A composite likelihood ratio approach to the analysis of correlated binary data in genetic association studies

The likelihood function represents statistical evidence in the context of data and a probability model. Considerable theory has demonstrated that evidence strength for different parameter values can be interpreted from the ratio of likelihoods at different points on the likelihood curve. The likelihood function can, however, be unknown or difficult to compute; e.g. for genetic association studies with a binary outcome in large multi-generational families. Composite likelihood is a convenient alternative to using the real likelihood and here we show composite likelihoods have valid evidential interpretation. We show that composite likelihoods, with a robust adjustment, have two large sample performance properties that enable reliable evaluation of relative evidence for different values on the likelihood curve: (1) The composite likelihood function will support the true value over the false value by an arbitrarily large factor; and (2) the probability of favouring a false value over a true value with high probability is small and bounded. Using an extensive simulation study, and in a genetic association analysis of reading disability in large complex pedigrees, we show that the composite approach yields valid statistical inference. Results are compared to analyses using generalized estimating equations and show similar inference is obtained, although the composite approach results in a full likelihood solution that provides additional complementary information.