zbMATH — the first resource for mathematics

Composite likelihood methods. (English) Zbl 0672.62069
Statistical inference from stochastic processes, Proc. AMS-IMS-SIAM Jt. Summer Res. Conf., Ithaca/NY 1987, Contemp. Math. 80, 221-239 (1988).
Summary: Composite likelihood, sometimes called pseudolikelihood, is a likelihood type object formed by adding together individual component log likelihoods, each of which corresponds to a marginal or conditional event. A partial survey is made of the applications of this method, with emphasis made on methods for assessing, comparing, and improving efficiency. It is shown how structural information can be incorporated by conditioning on sufficient statistics. A new application based on rank likelihoods is introduced, and methods for assessing its information is given. Also, it is shown how to construct a stochastic Taylor series in an autonormal problem, with concomitant improvement in efficiency.
[For the entire collection see Zbl 0667.00024.]

62H12 Estimation in multivariate analysis
62A01 Foundations and philosophical topics in statistics