A procedure which provides a highly effective computational solution to problems involving dependent data when the use of the likelihood function may be intractable.