More about Meta-Analysis

July 2, 2015

I read “Another Five Things to Know about Meta-Analysis.” The write up is a follow on to an earlier article called “5 Key Things to Know about Meta-Analysis.” The use of “5” in one headline and “five” in another is not my typing.

The points are interesting. I want to highlight three.

First, the article points out that meta-analyses are not more reliable than a single study. The notion that averaging five studies’ data can yield wonky results reminds me of comments made to me by my teachers. Basics are good.

Second, studies go out of date. Therefore, the analysis may not be timely. Make decisions on stale data and excitement can ensue. Another useful point.

Third, analyses which are not inclusive of germane studies can be misleading.

The write up is a reminder that old fashioned research, data analysis, and research planning are needed to make fancy math work as expected. My view is that most of the whippersnappers just want answers. Work is not sitting alone and figuring out what to do with what set of data and how best to move forward to get a result that is sort of useful.

Who has time for that?

Stephen E Arnold, July 2, 2015

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta