Table 3. Effects of Removing Two Studies ([34] and Burton, unpublished) at High Risk of Bias from the Pooled Estimate of Generalized Stress-directed Training Program Effectiveness.

Outcome (number of studies)	Pooled Std. MeanDiff,Random Effects(95% CI)	P Value	I2	Absolute Change in Effect Size and New Interpretation
Resilience
With Burton (13) 0.37(0.18, 0.57) 0.0002 41%	0.37 (0.18, 0.57)	0.0002	41%	+0.04; suggests a highly significant, moderately consistent, and moderate effect on improving resilience
Without Burton (12) 0.41 (0.20, 0.61) <0.0001 40%	0.41 (0.20, 0.61)	<0.0001	40%
Quality of Life
With Abbott (4)	0.34 (−0.03, 0.72)	0.07	10%	+0.28; suggests a significant, highly consistent, and moderate effect on improving quality of life
Without Abbott (3)	0.62 (0.14, 1.09)	0.01	0%
Depression
With Abbott/Burton (6)	−0.28 (−0.56, 0.01)	0.06	33%	−0.23; suggests a highly significant, highly consistent, and moderate effect on improving depression symptoms
Without Abbott/Burton (4)	−0.51 (−0.79, −0.22)	0.0005	0%
Stress
With Abbott/Burton (9)	−0.28 (−0.60, 0.04)	0.09	57%	−0.22; suggests a highly significant, highly consistent, and moderate effect on improving stress symptoms
Without Abbott/Burton (7)	−0.50 (−0.74, −0.26)	<0.0001	0%
Anxiety
With Abbott/Burton (5)	−0.11 (−0.41, 0.20)	0.48	17%	−0.26; suggests a borderline-significant, highly consistent, and small effect on improving anxiety symptoms
Without Abbott/Burton (3)	−0.37 (−0.75, 0.01)	0.06	0%