Skip to main content
Fig. 9. | BMC Biology

Fig. 9.

From: Publication bias impacts on effect size, statistical power, and magnitude (Type M) and sign (Type S) errors in ecology and evolutionary biology

Fig. 9.

Ecological and evolutionary studies’ median Type S error rates (sign error) in detecting ‘true’ effects that were approximated by meta-analytic mean effect size estimates (labels: Meta-analysis, Sampling) and their bias-corrected versions (labels: cMeta-analysis, cSampling). On the y-axis, effect size metrics with different subscripts represent different individual meta-analyses (see Fig. 2). Sampling = statistical power at sampling level (primary studies). See more details in the legend of Figs. 3 and 8. All figures were drawn via geom_tile() function in ggplot2 R package (version 2.0) [61]

Back to article page