*3.4. Quality Assessment*

Table 2 displays the results of the quality assessment, which identified two low quality studies, four neutral quality studies, and five high quality studies. The low-quality studies were published in 1995 and 2017 and represented small samples of children attending a residential school in Canada (N = 20) and an elementary school in Switzerland (N = 103), respectively [20,21]. These studies reported non-significant effect estimates for outcomes and had short duration of follow-up (≤3 years). The high-quality studies were published between 2001 and 2014, included large cohorts (>1000) from New Zealand, Norway, and The Netherlands, and a smaller cohort from the United States [22,25,26,28,29]. The effect measures for the high-quality studies were almost all significant (except one effect measure from a study that used the Youth Self Report & Child Behavior Checklist [29] and had greater length of follow up (range 5 to 29 years). The five high quality studies are summarized below:

• Moffitt, 2011 [26]

A prospective cohort study from the participants in the Dunedin Multidisciplinary Health and Development Study Cohort in New Zealand assessed childhood self-control, socioeconomic factors, and IQ using the Wechsler Intelligence Scales for Children, Revised (WISC-R; repeat measures at ages 3, 5, 7, 9, and 11), and the association with wealth at age 32. Statistical models included adjustment for socioeconomic factors and fixedeffects modeling applied to dizygotic same-gender twins to compare outcomes of siblings with differential self-control levels and thus isolate the effect of self-control. The study found that the intelligence assessment was significantly associated with four measures of wealth: socioeconomic status, income, financial planfulness, and financial issues (regression estimates −0.400, −0.291, −0.160, and 0.029, respectively; all *p* < 0.05).

• Sagatun, 2014 [28]

A retrospective cohort study that utilized data from a Norwegian registry to assess the association between the Strengths and Difficulties Questionnaire administered to 15 to 16-year-olds and academic attainment as recorded in the national registry of school completion at age 20–21. Statistical models included adjustments for children's ethnic background, county of residence, parents' education, income, and marital status. The study

found that this tool was significantly associated with odds of non-completion of school (ORs 1.11–1.48, all *p* < 0.001).

• Lamp, 2001 [22]

A prospective cohort study among families enrolled in the Head Start Program in the United States assessed intelligence using the Stanford Binet Intelligence scale at age 4 and its correlation with academic achievement at ages 5 to 10 years, measured by the Metropolitan Achievement Test. No information regarding the factors used for adjustment in statistical models was provided. The study found that intelligence as measured by this tool was significantly correlated with academic achievement (correlation coefficients 0.39–0.62, all *p* < 0.01).

• Fergusson, 2005 [25]

A retrospective cohort study involving participants from the Christchurch Child Development Study in New Zealand assessed intelligence using the WISC-R in 8 to 9 year-olds and analyzed its association with wealth, educational outcomes, and obtaining a university degree between the ages of 18 and 25 years. Statistical models included adjustment for a series of covariate factors including measures of childhood social and family disadvantage and behavior. The study found that intelligence was significantly associated with gross income (regression coefficient 1.595, *p* < 0.05) and gaining school or university qualifications (regression coefficients 0.67–0.82, *p* < 0.01).

• Veldman, 2014 [29]

A prospective cohort study to determine likelihood of educational attainment (measured by number of years of schooling completed) by age 19, using data from the Tracking Adolescent's Individual Lives Survey in The Netherlands, assessed 11 year-olds using the Child Behavior Checklist and its Youth Self Report. Statistical models included adjustment for children's sex, age, IQ, parental educational status, and physical health status. The study found that externalizing, internalizing, and attention problems, as assessed by these combined tools, were associated with higher odds of low (primary, lower vocational and lower secondary education) vs. medium (intermediate vocational and intermediate secondary) educational attainment at age 19 (OR 1.25–1.78; statistical significance varied—see Table S3 for details).


Assessment)

 for additional detail.
