Schooling and variation in the COMT gene: the devil is in the details. (19th August 2013)

• Record Type: Journal Article
• Authors: Campbell, Daniel; Bick, Johanna; Yrigollen, Carolyn M.; Lee, Maria; Joseph, Antony; Chang, Joseph T.; Grigorenko, Elena L.
• Is Part Of: Journal of child psychology and psychiatry and allied disciplines. Volume 54:Number 10(2013:Oct.)
• Page Start: 1056
• Page End: 1065
• Publication Date: 2013-08-19
• Subjects: Schooling -- nonverbal intelligence -- the COMT gene -- haplotype analysis -- haplo.glm -- interaction effects; Child psychology -- Periodicals; Child psychiatry -- Periodicals; 155.4
• DOI: 10.1111/jcpp.12120
• Languages: English
• ISSNs: 0021-9630

Abstract

Abstract : Background: Schooling is considered one of the major contributors to the development of intelligence within societies and individuals. Genetic variation might modulate the impact of schooling and explain, at least partially, the presence of individual differences in classrooms. Method: We studied a sample of 1, 502 children (mean age = 11.7 years) from Zambia. Approximately 57% of these children were enrolled in school, and the rest were not. To quantify genetic variation, we investigated a number of common polymorphisms in the catechol‐ O ‐methyltransferase ( COMT ) gene that controls the production of the protein thought to account for >60% of the dopamine degradation in the prefrontal cortex. Results: Haplotype analyses generated results ranging from the presence to absence of significant interactions between a number of COMT haplotypes and indicators of schooling (i.e., in‐ vs. out‐of‐school and grade completed) in the prediction of nonverbal intelligence, depending on the parameter specification. However, an investigation of the distribution of corresponding p ‐values suggested that these positive results were false. Conclusions: Convincing evidence that the variation in the COMT gene is associated with individual differences in nonverbal intelligence either directly or through interactions with schooling was not found. p ‐values produced by the method of testing for haplotype effects employed here may be sensitive to parameter settings, invalid under default settings, and should be checked for validity through simulation.