摘要:

We present results of a simulation study investigating the behavior of some exact, modified exact,and asymptotic tests of independence for two-way contingency tables having ordinal row and column classifications. The exact analysis utilizes the sufficient statistic for the association parameter in a loglinear model of linear-by-linear association, conditional on marginal totals. Approximate methods include a modified exact test using the mid-P value, a first-order asymptotic test usingthe efficient score statistic, and a higher-order asymptotic test using a double saddlepoint approximation. For most cases considered, the saddlepoint approximation to the exact P-value is outstanding. The extra discreteness induced by the conditioning in the exact test results in that test being conservative. For fixed sample size, the conservativeness diminishes as the numbers of rows and columns increase or as row scores and column scores for the statistic depart from equal-interval.The test based on the mid-P value gives a good compromise between ensuring exactness and eliminating conservativeness due to discreteness

ISSN:0361-0918    

DOI:10.1080/03610919308813078

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

The multivariate portmanteau test proposed by Hosking (1980) for testing the adequacy of an autoregressive moving average model is based on the first s residual autocovariances of the fitted model.In practice a value for s is chosen in dependence on the sample size n, mostly s = 20 for n between 50 and 200. In this paper it will be shown by simulations that the usual choiceof s = 20 oftenleads to a significant deviation of the sample distribution of the test statistic Pmfrom the asymptotic X2distribution. In the case of pure multivariate AR models the Kolmogorow-Smirnow test is used to find those values of s for which the sample distribution shows the best agreement with X2.In this manner s depends not only on the sample size n but also on the order of themodel p and the dimension m. A table for the best choice of s is given for n between 100 and 1000,p between 1 and 5 and m between 1 and 12.

ISSN:0361-0918    

DOI:10.1080/03610919308813079

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

An algorithm is proposed for determining 100 × (1 − α)% small-sample confidence intervals for the difference δ≡ p1− p2of two binomial success probabilities based on n1and n2trials, respectively. The interval covers the true δ with probability at least (1 − α) for all 0 < p1, p2< 1; it is invariant with respect to relabeling of the two populations and with respect to interchanging the outcomes of success and failure intervals. Coverage and expected length comparisons are made with the small sample 100 × (1 − α)% tail intervals of Santner and Snell (1980). A FORTRAN program implementing the algorithm is available from the authors

ISSN:0361-0918    

DOI:10.1080/03610919308813080

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

A common design in biological and toxicological experimentation consists of k groups, each of sample size n. Statistical procedures for the analysis of resulting data often assume normality and equal within group variance. We examined statistical procedures based on probabilityplotting techniques for assessing these assumptions through computer simulations.. To obtain critical values, we generated 10,000 homogeneous normal samples for k=2, 3 and 4, and for n=4 to 12. Power comparisons were made using 1,000 samples each from a variety of non-normal and heterogeneous variance situations.

ISSN:0361-0918    

DOI:10.1080/03610919308813081

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

An approximate method of computing the misclassification probabilities for quadratic discriminant functions (QDF) with known covariance matrices is suggested in this paper. Large scale simulation from empirical distribution of experimental data is used to test the effectivenessof the approximation and compare its performance against an existing method. It is shown that thisapproximation method performs well for multivariate normal distributions of dimension 2, 3. 4, 5,7, and 10. Exact expressions for computing these probabilities for univariate quadratic discriminant function are provided.

ISSN:0361-0918    

DOI:10.1080/03610919308813082

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

A-optimal design measures are obtained for some second-order polyno-micd regression models over cubic experimental regions using a method involving minimization of convex univariate functions. Performance of these design measures under model variation is examined. A-optimal design measures among the central composite type are also derived. The efficiency of these and some integer central composite designs requiring a practicable number of trials is investigated.

ISSN:0361-0918    

DOI:10.1080/03610919308813083

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

The internal norm measure DI(C−1k) introduced in Gray (1989) assesses the influencethat a size- subset of cases has on the least squares parameter estimates in a linear regression. The term internal norm refers to the fact that the assessment of influence is based on a comparison of the subset of interest to all size- subsets internal to the data set. Although this approachis intuitively

ISSN:0361-0918    

DOI:10.1080/03610919308813084

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

Simulations are used to show that the ART (Aligned Ranks Transformation) procedure,when testing for interaction, is robust and almost as powerful as the F-test when the data satisfy the classical assumptions. When these assumptions are violated the ART test is significantly more powerful than the F-test.

ISSN:0361-0918    

DOI:10.1080/03610919308813085

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

Much of the data used for the construction of time series models has been subjected to a rounding procedure prior to its use for modelling. In this work, the error induced by the rounding procedure is modelled as additive and uniformlydistributed, and its effect on ARMA(1,1) models is investigated. The impact on the three model parameters (autoregressive, moving average, and location) is developed; the primary effect is on the moving average term. The effect of roundingerrors on the small-sample properties of the estimators for ARMA(1,1) models is studied using Monte Carlo simulation techniques

ISSN:0361-0918    

DOI:10.1080/03610919308813086

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

Exact tests and confidence intervals are derived for the difference in failure intensities for two exponential populations, based on the conditional distribution of the total time-to-failure of one of the samples, given the total time-to-failure of the two samples. Moments of the distribution are studied, and approximate methods are discussed based on approximate percentiles and on the asymptotic normal distribution. An extension of the method to Type II censored samples is given, and an extension to the problem of comparing mean costs, when the costs of the products or failures are different, is also discussed.

ISSN:0361-0918    

DOI:10.1080/03610919308813087

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor