Some comments on Bayes factors

Difficulties arise when improper distributions are used in the evaluation of Bayes factors. Several attempts have been made to remove the difficulties by modifying the factors. This paper argues that the modifications are unnecessary and that it is the impropriety that is at fault. Bayes factors have another undesirable feature, even with propriety, and it is suggested that they be replaced by odds.     

Some comments on two-stage selection procedures

In this paper we study the procedures of Dudewicz and Dalal ( 1975 ), and the modifications suggested by Rinott ( 1978 ), for selecting the largest mean from k normal populations with unknown variances. We look at the case k = 2 in detail, because there is an optimal allocation scheme here. We do not really allocate the total number of samples into two groups, but we estimate this optimal sample size, as well, so as to guarantee the probability of correct selection (written as P(CS)) at least P^?, 1/2 < P^? < 1 . We prove that the procedure of Rinott is “asymptotically in-efficient” (to be defined below) in the sense of Chow and Robbins ( 1965 ) for any k  2. Next, we propose two-stage procedures having all the properties of Rinott's procedure, together with the property of “asymptotic efficiency” - which is highly desirable.     

Some comments on model development and posterior existence

We wish to congratulate John Geweke on producing such an interesting and complete paper. We are delighted to see that serious attempts to make Bayesian methods more generally understood and available are being undertaken. In addition, a number of quite challenging issues is addressed here. Whereas we agree with most of what is stated in the paper, it is our (perceived) duty to single out those things that we feel are more contentious. However, we hope that this can lead to a stimulating discussion of general interest and hopefully to our better understanding of these issues.     

Some comments on sample quantiles for dependent observations

Sample quantiles obtained from a linear process are examined and their asymptotic properties are derived by using simple elementary arguments.     

Some comments on the application of diagnostic techniques in empirical modelling

The application of diagnostic procedures in empirical modelling, raises several practical problems which seem to have received relatively little attention. The aim of this paper is to highlight some of these difficulties and to sketch a possible strategy to isolate the source of an outlier and to decide on appropriate correction action.     

Some comments on a class of simultaneous inference procedures in ancova

Methods for constructing simultaneous confidence intervals for contrasts of treatment effects in analysis of covariance (ANCOVA) models are discussed and compared. A simple procedure is given by which, under a normality assumption made on the covariates, any method appropriate in an analysis of variance (ANOVA) may be applied to a corresponding ANCOVA by means of a simple adjustment of the required critical values. Some properties of this procedure are noted.     

Some comments on efficiency gains from auxiliary information for right-censored data

Heavily right-censored time to event, or survival, data arise frequently in research areas such as medicine and industrial reliability. Recently, there have been suggestions that auxiliary outcomes which are more fully observed may be used to “enhance” or increase the efficiency of inferences for a primary survival time variable. However, efficiency gains from this approach have mostly been very small. Most of the situations considered have involved semiparametric models, so in this note we consider two very simple fully parametric models. In the one case involving a correlated auxiliary variable that is always observed, we find that efficiency gains are small unless the response and auxiliary variable are very highly correlated and the response is heavily censored. In the second case, which involves an intermediate stage in a three-stage model of failure, the efficiency gains can be more substantial. We suggest that careful study of specific situations is needed to identify opportunities for “enhanced” inferences, but that substantial gains seem more likely when auxiliary information involves structural information about the failure process.     

Some comments on escoufier's RV- coefficient as a sensitivity measure in principal component analysis

The RV-coefficient (Escoufier, 1973; Robert and Escoufier, 1976) is studied as a sensitivity coefficient of the subspace spanned by dominant eigenvectors in principal component analysis. We use the perturbation expansion up to second order term of the corresponding projection matrix. The relationship with the measures by Benasseni (1990) and Krzanowski (1979) is also discussed.     

Test for cointegration based on two-stage least squares

A residual-based test for cointegration is proposed. The method of two-stage least squares is used to estimate the cointegration model parameters. The residuals are then tested for the existence of a unit root using the augmented Dickey-Fuller test.     

A weighted symmetric cointegration test

Weighted symmetric estimation is employed to develop a new test for cointegration. Using Monte Carlo simulation, the resulting test is shown to possess greater power than alternative existing tests.     

Some comments on Latin squares and on Graeco-Latin squares,illustrated with postage stamps and old playing cards

We present some comments on Latin squares and on Graeco-Latin squares, with special emphasis on their use in statistics and in a historical context. We also comment on the Knut Vik square, the knight’s move design and the knight’s tour, as well as the Magic Card Puzzle. We consider the well-known 36 officers problem studied by Euler (Verhandelingen uitgegeven door het zeeuwsch Genootschap der Wetenschappen te Vlissingen, vol. 9 (Middleburg 1782), pp. 85–239, 1779/1782), and give two examples of diagonal Latin squares of order 6 due, respectively, to Abbé François-Guillaume Poignard (Chez Guillaume Fricx, Imprimeur &; libraire ruë Bergestract, à l’enseigne des quatre Evangelistes, Bruxelles [4] 79 pp. (p. 71 folded), 1704) and József Dénes (J Lond Math Soc Ser 2, 6(4):679–689, 1970). We illustrate our comments with images of postage stamps and old playing cards. An extensive annotated bibliography ends the paper.     

Some problems of optimal allocation in sample surveys involving inequality constraints

Four optimal sample allocation problems, involving inequality constraints of a special type, are investigated, Solutions are obtained by employing an extension of a result due to Thompson (1962). Applications are given.     

Estimating the Rayleigh distribution scale parameter from doubly censored samples—Some comments

In a recent paper in this journal, Lee, Kapadia and Brock (1980) developed maximum likelihood (ML) methods for estimating the scale parameter of the Rayleigh distribution from doubly censored samples. They reported convergence difficulties in attempting to solve numerically the nonlinear likelihood equation (LE). To mitigate these difficulties, they employed approximations to simplify the LE, but found that the solution of the resulting simplified equation can give rise to parameter estimates of erratic accuracy. We show that the use of approximations to simplify the LE is unnecessary. In fact, under suitable parametric transformation, the log-likelihood function is strictly concave, the ML estimate always exists, is unique and finite. Furthermore, the LE is easy to solve numerically. A numerical example is given to illustrate the computations involved.     

A general inversion theorem for cointegration

A generalization of the Granger and the Johansen Representation Theorems valid for any (possibly fractional) order of integration is presented. This Representation Theorem is based on inversion results that characterize the order of the pole and the coefficients of the Laurent series representation of the inverse of a matrix function around a singular point. Explicit expressions of the matrix coefficients of the (polynomial) cointegrating relations, of the Common Trends and of the Triangular representations are provided, either starting from the Moving Average or the Auto Regressive form. This contribution unifies different approaches in the literature and extends them to an arbitrary order of integration. The role of deterministic terms is discussed in detail.     

Bootstrap tests for time varying cointegration

This article proposes wild and the independent and identically distibuted (i.i.d.) parametric bootstrap implementations of the time-varying cointegration test of Bierens and Martins (2010 Bierens, H. J., Martins, L. F. (2010). Time varying cointegration. Econometric Theory 26:1453–1490.[Crossref], [Web of Science ®] , [Google Scholar]). The bootstrap statistics and the original likelihood ratio test share the same first-order asymptotic null distribution. Monte Carlo results suggest that the bootstrap approximation to the finite-sample distribution is very accurate, in particular for the wild bootstrap case. The tests are applied to study the purchasing power parity hypothesis for twelve Organisation for Economic Cooperation and Development (OECD) countries and we only find evidence of a constant long-term equilibrium for the U.S.–U.K. relationship.