Multiple Endpoints

In ANOVA we are mainly based on inter-treatment comparisons. Another common problems arising in biometric studies (especially in biomedical studies) is that of comparing two groups of patients (treatment and a control group) based on multiple response (called multiple endpoints). In this paper we present the continuos and discrete approaches to multiple endpoints. In the case of continuous multiple endpoints we have common assumption in that the covariance matrices in group of the control and observation are equal. Let ρ be the correlation coefficient between Y₁, and Yj endpoints and ρ₁, be the raw ρ-value obtained using some tests statistics for the i-th endpoints. We can also proposed a general bootstrap approach which can be used to estimate the ρ-value without making any parametric and distributional or correctional assumptions. Binary outcomes are common in medical studies. We present the modified Bonfferroni procedures and permutational procedures and we compare these procedures to each other.