Associations between polygenic index (PGI) of education and years of education (linear coefficients) and between PGI and educational transitions (odds ratios), with gender interactions: Linear (A), multinomial logistic (B and C), and binary logistic (D) regression models
| . | A: Years of Education . | B: Academic Secondary/Vocational Given Basic Degree . | C: Lower/Higher Tertiary Given Academic Secondary Degree . | D: Tertiary Given Vocational Degree . | ||
|---|---|---|---|---|---|---|
| . | . | Vocational . | Academic Secondary . | Lower Tertiary . | Higher Tertiary . | . |
| PGI of Education | 0.74*** | 1.40*** | 2.57*** | 0.97 | 1.51*** | 1.48*** |
| (0.71–0.78) | (1.34–1.46) | (2.44–2.71) | (0.89–1.06) | (1.38–1.66) | (1.40–1.56) | |
| Female | 0.36*** | 0.97 | 1.91*** | 1.42*** | 0.90† | 1.01 |
| (0.31–0.41) | (0.92–1.03) | (1.79–2.05) | (1.27–1.58) | (0.80–1.01) | (0.95–1.09) | |
| PGI × Female | −0.03 | 0.90*** | 0.87*** | 1.14* | 1.08 | 0.94† |
| (−0.08 to 0.02) | (0.85–0.96) | (0.81–0.93) | (1.02–1.28) | (0.96–1.21) | (0.87–1.01) | |
| N | 37,416 | 37,416 | 11,382 | 16,373 | ||
| . | A: Years of Education . | B: Academic Secondary/Vocational Given Basic Degree . | C: Lower/Higher Tertiary Given Academic Secondary Degree . | D: Tertiary Given Vocational Degree . | ||
|---|---|---|---|---|---|---|
| . | . | Vocational . | Academic Secondary . | Lower Tertiary . | Higher Tertiary . | . |
| PGI of Education | 0.74*** | 1.40*** | 2.57*** | 0.97 | 1.51*** | 1.48*** |
| (0.71–0.78) | (1.34–1.46) | (2.44–2.71) | (0.89–1.06) | (1.38–1.66) | (1.40–1.56) | |
| Female | 0.36*** | 0.97 | 1.91*** | 1.42*** | 0.90† | 1.01 |
| (0.31–0.41) | (0.92–1.03) | (1.79–2.05) | (1.27–1.58) | (0.80–1.01) | (0.95–1.09) | |
| PGI × Female | −0.03 | 0.90*** | 0.87*** | 1.14* | 1.08 | 0.94† |
| (−0.08 to 0.02) | (0.85–0.96) | (0.81–0.93) | (1.02–1.28) | (0.96–1.21) | (0.87–1.01) | |
| N | 37,416 | 37,416 | 11,382 | 16,373 | ||
Notes: All models include year of birth cubic splines, 10 first principal components of the genome, region of residence, and combination of data collection round and genotyping batch. Linear Model A reports conventional regression coefficients (with Huber–White standard errors), multinomial logistic Models B and C report multinomial odds ratios, and binary logistic Model D reports odds ratios. 95% confidence intervals are shown in parentheses.
†p < .10; *p < .05; ***p < .001