Use of multiple logistic analysis in twin zygosity diagnosis

Stochastic methods were applied to the diagnosis of twin zygosity by mailed questionnaire in a study on adult twins. Multiple logistic discriminant analysis was able to classify twin pairs previously left unclassified (XZ) by a deterministic method, and accuracy of classification was verified by blood testing. The mean intrapair differences of height and weight of XZ pairs were in between dizygotic and monozygotic pairs, but the frequency of personal contact between XZ twins was significantly less than that of MZ or DZ pairs. Both classification methods, when used separately with the same questions, left unclassified 6.5-7.6% of all 11,542 respondent pairs in the Finnish Twin Registry, but deterministic classification followed by logistic analysis misclassified only 1.9% of all twin pairs.