The main result shows that family bereavement has a clinically significant, medium sized effect on the increase of internalizing problems within 2 years in comparison to non-bereaved adolescents while accounting for pre (loss) internalizing problems.
[30] At endpoint, we also observed medium sized effects for the differences between placebo and control groups on symptom severity (d = 0.53) and quality of life (d = 0.40).
Evidence for small and medium-sized effects of lower maternal education levels on increased risk for SSD was identified in the population cohort [ 59], the birth cohort [ 91] (although note the lack of evidence within a subset of this cohort [ 63]), and one of the case-control cohorts [ 45].
The standard deviation in each cell is about 6, corresponding to a medium sized effect size of 0.36.
Our previous genetic studies demonstrated that these lines differ in large part due to cumulative action of multiple quantitative trait loci (QTLs) of small to medium sized effects, confirming that these lines may present a valuable model of the polygenic form of leanness/obesity [ 17- 19].
Detectable differences for all measures were between 0.30 and 0.43 standard deviation units (small to medium sized effects), which translated to detectable differences of 0.62 minutes of MVPA per hour, 1.63 minutes of sedentary time per hour, and 0.33 ml/kg/min of energy expenditure, assuming a conservative estimate of measuring 25 children per preschool.
It is widely accepted that Cohen's d values of 0.2 – 0.49 denote small sized effects; 0.5 – 0.79 denote medium sized effects; and >0.8 denote large effect sizes.
Moreover, in a meta-analysis of two OCD studies representing the net contribution of exposure in vivo, we found moderate evidence of a medium-sized effect on work-related outcome.
In a 1992 meta-analysis, Cohen et al. found a "medium-sized effect" of computer-assisted instruction on student learning and recommended more research to identify specific features of computer-assisted instruction that lead to improved student performance [5].
In general, a statistical power to detect a medium-sized effect of 80%% is considered adequate [ 54].
Assuming a common standard deviation of 10 points for two-group comparisons yields a medium-sized effect of 0.5 (Cohen's d).
