The question of whether all children, regardless of background, can reach their full potential is central to educational equity. Ability tracking, the sorting of students into different educational streams based on ability, is a common practice with varying implementation ages across countries. While proponents argue that tracking optimizes learning environments, concerns exist that it exacerbates inequalities based on family background, particularly when implemented early. Existing research often relies on parental socioeconomic status (SES) and IQ scores to assess family influence and potential ability, respectively. This approach is limited because it fails to capture the multifaceted nature of family influence and individual abilities. Furthermore, early ability assessments are themselves influenced by family background, making disentanglement difficult. This study employs a behavioral genetics approach using twin data to overcome these limitations. The classical twin design allows for the estimation of heritability (genetic influence) and shared environmental influence (family background influence) on educational attainment. The researchers hypothesize that delaying tracking to a later age increases the importance of potential ability (genetic influence) and reduces the importance of family background (shared environmental influence). This hypothesis is supported by existing literature suggesting that greater equality of opportunity leads to larger genetic and lower shared environmental influences on educational outcomes. However, the researchers acknowledge caveats, such as genetic influence potentially reflecting both positive and negative traits and the shared environment representing a composite of family influences. The study further explores the mechanisms through which tracking age affects inequality of opportunity by differentiating between primary and secondary effects of family background. Primary effects refer to the direct impact of family background on student performance, whereas secondary effects relate to the influence of family background on educational decisions even when performance is equal. The study examines whether early tracking amplifies secondary effects through either hampering (disadvantaged children tracked below their potential) or compensation (advantaged children tracked above their potential) mechanisms. The Netherlands, with its varying tracking ages across schools, provides an ideal context for this investigation. The researchers address potential confounding factors, such as school choice and parental influence on school selection, acknowledging that the choice of immediate versus delayed tracking may not be completely exogenous.

The literature review highlights the existing debate surrounding ability tracking and its impact on educational equity. Studies consistently demonstrate a stronger influence of family background on educational attainment in systems with earlier tracking ages, despite no corresponding increase in average performance. This suggests that early tracking hinders some students from reaching their potential. However, the methodological challenges of disentangling family influence and inherent ability using traditional measures of SES and IQ are emphasized. The authors discuss the limitations of relying on sibling similarity as an omnibus measure of family background, as it confounds genetic and environmental influences. They also criticize the use of general ability tests, which are themselves influenced by early family environment. The study draws upon existing research that uses heritability as a measure of opportunity to realize potential and shared environmental influence as an indicator of family background impact. Previous studies have suggested a correlation between equality of opportunity and the relative importance of genetic and shared environmental influences, but empirical evidence remains scarce, especially concerning tracking age practices. The review also discusses transactional models of development, emphasizing the interplay between genetic predispositions and environmental opportunities, and how an individual's autonomy in shaping their environment can impact the expression of genetic potential. Finally, the literature review covers primary and secondary effects of family background on educational attainment, highlighting the challenges in measuring these effects and understanding the mechanisms by which early tracking might affect them.

The study utilizes data from the Netherlands Twin Register (NTR), a longitudinal study encompassing 8847 twins born between 1986 and 1999. The Dutch educational system, with its ability tracking at the end of primary school (around age 12), provides a valuable context. The researchers leverage variation in tracking practices across schools, distinguishing between immediate tracking (homogeneous classes) and delayed tracking (heterogeneous classes with later track assignment). The dependent variable is the final secondary educational track level, assigned numerical scores based on the educational pathway (VMBO, HAVO, VWO). A dichotomous variable indicates whether tracking was delayed. Educational performance is measured using standardized CITO test scores obtained at the end of primary school. Parental education serves as a measure of family socioeconomic background. The study controls for sex and year of birth. The analysis employs genetic structural equation modeling using the classical twin design. This approach compares monozygotic (MZ) and dizygotic (DZ) twins to partition the variance of educational attainment into additive genetic (A), shared environmental (C), and non-shared environmental (E) components. A multigroup model compares the A, C, and E components for twins with immediate versus delayed tracking. To investigate the interaction between tracking age and student performance, a bivariate model is used, incorporating educational performance as a continuous moderator. This bivariate model further decomposes the genetic and environmental effects on educational attainment into components common to performance and components unique to attainment. The Cholesky decomposition is used to model these relationships, allowing the effects of genetic and environmental factors on attainment to depend on performance levels. The researchers discuss assumptions of the twin models, including the equal environments assumption, additivity of genetic effects, absence of assortative mating, and absence of gene-environment correlation. They argue that any potential biases due to violations of these assumptions are unlikely to differentially affect the immediate and delayed tracking groups. Finally, the paper examines potential confounding effects of parental socioeconomic status and student performance on the selection into immediate versus delayed tracking.

