by going beyond single studies, and single studies of single studies, it is truly becoming "too meta", too grand. 108.162.229.207 21:17, 31 March 2015 (UTC)
My interpretation is that it could mean both. For example, "too meta" = "the strenght of meta is too high" for being too self referential or "too meta" = "there's too many metas" for the excess or abstraction. I'm now experiencing Semantic satiation. 188.114.99.189 18:43, 1 December 2015 (UTC)

How much paper is there?

If the paper is a meta-meta-meta-analysis, then the paper is discussing the paper is discussing the paper? Is the Methods section (near-)infinite? How much paper does Randall have? 108.162.246.164 17:22, 14 September 2022 (UTC)

Papers are mostly paperless these days. 172.69.34.10 20:34, 14 September 2022 (UTC)
Maybe he's a quitter. L-Space Traveler (talk) 14:17, 20 March 2023 (UTC)

Meta-meta-…-meta-analysis

Some meta-analyses may use other permutations (like "We searched Embase, Cochrane and Medline for…"). Since there are 3 databases, the number of permutations is 6 for each iteration. All iterations are independent from the others, so there are 36 possibilities in a meta-meta-meta-analysis, and 46,656 in a meta-meta-meta-meta-meta-meta-meta-analysis.

Since the number of possibilities increases exponentially, a meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-analysis would have over 1014 permutations, much more than what can be handled by hand, a meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-analysis has over 3.6 · 1022, and a meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-meta-analysis has about as many possibilites as the estimated number of elementary particles (although I found no meta-meta-meta-meta-analysis or beyond on the web and only 1 study said to be a meta-meta-meta-analysis). 172.70.42.25 18:14, 22 May 2024 (UTC)