Thread by @ollie_lovell

1) I’ve read a bit recently critiquing the approach of the ‘meta-analysis’. So I’m trying to look into it. Stop one, Gene Glass’ 1976 paper ‘Primary, secondary, and meta-analysis of research’. Here’s what he’s saying. pdfs.semanticscholar.org/e990/a41e8f09e…

2) Meta-analyses are necessitated by the sheer volume of research that’s out there now (‘now’ being 1976). And then-current methods weren’t up to scratch. (but note how he himself suggests that combining 500 studies will ‘defy simple summary’)

3) As an aside, a pretty funny characterisation of those then-current methods.

4) Then he makes some really interesting comments regarding his thoughts about studies with design and analysis flaws. I feel like further discussion is needed to work out exactly what he means by this...

5) He moves on to then highlight the value/importance of summarising the research, and getting it out there.

6) Some interesting comments are made regarding how early meta-analyses were approached. Curious to know into what level of detail that recent meta-analyses go to in order to separate studies based upon design quality, significance of results, contextual factors, etc.

7) Now, here it starts to get interesting. By my reading, Glass suggests that it doesn’t make sense to compare studies that don’t have the same contexts, and, in order to compare A and B, settles on a meta-analytic approach that ONLY includes A,B,C studies (‘c’ being control).

8) Then he makes a funny comment.

9) Wraps up with a great word I've never heard before...

10) Short but very interesting paper. Provides an interesting historical context for the birth of the meta-analysis.