Cramér‐von Mises statistics for discrete distributions with unknown parameters

Choulakian, Lockhart & Stephens (1994) proposed Cramér‐von Mises statistics for testing fit to a fully specified discrete distribution. The authors give slightly modified definitions for these statistics and determine their asymptotic behaviour in the case when unknown parameters in the distribution must be estimated from the sample data. They also present two examples of applications.     

Cramér-von Mises tests of fit for the Poisson distribution

Goodness-of-fit tests based on the Cramér-von Mises statistics are given for the Poisson distribution. Power comparisons show that these statistics, particularly A², give good overall tests of fit. The statistic A² will be particularly useful for detecting distributions where the variance is close to the mean, but which are not Poisson.     

Cramér‐von Mises regression

Consider a linear regression model with unknown regression parameters β₀ and independent errors of unknown distribution. Block the observations into q groups whose independent variables have a common value and measure the homogeneity of the blocks of residuals by a Cramér‐von Mises q‐sample statistic T_q(β). This statistic is designed so that its expected value as a function of the chosen regression parameter β has a minimum value of zero precisely at the true value β₀. The minimizer β of T_q(β) over all β is shown to be a consistent estimate of β₀. It is also shown that the bootstrap distribution of T_q(β₀) can be used to do a lack of fit test of the regression model and to construct a confidence region for β₀     

Goodness of fit tests for discrete distributions

Some alternative procedures for testing goodness of fit in discrete distributions are discussed here.. These procedures are based on the probability generating functions.. The methods considered are quite general, being applicable in multidimensional situations., The strength of the tests lies in that no ambiguity as to classification of the data arises.. Hov-ever, some difficulties in the proposed procedures are also pointed out.     

Random partition indicator and spacing statistics

Several conventional and new goodness-of-fit techniques are developed to assess random partitioning of a finite population into disjoint subsets. These nonparametric methods focus on indicator, spacing, and placement statistics. Mathematical development reminiscent of Fourier analysis yields distribution-free orthogonal indicator and spacing component goodness-of-fit procedures for simple random sampling assessment. Providing directional and omnibus criteria to detect between-subset differences, a main contribution of this survey is a unified treatment of the rank indicator procedures in Boos (1986) and the rank spacing methods in Kaigh (1994) with immediate applications to the nonparametric K-sample problem     

Robust weighted Cramér-von Mises Estimators of location,with minimax variance in ϵ-contamination neighbourhoods

We construct weighted Cramér-von Mises location estimators which are asymptotically normally distributed throughout an ?e-contamination neighbourhood of a given, strongly unimodal distribution function, and which minimize the maximum asymptotic variance in such neighbourhoods. Applications to the estimation of a normal or logistic mean are given.     

On the asymptotic distributions of high-order spacings statistics

Goodness-of-fit tests for the uniform distribution based on sums of smooth functions of m-spacings are studied. A limiting sum-of-weighted-chi-squareds approximation is shown to be accurate uniformly in m for the special cases of analogues of Greenwoo?s statistic and Moran's statistic. Asymptotic critical points are provided; theory and Monte Carlo studies show they are accurate for all m provided n is moderately large.     

On a cramér-von mises type statistic for testing bivariate independence

A Cramér-von Mises type statistic for testing bivariate independence, proposed by Hoeffding (1948) and by Blum, Kiefer, and Rosenblatt (1961), is examined in greater detail. The statistic is decomposed into components in the manner of Durbin and Knott (1972), and the components are shown to be related to linear rank statistics. Asymptotic power properties of the Hoeffding statistic and its components in testing for independence with bivariate normal random observations are described; a Monte Carlo study comparing these statistics with other nonparametric statistics for bivariate independence is also reported.     

Improved deviance goodness of fit statistics for a gamma regression model

A gamma regression model with an exponential link function for the means Is considered. Moment properties of the deviance statistics based on maximum likelihood and weighted least squares fits are used to define modified deviance statistics which provide alternative global goodness of fit tests. The null distribution properties of the deviances and modified deviances are compared with those of the approximating chi-square distribution and It is shown that the use of the modified deviances gives much better control over the significance levels of the tests.     

Assessing the fit of finite mixture distributions

Mixture distributions have become a very flexible and common class of distributions, used in many different applications, but hardly any literature can be found on tests for assessing their goodness of fit. We propose two types of smooth tests of goodness of fit for mixture distributions. The first test is a genuine smooth test, and the second test makes explicit use of the mixture structure. In a simulation study the tests are compared to some traditional goodness of fit tests that, however, are not customised for mixture distributions. The first smooth test has overall good power and generally outperforms the other tests. The second smooth test is particularly suitable for assessing the fit of each component distribution separately. The tests are applicable to both continuous and discrete distributions and they are illustrated on three medical data sets.     

