Bounds for p-values of Combinatorial Tests for Clustering in Epidemiology
The data basis for finding temporal or spatial disease clusters in epidemiologic investigations is quite often a large one, and, in this case, the usage of exact statistical tests is not feasible. However, sometimes only small samples of patients are available and it becomes possible to apply exact statistical methods. Grimson et al. (1992, 1993) derived combinatorial tests for different kinds of null hypotheses and test statistics. In general, these tests are only feasible for very small sample sizes. For some situations, we derive exact bounds for the p-values of combinatorial tests for clustering which allow the application of these tests not only for small, but also for medium and large sample sizes.
Unable to display preview. Download preview PDF.
- Kounias, E. and Marin, D. (1974): Best linear Bonferroni bounds. In: Proceedings of the Prague Symposium on Asymptotic Statistics, Vol. II, Charles University, Prague, 179–213.Google Scholar
- Krauth, J. (1991): Bounds for the upper tail probabilities of the multivariate disjoint test. Biometrie und Informatik in Medizin und Biologie, 22, 147–155.Google Scholar
- Krauth, J. (1992): Bounds for the tail probabilities of the linear ratchet scan statistic. In: M. Schader (ed.): Analyzing and Modeling Data and Knowledge. Springer, Berlin, 55–61.Google Scholar
- Krauth, J. (1995): Spatial clustering of neurons by hypergeometric disjoint statistics. In: D. Pfeifer and W. Gaul (eds.): From Data to Knowledge. Springer, Berlin, 253–261.Google Scholar
- Kwerel, S. (1975): Most stringent bounds on aggregated probabilities of partially specified dependent probability systems. Journal of the American Statistical Association, 70,Google Scholar