Accession Number : AD0765206

Title :   Widely Usable Analysis of Variance Models.

Descriptive Note : Final rept.,

Corporate Author : SOUTHERN METHODIST UNIV DALLAS TEX

Personal Author(s) : Walsh,John E.

Report Date : JUL 1973

Pagination or Media Count : 72

Abstract : Many results for statistical investigations are valid for more general models than were used in their development. This is the case for situations where independence (or zero correlation), equal variances, and sometimes joint normality are assumed for sets of random variables. This research investigates extended uses of standard results for some analysis of variance situations. Many methods are found to be usable with much more general models, although not all types of effects can be investigated. The research also provides investigation methods that are usable for all the effects. Moreover, the extended models associated with these methods are much more general than those for the standard methods. Section 1 is introductory and includes a statement of the concepts involved. The remaining sections are, with one exception, concerned entirely with analysis of variance. Section 2 describes nonparametric models and methods for one-way ANOVA. Sections 3, 4, and 5 deal with two-way ANOVA with fixed effects. Section 6 deals with n-way ANOVA with fixed effects and no interactions. (Author)

Descriptors :   (*ANALYSIS OF VARIANCE, MATHEMATICAL MODELS), RANDOM VARIABLES, DISTRIBUTION FUNCTIONS, SAMPLING, CONFIDENCE LIMITS, REGRESSION ANALYSIS

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE