Conditional Inference for Treatment and Error in Multivariate Analysis
将方向性条件推断方法应用于多元方差分析，推导出处理效应、误差方差矩阵和处理改进的精确且易实施的检验，为每个检验提供偏离假设的物理度量，并基于指示的偏离方向进行条件推断。
Directional methods of conditional inference are applied to the multivariate analysis of variance. Exact and easily implemented tests are derived for treatment effect, error variance matrix, and treatment improvement. Each test provides a physical measure of the magnitude of the departure from expectation under the hypothesis and is conditional on the indicated direction of departure. For treatment effect the test provides a geometrically understandable alternative to tests such as the largest root, the trace, and others. For treatment improvement the test attempts to take account reasonably of coordinates that contain no apparent effect.