Genotype-Environment Interaction in Mathematics Performance: Application of a Heteroscedastic Multilevel IRT Model on Twin Data

Results of international comparisons such as PISA (Program for International Student Assessment) and TIMSS (Trends in International Mathematics and Science Study) suggest that Dutch students are underperforming in mathematics:... [ view full abstract ]

Results of international comparisons such as PISA (Program for International Student Assessment) and TIMSS (Trends in International Mathematics and Science Study) suggest that Dutch students are underperforming in mathematics: While, overall, their performance is not bad, there is only a small percentage of high-performing students in the Netherlands when compared to other participating countries. These results are often taken to indicate that mathematical education in Dutch schools is not appropriate for mathematically talented students as they are not nurtured to their maximum potential. However, so far, there has been no empirical study yet that investigated this hypothesis.     
     We investigated the underperformance of Dutch mathematically talented students from a behavior genetics perspective where we operationalized mathematical talent as the genotypic value, a genetic concept representing the sum of average effects of genes that influence mathematical achievement. If indeed, Dutch students with a genetic predisposition (i.e., a high genotypic value) are not nurtured to their full potential, their mathematics performance should be more affected by environmental factors (e.g., being at the mercy of random events, like having a teacher that is interested in their abilities) than that of children with a genetic predisposition for low mathematical ability. Such a situation induces the presence of genotype-environment interaction: conditional on a student’s genotypic value for mathematical ability, environmental influences can be more or less important.
     To investigate this hypothesis, we analyzed mathematics performance of 2110 Dutch twin pairs on a Dutch national educational achievement test. To prevent the spurious finding of any genotype-environment interactions based on properties of the measurement scale, we corrected for error variance heterogeneity in the measurement of mathematics performance through the application of an item response theory model (IRT). As hypothesized, we found a positive genotype-environment interaction (= 2.0837 (SD: 0.4939, 95% highest posterior density interval: [1.0886; 3,0344]). These results indicate that, in the Netherlands, environmental influences are relatively more important in explaining individual differences in students with a genetic predisposition for high mathematical ability. Implications for our understanding of the issue at hand and possible conclusions for policy measures will be discussed at the end of the presentation.