am Donnerstag, 08. Mai 2008
16:00 Uhr
Institut für Statistik / TU Graz
Raum 407
Steyrergasse 17/IV, 8010 Graz

Testing marginal homogeneity in multivariate ordinal data, with application to toxicity and drug safety

von Bernhard Klingenberg
Department of Mathematics and Statistics, Williams College, Bronfman Science Center / Williamstown, MA (USA)


Many assessment instruments used in the evaluation of toxicity, safety, pain or disease progression consider multiple ordinal endpoints to fully capture the presence and severity of treatment effects. Contingency tables underlying these correlated responses are often sparse and imbalanced, rendering asymptotic results unreliable or model fitting prohibitively complex without overly simplistic assumptions on the marginal and joint distribution. Instead, we look at stochastic order and marginal inhomogeneity as an expression or manifestation of a treatment effect under much weaker assumptions. The permutation or bootstrap distribution is used throughout to obtain global, subgroup and individual significance levels and we provide a theorem that establishes a connection between marginal homogeneity and the stronger exchangeability assumption. Multiplicity adjustments for the individual endpoints are obtained via step-down procedures, while subgroup significance levels are adjusted via full closed testing. The proposed methodology is illustrated using a collection of 25 correlated ordinal endpoints, grouped into 6 domains, to evaluate toxicity of a chemical compound.

Zurück zur Seite der biometrischen Sektion Steiermark-Kärnten