An importance sampling algorithm for exact conditional tests in log-linear models

摘要: SUMMARY A simple but quite general simulation method for conducting exact conditional lackof-fit tests in log-linear models is proposed. Our Monte Carlo approximation utilises an importance sampling motivated by the crude normal to Poisson distribution. Examples considered include of quasi-symmetry and related square tables concerning higher-order interactions multi-way tables. The competitive with direct from distribution when this feasible outperforms alternative procedures infeasible provided number degrees freedom test not too large. Extension against non-saturated alternatives straightforward briefly discussed illustrated.