-
作者：SPIEGELHALTER, DJ
摘要：An extension of the test statistic of Spiegelhalter (1977) is investigated. The Bayesian background to the test is discussed, and numerical examples display the adaptive nature of the statistic. A small simulation study indicates good power properties against all alternatives for n = 20, and symmetric alternatives for n = 50.
-
作者：GORIA, MN
摘要：Several well-known rank tests for location and scale are generalized by raising the score generating functions of the tests to an exponent .alpha.. Asymptotic efficiencies of the resulting statistics are relative to the tests for .alpha. = 1, taking various densities. The distributions for which the new tests are locally most powerful are derived.
-
作者：REGAL, R
摘要：The 2 sample problem for exponential populations can be tested exactly using an F statistic when the samples are complete or type II censored. An analogous F test was used as an approximate test when the data are type I censored. Sampling theory from Bartholomew (1963) is used to investigate the accuracy of the F approximation for small sample sizes.
-
作者：SASABUCHI, S
摘要：A new multivariate generalization of a 1-sided test that is different from that of Kudo (1963) is proposed. Let X be a p-variate normal random variable with the mean vector .mu. and a known covariance matrix. The null hypothesis that .mu. lies on the boundary of a convex polyhedral cone determined by linear inequalities is considered; the alternative is that .mu. lies in its interior. A 2-sided version is also discussed. Likelihood ratio tests are provided and some applications are discussed, ...
-
作者：HANUMARA, RC; STRAIN, WF
摘要：The actual percentage coverages attained in simultaneous confidence intervals on variance components with confidence level 1-2.alpha. are investigated by Monte Carlo techniques. The percentage points l and u of the extreme roots of a Wishart matrix at .alpha. levels of 0.10 and 0.15 yield 0.95 and 0.90 confidence intervals, respectively.
-
作者：WINTERBOTTOM, A
摘要：Asymptotic expansions are used to provide 1st- and 2nd-order adjustments to large sample approximations for system reliability confidence intervals obtained from component test data. The effectiveness of the adjustments for sample sizes of practical interest is demonstrated by numerical comparisons with results obtained using exact methods and also by a Monte Carlo study for systems with known reliabilities.
-
作者：LAWAL, HB; UPTON, GJG
作者单位：University of Essex
摘要：The X2 goodness-of-fit test is considered for cases in which the observed frequencies are supposed to be distributed according to a specified multinomial distribution. A log normal approximation to the distribution of X2 is proposed. The approximation is valid providing that the smallest expectation is greater than r/d3/2, where r is the number of expectations < 5, and d is the number of degrees of freedom.
-
作者：HINKLEY, DV
摘要：For a single-parameter curved exponential family, the conditional distribution of the maximum likelihood estimator is the normalized likelihood function, except for a relative error proportional to inverse sample size.
-
作者：WILD, CJ
摘要：The variance .sigma.2 from n pairs (Xi, Yi) which are independent normal random variables with unknown mean .mu.i and variance .sigma.2 was estimated. According to a marginal or conditional sufficiency argument, all useful information about .sigma.2 is contained in the differences Xi-Yi. Under fairly weak assumptions about the means, a considerable amount of information is also extractable from the sums Xi + Yi.
-
作者：ATKINSON, AC
摘要：One way of selecting models is to choose that model for which the maximized log likelihood minus a multiple of the number of parameters estimated is a maximum. The choice of values for the multiplier is explored, with the value 1 corresponding to Akaike''s information criterion. The relationship with Bayesian procedures is mentioned.
