Bias estimation of p-values in analytic and simulated Cox Tests for non-nested models

In this paper we show that the estimation of p-values in both Cox’s test for non-nested models and its simulation based analogues is biased and that whilst simulation based Cox tests may be extended to nested models the consequent level of bias is so large as to render the test useless, but that this bias may be removed by adapting the null hypothesis to be simple.

Citation

Wilson, P. (2008) Bias estimation of p-values in analytic and simulated Cox Tests for non-nested models, in Eilers, Paul H. C. (ed.) Proceedings of the 23rd International Workshop on Statistical Modelling, July 7-11, 2008, Utrecht. Utrecht: Statistical Modelling Society, pp. 443-453.

Type

Conference contribution

Language

en

Collections

Faculty of Science and Engineering