Exercise Session 3

Questions.

The data for this question are standardized mean differences from a meta-analysis of 49 experimental effects of teacher expectancy on pupil IQ (Raudenbush, 1984). They are part of the metaforpackage and you can find out more about them in the help file in the metafor package.

The typical study in this dataset administered a pre-test to a sample of students. Subsequently, the teachers were told that a randomly selected subsample were ‘bloomers’ (students with substantial potential intellectual growth). All of the students were then administered a post-test and it was expected that the ones identified as bloomers would show a significantly higher increment in IQ growth than the control group (i.e., students not identified as bloomers). You might have heard of the Pygmalion effect, (see here)

Inspect the data and note that it has yi= standardized mean difference and vi= corresponding sampling variance.

• Conduct a fixed-effects meta-analysis. Note that you’ll need to calculate the standard error.
• Conduct a random-effects model with the DerSimonian-Laird method.
• Conduct a random-effects model with the Hartung-Knapp-Sidik-Jonkman method.
• Conduct a random-effects model with Restricted Maximum Likelihood (REML) estimation, look at the meta package, if you don’t know how to.
• Make a ‘nice’ forest plot for one of your models. (Bonus: see if you manage to combine author and year on your forest plot).