Order statistics from discrete distributions

Several recurrence relations and identities available for single and product moments of order1 statistics in a sample size n from an arbitrary continuous distribution are extended for the discrete case,, Making use of these recurrence relations it is shown that it is sufficient to evaluate just two single moments and (n-l)/2 product moments when n is odd and two single moments and {n-2)/2 product moments when n is even, in order to evaluate the first, second and product moments of order statistics in a sample of size n drawn from an arbitrary discrete distribution, given these moments in samples of sizes n-1 and less.. A series representation for the product moments of order statistics is derived.. Besides enabling us to obtain an exact and explicit expression for the product moments of order statistics from the geometric distribution, it. makes the computation of the product moments of order statistics from other discrete distributions easy too.     

Exact and approximate distributions of the chi-square statistic for equiprobability

The distribution of the chi-square goodness-of-fit statistic is studied in the equiprobable case. Tables of exact critical values are given for a = .1, .05, .01, .005; k = 2(1)4, N = 26(1)50; k = 5, N = 26(1)40; k = 6(1)10, N = 26(1)30, where a is the desired significance level, k is the number of cells and N is the sample size. Methods of fitting the true distribution are compared. If k> 3, it is found that a simple additive adjustment to the asymptotic chi-square fit leads to high accuracy even for N between 10 and 20. For k = 2, the Yates corrected chi-square statistic is very accurately fitted by the usual chi-square distribution.     

A correlation type procedure for assessing multivariate normality

A correlation-type statistic for assessing multivariate normality is described. Its estimated finite sample distribution is tabulated, and its performance against certain alternatives is compared with that of a competing Cramer-von Mises type statistic in a Monte Carlo power study. A set of quadrivariate data is examined as illustration of the procedure.     

On distributions Of generalized order statistics

In a wide subclass of generalized order statistics, representations of marginal density and distribution functions are developed. The results are applied to obtain several relations, such as recurrence relations, and explicit expressions for the moments of generalized order statistics from Pareto, power function and Weibull distributions Moreover, characterizations of exponential distributions are shown by means of a distributional identity as well as by* an identity of expectations involving a subrange and a corresponding generalized order statistic.     

Inequalities for generalized order statistics from some restricted family of distributions

The Steffensen inequality is applied to derive quantile bounds for the expectations of generalized order statistics from a distribution belonging to a particular subclass of distributions. The subclass consists of F having the property that F^?1(0+)=x₀>0 and that x →[1? F(x)]x^z is nonincreasing for all x > X₀ and some z > 0.     

A general class of flexible Weibull distributions

We consider a linear combination of two logarithms of cumulative hazard functions and propose a general class of flexible Weibull distribution functions which includes some well-known modified Weibull distributions (MWDs). We suggest a very flexible Weibull distribution, which belongs to the class, and show that its hazard function is monotone, bathtub-shaped, modified bathtub-shaped, or even upside-down bathtub-shaped. We also discuss the methods of least square estimation and maximum likelihood estimation of the unknown parameters. We take two illustrated examples to compare the suggested distribution with some current MWDs, and show that the suggested distribution shows good performances.     

Random translation of discrete order statistics

We investigate the existence and uniqueness of a discrete parent distribution supported on the integers whose order statistics are related by a random translation. We also provide some examples using the constructive method that we propose.     

On limiting distribution laws of statistics analogous to pearson's chi-square

Two test-statistics analogous to Pearson's chi-square test function - given in (1.6) and (1.7) - are investigated. These statistics utilize, apart from the number of sample elements lying in the respective intervals of the partition, their positions within the intervals too. It is shown that the test-statistics are asymptotically distributed - as the sample size N tends to infinity - according to the x ²distribution with parameter r, i.e. the number of intervals chosen. The limiting distribution of the test statistics under the null-hypothesis when N tends to the infinity and r =O(N ^α) (0<α<1), further the consistency of the tests based on these statistics is considered. Some remarks are made concerning the efficiency of the corresponding goodness of fit tests also; the authors intend to return to a more detailed treatment of the efficiency later.     

Domains of attraction of asymptotic distributions of extreme generalized order statistics

Abstract

In extreme value theory for ordinary order statistics, there are many results that characterize the domains of attraction of the three extreme value distributions. In this article, we consider a subclass of generalized order statistics for which also three types of limit distributions occur. We characterize the domains of attraction of these limit distributions by means of necessary and/or sufficient conditions for an underlying distribution function to belong to the respective domain of attraction. Moreover, we compare the domains of attraction of the limit distributions for extreme generalized order statistics with the domains of attraction of the extreme value distributions.