Table 4. Standardized coefficients from educational attainment mediation model.
| Random Intercept Factors | ||||||
|---|---|---|---|---|---|---|
| GPA | Academic Motivation (Self/Parent) | Academic Motivation (Teacher) | Disciplinary Problems | Cigarettes per day | Educational Attainment | |
| Smoking PGS | ||||||
| PGS → RI | -.09 | -.13 | -.13 | .15 | .18 | -- |
| [-.13, -.05] | [-.17, -.09] | [-.17, -.09] | [.11, .19] | [.14, .22] | ||
| PGS → Educational Attainment | -- | -- | -- | -- | -- | -.07 |
| [-.11, -.04] | ||||||
| PGS → RI → Educational Attainment | -.03 | .00 | -.03 | -.01 | -.01 | -.08 |
| [-.04, -.01] | [-.01, .00] | [-.04, -.02] | [-.02, -.01] | [-.02, .00] | [-.10, -.05] | |
| Education PGS | ||||||
| PGS → RI | .22 | .09 | .19 | -.05 | -.10 | -- |
| [.18, .27] | [.05, .13] | [.15, .23] | [-.10, -.01] | [-.14, -.06] | ||
| PGS → Educational Attainment | -- | -- | -- | -- | -- | .11 |
| [.08, .15] | ||||||
| PGS → RI → Educational Attainment | .06 | .00 | .04 | .00 | -.01 | .12 |
| [.05, .08] | [.00, .01] | [.03, .05] | [.00, .01] | [.00, .01] | [.10, .14] | |
| Educational Attainment | ||||||
| RI → Educational Attainment | .28 | .03 | .21 | -.08 | -.05 | -- |
| [.24, .33] | [-.02, .07] | [.16, .26] | [-.12, -.04] | [-.09, -.02] | ||
Note. PGS = polygenic score; RI = random intercept; GPA = grade point average; PGS → RI = paths from the PGS to the random intercept; PGS → Educational Attainment = paths from the PGS to educational attainment (i.e., the direct effect); PGS → RI → Educational Attainment = indirect effects from the PGS to educational attainment through the corresponding RI (final entry in these rows correspond to the total indirect effect through all RIs); RI → Educational Attainment = paths from the RI to educational attainment; Bold = 95% confidence interval does not include 0. All paths come from a single model with one outcome variable (educational attainment), five intervening variables (the random intercepts), and 6 independent variables (the PGSs and covariates of sex and 5 genetic principal components to adjust for ancestry; coefficients for control variables not presented). In this model, the random intercepts were represented via random intercept factor scores from the initial unconditional random intercept models. Confidence intervals derived via clustered non-parametric percentile bootstrap with 1,000 draws. Educational attainment was adjusted for age (mean age 29.4 years, SD = 3.9 years, range 19.7 to 39.9 years). R2 = .36.