\label{preliminary-results}
Table 2reports the results from a series of OLS regression models where
we study the association between the proportion aged 25+ with college
degree or more at the census block group level and educational
attainment at Wave IV. Model 1 reports the results of a regression model
without interaction term. Model 2 adds parental education in the
estimating equation and Model 3 includes both parental education and
interaction between PGS and contextual variables. Results indicate that
the polygenic score is positively associated with education, with 1SD
associated with 0.4 years of education (Model 1). Educational level at
the neighborhood level is also positively associated with individual
educational attainment. An increase of 10% in the proportion of adults
with college education is associated with 0.5 years of education at the
individual level (Model 1). To control for selection problems, we
include parental education as covariate in Model 2, bearing in mind that
the correlation between respondents and parental PGS is on average 0.5
because biological parents share on average 50% of genetic variation.
Controlling for parental education reduces substantially the coefficient
of education at the block level, but has a limited effect on PGS.
Lastly, we included the interaction between PGS and proportion of
college educated adults. Results indicate a positive significant
interaction between the two variables, while the main effect of PGS
reduces becomes not statistically different from zero.