Table 3. Tests of Association Between Childhood Behavioral Ratings and Adult Earnings^a.

Variable	β	SE	P Value	95% CI	Effect Size
Male Participants (n = 1470)
Inattention	−1271.49	325.09	.001	−1908.67 to −634.30	3077
Hyperactivity	−542.41	593.02	.36	−1704.71 to 619.89	797
Aggression-opposition	−699.83	287.07	.01	−1262.49 to −137.17	1764
Anxiety	275.02	468.25	.56	−642.74 to 1192.77	418
Prosociality	476.75	150.62	.002	181.53 to 771.96	2050
IQ	1073.41	505.97	.03	77.28 to 2069.54	1771
Family adversity	−13,859.23	2825.57	.001	−19,397.54 to −8320.92	3603
Female Participants (n = 1380)
Inattention	−924.95	254.62	.001	−1424.44 to −425.46	1915
Hyperactivity	43.93	486.05	.93	−909.56 to 997.42	65
Aggression-opposition	−109.72	307.66	.72	−713.25 to 493.84	190
Anxiety	−261.52	362.98	.47	−973.57 to 450.55	361
Prosociality	166.15	112.52	.14	−54.58 to 386.88	758
IQ	916.84	456.22	.045	18.77 to 1814.90	1265
Family adversity	−16,048.02	1898.10	.001	−19,771.76 to −12,324.28	4172

Abbreviation: β, unstandardized beta coefficient.

^aMixed-effects linear regression models were used to examine the association between behaviors at age 6 years and adult earnings, controlling for child IQ and family background. A 1-unit change in behavioral scores at age 6 years (Table 1) was associated with a corresponding change in annual earnings (β) at age 33 to 35 years. Units for the aggression-opposition variable were mean (SD) = 2.22 (2.52) for male participants, and mean (SD) = 1.05 (1.73) for female participants. Continuous interactions between all variables were tested. Effect sizes represent the ratio of the coefficient estimate to the variable’s SD. Effect sizes (f²) for the overall models were 0.11 for male and 0.12 for female participants.