Skip to main content
. Author manuscript; available in PMC: 2022 Feb 1.
Published in final edited form as: Neurosci Biobehav Rev. 2020 Dec 25;121:291–306. doi: 10.1016/j.neubiorev.2020.12.020

Table 3.

Meta-analytic results.

Outcome k ES g [95% CI] p τ2 Q I2
Brain
Neural Effect During Training Taska 11 15 .59 [.44, .75] <.0001 0 8.07 0%
Neural Effect During Transfer Taska 8 8 .84 [.37, 1.31] .005 .09 10.51 31.51%
Behavior
Symptomsb 17 62 .37 [.16, .58] .002 .09 75.16 36.48%
Cognition 4 18 .23 [−.33, .78] .288 .21 37.19* 54.43%
RDoC Negative Valence Systemsb 12 37 .41 [.15, .68] .006 .09 38.60 35.51%
RDoC Positive Valence Systemsa 8 9 .13 [−.42, .67] .576 .11 11.62 39.76%
RDoC Cognitive Systems 5 20 .22 [−.28, .72] .289 .14 32.77* 42.82%
RDoC Social Processes 3 8 .02 [−.44, .48] .871 .05 10.20 30.34%
RDoC Sensorimotor Systems 2 3 .64 [.39, .88] .020 0 .74 0%

Note. Statistics are from models in which influential outliers were removed. k= number of studies, ES=number of effect sizes.

a

N= 1 influential outlier identified and removed. Hedges’ g and model statistics are reported without this influential outlier.

b

Data were fit with a three-level model. The sum of the variance components (across levels 2 and 3; i.e., σ2), and the total proportion of variance attributable to heterogeneity in the true effects is provided in the τ2 and I2 columns, respectively.

*

p<.05

**

p<.01