We computed effect sizes and conducted the metaanalysis in the Comprehensive
Meta-Analysis software package (45). All reported P values are
two-tailed, unless noted.
We used a random effects model (46, 47) to compare effect sizes. The
random effect size model was appropriate because conditions that could
affect learning gains varied among studies in the analysis, including the (i)
type (e.g., PBL versus clickers), intensity (percentage of class time devoted to
constructivist activities), and implementation (e.g., graded or ungraded) of
active learning; (ii) student population; (iii) course level and discipline; and
(iv) type, cognitive level, and timing—relative to the active learning exercise—
of examinations or other assessments.
We calculated effect sizes as (i) the weighted standardized mean difference
as Hedges’ g (48) for data on examination scores, and (ii) the log-odds
for data on failure rates. For ease of interpretation, we then converted logodds
values to odds ratio, risk ratio, or relative risk (49).
To evaluate the influence of publication bias on the results, we assessed
funnel plots visually (50) and statistically (51), applied Duval and Tweedie’s
trim and fill method (51), and calculated fail-safe Ns (45).