Multiple Comparisons within Anova Using Orthogonal or Nonorthogonal Components

Abstract

The individual contemplating the use of a multiple comparison procedure is faced with a bewildering array of potential methods, each presenting unique features slightly different than the others. The development of these procedures has been useful in emphasizing the error rates involved in a number of comparisons; however, anova procedures using orthogonal or nonorthogonal components can handle these aspects also. Comparisons involving nonorthogonal components provide a flexible approach to multiple comparisons. The desired error rate can be attained or approximated by modification of the α level.