Table 4.

Categorical and continuous moderator analyses.

Categorical moderator	k	ES—r	95% CI	z	p	Q bet, df (p)	I2
FTD measure a	14	−0.23	[−0.33, −0.12]	−3.99	<0.001	1.90, 1 (0.168)	75.77
Clinician-rated	9	−0.28	[−0.40, −0.14]	−3.97	<0.001		83.73
Behaviorally based	5	−0.13	[−0.31, 0.07]	−1.31	0.192		0.00
Social function measure b	17	−0.21	[−0.30, −0.12]	−4.34	<0.001	1.15, 2 (0.563)	72.63
Self-report	3	−0.25	[−0.46, −0.02]	−2.10	0.035		87.47
Clinician-rated	10	−0.21	[−0.33, −0.08]	−3.27	0.001		75.20
Performance-based	4	−0.20	[−0.38, −0.01]	−1.98	0.048		40.67
Study quality	13	−0.23	[−0.33, −0.12]	−3.88	<0.001	3.35, 1 (0.067)	77.25
High	6	−0.28	[−0.44, −0.11]	−3.16	0.002		76.44
Moderate	7	−0.19	[−0.33, −0.04]	−2.40	0.016		78.72
Continuous moderator	k	B	R 2	SE	95% CI	z	p	I2 c
%Female	12	−0.21	0.03	0.72	[−1.62, 1.20]	−0.29	0.77	78.03
Age	11	0.01	0.00	0.09	[−0.01, 0.03]	0.84	0.40	80.02
Total symptoms	7	−0.01	0.00	0.01	[−0.03, 0.02]	−0.34	0.74	70.84
Positive symptoms	9	−0.26	0.00	0.27	[−0.79, 0.27]	−0.95	0.34	82.42
Negative symptoms	9	0.07	0.00	0.18	[−0.28, 0.42]	0.42	0.68	82.42

The p-value of Q _bet indicates whether the subgroups are significantly different.

Abbreviations: B, regression coefficient; CI, confidence interval; ES, r effect size statistic; I², extent of between-study variability; k, number of studies used in the calculation of the mean effect size; p, two-tailed p-value associated with the test of statistical significance; Q _bet, variance between subgroups; R², R² analogue; SE, standard error, z, test for statistical significance of the mean effect size.

^aFor this analysis, one study [98] contributed effect sizes to both clinician-rated and behaviorally-based subgroups.

^bFor this analysis three studies contributed multiple effect sizes.

^cThe I ² value might be larger than that of the overall main analyses because not all of the studies were included in the moderator analyses.