Sample Size Requirements for the Capron and Duyme Balanced Fostering Study of IQ

A simplified method is used to estimate the appropriate sample sizes needed to detect main effects and an interaction effect in analysis of variance, using the IQ data from the Capron and Duyme (1991) adoption study as an example. To achieve power of 80% to reject an hypothesis of no interaction when there is in reality a modest interaction requires about 215 children in each of four groups in a 2 × 2 design, whereas only 9 to 10 children per group are needed to detect main effects. Only a transnational collaborative study could hope to find this many children in the condition where a child from high socioeconomic status background is adopted into a low status family.