-
作者：RAO, CR
摘要：A unified approach to the problem is given, making use of the concept of analysis of dispersion in analogy with the analysis of variance for the univariate case. The matrix of mean products giving unbiased estimates of the variances and covariances is called the error matrix; that of mean products yielding unbiased estimates only under the null hypothesis, the matrix due to deviation from the hypothesis. The approp. test criterion is a root of a determinantal equation. Wilk''s A criterion is r...
-
作者：WISHART, J
摘要：The various proofs given in the literature of the distr. of 2d order moment statistics in samples from a normal multivariate population are catalogued and a new geometrical argument due to David Fog is discussed in some detail.
-
作者：NAIR, KR
摘要：The distrs. of McKay''s statistic [nu] = (xn - [image])/[sigma]rand of its studentized form (xn - [image])/s[nu] , where xn is the largest observation and [image] the mean in a sample of n from a normal distr. and sp is an independent estimate of cr based on y degrees of freedom, are obtained. An extensive table of the probability integral of [mu] and a less extensive table of the first 3 coefficients in Hartley''s (Biometrika, 1944) expansion for the studentized form including terms in [nu]-2...
-
作者：DAVID, FN; JOHNSON, NL
摘要：The probability integral transformation, which leads to a rectangular distr. in case a simple hypothesis specifies the parameters of a known probability law, is investigated for composite hypotheses in which a location parameter or a location parameter and a scale parameter are to be estimated from the observed values. The distr. of the transformed variate does not depend on the parameters, but does depend on the functional form of the probability law. All of the distrs. obtained appear to be ...
-
作者：FINNEY, DJ
摘要：Tables are given for the test of] significance of deviation from proportionality in a 2 x 2 table according to Fisher''s exact method at the 0.05, 0.025, 0.01 and 0.005 significance levels (single-tailed test) for tables in which one pair of marginal totals each do not exceed 15. It is noted that certain somewhat larger 2x2 tables may be handled by observing that they are more extreme than significant 2x2 tables which are given or less extreme than non-significant ones. Tables which are much l...
-
作者：YATES, F
摘要：A p x q contingency table is usually tested for independence by a Chi-square test with (p-1)(q-1) degrees of freedom. If the groupings are directly or indirectly based on numerical scales, a test which takes into account trend effects is desirable. It is suggested that a test of significance of the regression coefficients might be useful. The calculations for this test are formalized.
