This article estimates for the Netherlands linear regression models for an individual’s level of education by bringing in parental as well as grandparental level of education. The prime question is that of the association between education in the first and third generation after
holding constant education in the second generation. For the empirical analysis, we used the five available waves of the Family Survey Dutch Population as conducted in 1992-1993, 1998, 2000, 2003 and 2009. Since both primary respondents and their partner were interviewed about their father,
mother and offspring, we could estimate statistical models in which the educational level of all grown-up children was regressed on the educational level of both parents and all four grandparents. The results show that grandfather’s level of education has a direct effect on his grandchild’s
level of education, controlled for the educational level of one of the parents. In a model with both mother’s and father’s education, no longer a direct effect of grandfather’s education is found. This suggests that the direct effect of grandfather’s education has become
indirect. Furthermore, the positive effect of grandfather’s education on his grandchild’s education declines as a function of parental education. In fact, only children with low educated parents profit from a high educated grandfather.