|
1. |
A branching process method in Lagrance random variate generation |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 1-14
Luc Devroye,
Preview
|
PDF (545KB)
|
|
摘要:
The generalized Lagrange probability distributions include the Borel-Tanner distribution, Haight's distribution, the Poisson-Poisson distribution and Consul's distribution, to name a few. We introduce two universally applicable random variate generators for this family of distributions. In the branching process method, we produce the generation sizes in a Galton-Watson branching process. In the uniform bounding method, we employ the rejection method based upon a simple probability inequality that is valid for id members in a given subfamily.
ISSN:0361-0918
DOI:10.1080/03610919208813005
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
2. |
A new method for assessing multivariate normality with graphical applications |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 15-34
Aydin Ozturk,
Jorge L. Romeu,
Preview
|
PDF (765KB)
|
|
摘要:
A new methodology for assessing distributional assumptions of multivariate data, with graphical applications, is developed. The underlying procedure is based on transforming the multivariate sample into a set of uncorrelated samples and representing the order statistics of each transformed sample by linked vectors in a two dimensional space. The multivariate normality tests are reviewed in detail, the proposed method is described and its properties are discussed.
ISSN:0361-0918
DOI:10.1080/03610919208813006
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
3. |
Robustness to normality of a selection rule |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 35-56
Kerrie Mengersen,
Preview
|
PDF (653KB)
|
|
摘要:
The robustness to the assumption of normality is considered for a special case of the procedure proposed by Bofinger and Mengersen (1986) for selecting the t best from k populations. These authors found a lower bound on the probability of correct selection, under the assumption of independent normal distributions with common unknown variance and equal numbers of observations from each population. Using Tukeys (1960) Generalised Lambda Distribution, it is shown that for most symmetric distributions the bound, as calculated for the normal distribution, is conservative in the face of nonnormality and in particular for heavier tails. For asymmetric distributions, however, the bound performs badly, although this does not necessarily indicate that the procedure itself is at fault. General methods for assessing the effect of nonnormality for particular datasets are also considered.
ISSN:0361-0918
DOI:10.1080/03610919308813007
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
4. |
Computing expected values of normal order statistics |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 57-70
Rudolph S. Parrish,
Preview
|
PDF (626KB)
|
|
摘要:
Expected values and standard deviations of order statistics in samples from a normal parent distribution are developed. Values were coinputed on the basis of a Gauss-Legendre quadrature technique to 25 decimal places for sample sizes of 2(1)50, 60(10)200, and 225(25)500. These values are more precise than those of other available tables and are more accurate for higher sample sizes. An abbreviated table of expected values is presented covering sample sizes up to 50.
ISSN:0361-0918
DOI:10.1080/03610919208813008
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
5. |
Computing variances and covariances of normal order statistics |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 71-101
Rudolph S. Parrish,
Preview
|
PDF (1292KB)
|
|
摘要:
A technique for computation of variances and covariances of normal order statistics is presented. This method provides the means to extend the precision of and correct errors in current tables. A numerical integration approach is employed for the calculations and associated error bounds are developed. Tables were constructed for samples sizes up to M with precision as follows: 25 decimal places (d.p.) for samples sizes of 2(1)20; 20 d.p. for 21(1)30; 15 d.p. for 31(1)40; 10 d.p. for 41(1)50. A table of variances and covariances for sample sizes up to 20 and a table of product moments of normal order statistics for samples sizes of 20(10)50 are presented.
ISSN:0361-0918
DOI:10.1080/03610919208813009
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
6. |
Small sample properties of random coefficient regression estimators: a monte carlo simulation |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 103-132
Terry E. Dielman,
Preview
|
PDF (738KB)
|
|
摘要:
Data measured over time on a number of individuals are referred to in the econometrics literature as pooled cross-sectional and time series data. A number of methods have been suggested for analyzing pooled data. This paper examines the performance of Swamy's random coefficient regression model under a variety of assumptions. The behavior of estimators and tests in small samples is investigated with a Monte Carlo simulation.
ISSN:0361-0918
DOI:10.1080/03610919308813010
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
7. |
Some robust two-sample test statistics based on m-estimators of location |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 133-148
M. A. F. Aboukalam,
Preview
|
PDF (431KB)
|
|
摘要:
This paper describes a simulation experiment that compares the performance, in terms of the size and a function of the power, of four two-sample test statistics based on M-estimators for location. M-esti-mates are chosen to ensure similar levels of breakdown point, gross error sensitivity and as far as possible, similar rejection point. Two pairs of sample size and six different distributions are involved. Matching 97.5% critical values for the statistics are determined.
ISSN:0361-0918
DOI:10.1080/03610919208813011
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
8. |
Estimation, large-sample parametric tests and diagnostics for non-exponential family nonlinear models |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 149-172
Gauss M. Cordeiro,
Gilberto A. Paula,
Preview
|
PDF (885KB)
|
|
摘要:
We study large sample inference for the non-exponential family non-linear models and show a simple procedure to fit these models in the GLIM regression package. We derive general expressions of three statis-tics (Wald, likelihood ratio and score) for testing a subset of parameters of interest, which may be implemented in GLIM. We review various diag-nostics for generalized linear models and extend to more general models. Some examples of analysis of real data are provided.
ISSN:0361-0918
DOI:10.1080/03610919208813012
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
9. |
Estimation of generalized poisson distribution |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 173-188
Famoye Felix,
Carl M.-S. Lee,
Preview
|
PDF (438KB)
|
|
摘要:
The generalized Poisson distribution (GPD) has been found to be a very versatile discrete distribution with applications in various areas of study such as engineering, manufacturing, survival analysis, genetic and branching processes. In this paper, we study the estimation of generalized Poisson distribution by the method of weighted discrepancies between observed and expected frequencies. The methods of maximum likelihood (ML), minimum chi-square and the minimum discrimination information estimation are special cases of the weighted discrepancies method. A new weighting technique, the empirical weighted rates of change, for estimating the GPD parameters is discussed. It is observed that the bias under this new estimation method performs equally good or better than the ML, minimum chi-square and the weighted discrepancies methods, but its variance is the largest of all.
ISSN:0361-0918
DOI:10.1080/03610919208813013
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
10. |
The envelope probability plot as a goodness-of-fit test |
|
Communications in Statistics - Simulation and Computation,
Volume 21,
Issue 1,
1992,
Page 189-202
Richard F. Raubertas,
Preview
|
PDF (414KB)
|
|
摘要:
An envelope plot is a modification of the familiar probability plot that compares a set of data to a probability distribution. Simulation is used to assess and display the variability inherent in a probability plot. The resulting plot has been used as an informal check of the hypothesis that the data were sampled from the distribution. This paper examines the properties of the envelope plot as a formal goodness-of-fit test. Advantages include its versatility and graphical nature. However, it is shown that commonly used numbers of simulations lead to rather high Type I error rates; when the significance level is controlled, the power of the test may be low.
ISSN:0361-0918
DOI:10.1080/03610919208813014
出版商:Marcel Dekker, Inc.
年代:1992
数据来源: Taylor
|
|