Validation of zygosity assessment by a self-report questionnaire in a sample of adult Serbian TwInS2

Abstract

Validation of a twin zygosity-estimating questionnaire, The Questionnaire of Twins' Physical Resemblance, created by Oniszczenko et al. and used in European and Serbian twin studies, was carried out on a sample of 222 pairs (176 monozygotic, 46 dizygotic) of adult twins (average age 24.6). Four discriminant functions, use of different sets of indicators (zygosity questionnaire items), were applied in order to obtain the most correct and accurate estimates of zygosity. The first function was a predefined function used in European twin studies, the following two functions contained sets of 18 and 24 freely estimated indicators respectively, while the last one utilized the items with most consistent contributions to zygosity prediction. The analytic procedure included cross-validation, whereby the sample was randomly split into two subsamples, comprising 107 and 115 twin pairs. The results pointed to successful (over 90% correct) identification of monozygotic twins, and sizeably lower correctness in identifying dizygotic twins. Overall correctness of estimation exceeded 90%, with the small set of best-performing indicators. The results encourage questionnaire estimation of zygosity, and raise the issue of improving the classification procedure in dizygotic twins.