摘要:

The power of the Fisher permutation test extended to 2 × k tables is evaluated unconditionally as a function of the under-lying cell probabilities in the table. These results are then applied in assessing the sensitivity of two-generation cancer bioassays in which a fixed number of pups from each litter born in the first generation are selected to continue on test in the second generation. In this case, the two rows of the table correspond to two treatment groups and the k columns correspond to the number of animals responding in a litter. The cell probabilities in this application are based on a suitable beta-binomial superpopulation model.

ISSN:0361-0918    

DOI:10.1080/03610918408812388

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

This paper uses a modified block Choleski decomposition method and tree pruning algorithms to attain the best multivariate subset autoregression for each size (number of non-zero coefficient matrices). Model selection criteria are then employed to select the optimum multivariate subset AR. A Monte Carlo study of these techniques has been investigated to assess their perform-ance, and comparisons of computational efficiency of the proposed procedures are also provided.

ISSN:0361-0918    

DOI:10.1080/03610918408812389

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

Pukkila (1982) introduced a new portmanteau-type test of white noise. The test is based on the differences between the estimated autocorrelations and the corresponding partial autocorrelations. In the present paper the distributions of these differences are consider in the case of normal white noise. Approximations to these distributions are given. Goodness of fit of these theoretical distributions and the corresponding observed distributions is studied using simulated time series of various lengths.

ISSN:0361-0918    

DOI:10.1080/03610918408812390

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

Two methods of estimation for the parameters of an AR(1) process which are based on a non-linear least-squares approach are presented. On the basis of some simulation results they are compared with two maximum likelihood estimates and their relative merits are discussed.

ISSN:0361-0918    

DOI:10.1080/03610918408812391

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

In the present investigation, a series expansion for the cumulants of the xf-distribution is derived in terms of f-jup to the order f-10. As an application a simple Cornish-Fisher-approximation to the noncentrality parameter of noncentral t is given.

ISSN:0361-0918    

DOI:10.1080/03610918408812392

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

Several alternatives to the most common approximation to the inverse moments of the positive binomial distribution are obtained. The method is based on equating moments and gives considerably better approximations for some values of the parameters.

ISSN:0361-0918    

DOI:10.1080/03610918408812393

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

An assumption which is often violated in the application of experimental designs is equality of variances. There are several methods available for estimating the unequal variances. This paper covers incorporating different estimators of the variances with the ordinary least squares and generalized least squares. A Monte Carlo study provides more insight into the behavior of these procedures. For some small sample sizes, the incorporations with the ordinary least squares perform satisfactorily, but with the generalized least squares they do not.

ISSN:0361-0918    

DOI:10.1080/03610918408812394

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

An algorithm, in the form of a Fortran subroutine TRIPLE, is given to compute statistics for the triples test for symmetry, The computational complexity of the algorithm is O(n2) which is an improvement over the straight for ward method, which is O(n3).

ISSN:0361-0918    

DOI:10.1080/03610918408812395

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

摘要:

An algorithm, in t h e form of a Fortran subroutine TRIPLE,is given to compute statistics forthetriples test for symmetry, The 2 computational complexity of the algorithm is O(n2) which is an 3 improvement over the straightforward method, which is O(n3).

ISSN:0361-0918    

DOI:10.1080/03610918408812397

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor

ISSN:0361-0918    

DOI:10.1080/03610918408812387

出版商:Marcel Dekker, Inc.    

年代:1984

数据来源: Taylor