Zeitschrift für Biometrie und Biostatistik

Dose Response Relationship from Four-Way Cross Over Trials

Abstract

Itrat Batool Naqvi and Johan Bring

This paper is an application of randomization test for clinical, four ways cross over, trials. The response variable was the proportion of nights with hypoglycaemic episode i.e., lowering the concentration of sugar in the blood. The hypothetical data has been used to examine how persuasively the probability of an episode depends on doses. We also observed how the power, of nonparametric randomization test, was affected when data possessed missing observations with varying sample sizes at 5% level of significance. One consequence in case of missing observations was the reduction of the power of the test, due to the reduction in the actual sample size. As a remedial mean imputation approach used, dose wise and found better power results.