The analysis reveals significant differences in the genetic and environmental influences on educational attainment based on the timing of tracking. Overall, identical twins show greater similarity in educational attainment (ρMZ = 0.80) than fraternal twins (ρDZ = 0.48), the same being true for educational performance. An ACE model applied to the data indicates that 58% of the variance in educational attainment is due to genetic differences (VA = 0.74), 22% to shared environmental factors (VC = 0.28), and 19% to non-shared environmental factors (VE = 0.25). The genetic influence on educational attainment is significantly larger for those with delayed tracking (74%) compared to those with immediate tracking (55%), whereas the shared environmental influence is significantly lower for those with delayed tracking (2%) than for those with immediate tracking (27%). This difference in variance components between immediate and delayed tracking is statistically significant. A substantial portion of the genetic influence (57%) on attainment is shared with performance, but a considerable portion (43%) is unique to attainment, suggesting factors beyond primary school performance influence educational outcomes. The differences between immediate and delayed tracking are primarily driven by primary effects (shared genetic and environmental factors between attainment and performance), rather than secondary effects (factors unique to attainment). Further analysis examining the interaction between tracking age and student performance shows that the difference in unique genetic effects on attainment between immediate and delayed tracking is stronger for low-performing students than high-performing students. This supports the notion that early tracking particularly hinders low-performing students from realizing their genetic potential, potentially due to bias. However, the expected interaction effect for shared environmental factors was not found, but the interaction effect for non-shared environmental factors was found, suggesting that delayed tracking may mitigate random errors in placement for low performers. Analysis of selection into immediate versus delayed tracking reveals that parental education is not associated with the choice of tracking method, while student performance is. Controlling for performance, the differences in genetic and shared environmental influences between immediate and delayed tracking groups remain.

The findings support the hypothesis that delaying tracking increases equality of opportunity. The larger genetic influence and lower shared environmental influence associated with delayed tracking indicate that students are better able to realize their genetic potential and that their educational attainment is less dependent on family background when tracking is delayed. The reduction in primary effects, as indicated by the significant differences in genetic and environmental factors common to both performance and attainment, is particularly noteworthy. This suggests that delayed tracking minimizes the impact of family background on student performance, which influences track assignment. This supports transactional models of development, highlighting the importance of providing children with more autonomy to develop their skills and talents in alignment with their genetic predispositions. These findings show a significant impact even with an age difference of just one to three years, potentially due to the change in assessment settings (secondary versus primary school). While the study addresses potential confounding factors, the non-exogenous nature of the choice of immediate or delayed tracking remains a limitation. The finding that low-performing students have a lower genetic influence under immediate tracking, along with the interaction effect observed for non-shared environmental factors, suggests a mitigating effect of delayed tracking on random placement errors.

This study demonstrates that delaying educational tracking leads to a greater expression of genetic potential and a reduced influence of family background on educational attainment. This provides strong evidence for the positive impact of delaying tracking on equality of opportunity. The study highlights the importance of considering both primary and secondary effects of family background on educational decisions. While limitations related to the non-exogenous nature of tracking choice exist, this research has implications for educational policy. Future research should incorporate more detailed measures of parental involvement and teacher bias, as well as explore the use of polygenic scores to further investigate the genetic basis of educational attainment across different tracking contexts.

The most significant limitation is the non-exogenous nature of the decision to delay tracking, which might introduce confounding factors. Although parental SES and student performance were examined as potential confounders, other unmeasured variables could influence the association between tracking age and educational outcomes. While the study uses a large twin sample, the overrepresentation of high-SES families and children without migration backgrounds in the NTR might limit the generalizability of the findings to the entire population. The study also relies on several assumptions inherent to the twin modeling approach which might slightly influence the interpretation of the results.