Pairwise likelihood approach to grouped continuous model and its extension

A pseudo-likelihood estimation method for the grouped continuous model and its extension to mixed ordinal and continuous data is proposed as an alternative to maximum likelihood estimation. The method, based on the pairwise likelihood approach, advocates simply pooling marginal pairwise likelihoods to approximate the full likelihood. In addition to being consistent and asymptotically normally distributed, maximum pairwise likelihood estimates are computationally simple to obtain. Simulations show that the estimates are quite accurate, yielding minimal bias and small root mean-squared errors. The methodology is illustrated using real-data examples.