Lecture Notes in Statistics – serie

This concise, easy-to-follow book stimulates interest and develops proficiency in statistical analysis. Packed full of helpful exercises-covering a wide variety of conditions,random patterns, and simple models-Understanding Randomness presents a logical sequence of study, through practice in interpreting random noise before progressing on toreal life problems ... demonstrates proper applications of numerous techniques throughworked out problems ... facilitates further work in statistics, keyed to standard references. . . and strengthens experience with unexpected results-fundamental for workingwith random events.Understanding Randomness serves as vital supplementary reading for both graduate andundergraduate students of statistics-with a standard statistics course as a prerequisiteandas a primary source for exercises in statistics laboratories. Moreover, it is importantfor industrial and mathematical training courses and society or association seminars, aswell as an invaluable workbook for statisticians, biostatisticians, biometricians, socialscientists concerned with improving their data analysis techniques-or anyone dealingwith evaluation of experimental data!

Spatial statistics and Markov Chain Monte Carlo (MCMC) techniques have each undergone major developments in the last decade. Also, these two areas are mutually reinforcing, because MCMC methods are often necessary for the practical implementation of spatial statistical inference, while new spatial stochastic models in turn motivate the development of improved MCMC algorithms. This volume shows how sophisticated spatial statistical and computational methods apply to a range of problems of increasing importance for applications in science and technology. It consists of four chapters: 1. Petros Dellaportas and Gareth O. Roberts give a tutorial on MCMC methods, the computational methodology which is essential for virtually all the complex spatial models to be considered in subsequent chapters. 2. Peter J. Diggle, Paulo J, Ribeiro Jr., and Ole F. Christensen introduce the reader to the model- based approach to geostatistics, i.e. the application of general statistical principles to the formulation of explicit stochastic models for geostatistical data, and to inference within a declared class of models. 3. Merrilee A. Hurn, Oddvar K. Husby, and H?vard Rue discuss various aspects of image analysis, ranging from low to high level tasks, and illustrated with different examples of applications. 4. Jesper Moller and Rasmus P. Waggepetersen collect recent theoretical advances in simulation-based inference for spatial point processes, and discuss some examples of applications. The volume introduces topics of current interest in spatial and computational statistics, which should be accessible to postgraduate students as well as to experienced statistical researchers. It is partly based on the course material for the "TMR and MaPhySto Summer School on Spatial Statistics and Computational Methods," held at Aalborg University, Denmark, August 19-22, 2001.

As statisticians, we are constantly trying to make inferences about the underlying population from which data are observed. This includes estimation and prediction about the underlying population parameters from both complete and incomplete data. Recently, methodology for estimation and prediction from incomplete data has been found useful for what is known as "record-breaking data," that is, data generated from setting new records. There has long been a keen interest in observing all kinds of records-in particular, sports records, financial records, flood records, and daily temperature records, to mention a few. The well-known Guinness Book of World Records is full of this kind of record information. As usual, beyond the general interest in knowing the last or current record value, the statistical problem of prediction of the next record based on past records has also been an important area of record research. Probabilistic and statistical models to describe behavior and make predictions from record-breaking data have been developed only within the last fifty or so years, with a relatively large amount of literature appearing on the subject in the last couple of decades. This book, written from a statistician's perspective, is not a compilation of "records," rather, it deals with the statistical issues of inference from a type of incomplete data, record-breaking data, observed as successive record values (maxima or minima) arising from a phenomenon or situation under study. Prediction is just one aspect of statistical inference based on observed record values.

The special manifolds of interest in this book are the Stiefel manifold and the Grassmann manifold. Formally, the Stiefel manifold Vk,m is the space of k­ frames in the m-dimensional real Euclidean space Rm, represented by the set of m x k matrices X such that X' X = I , where Ik is the k x k identity matrix, k and the Grassmann manifold Gk,m-k is the space of k-planes (k-dimensional hyperplanes) in Rm. We see that the manifold Pk,m-k of m x m orthogonal projection matrices idempotent of rank k corresponds uniquely to Gk,m-k. This book is concerned with statistical analysis on the manifolds Vk,m and Pk,m-k as statistical sample spaces consisting of matrices. The discussion is carried out on the real spaces so that scalars, vectors, and matrices treated in this book are all real, unless explicitly stated otherwise. For the special case k = 1, the observations from V1,m and G1,m-l are regarded as directed vectors on a unit sphere and as undirected axes or lines, respectively. There exists a large literature of applications of directional statis­ tics and its statistical analysis, mostly occurring for m = 2 or 3 in practice, in the Earth (or Geological) Sciences, Astrophysics, Medicine, Biology, Meteo­ rology, Animal Behavior, and many other fields. Examples of observations on the general Grassmann manifold Gk,m-k arise in the signal processing of radar with m elements observing k targets.

Probability matching priors, ensuring frequentist validity of posterior credible sets up to the desired order of asymptotics, are of substantial current interest. They can form the basis of an objective Bayesian analysis. In addition, they provide a route for obtaining accurate frequentist confidence sets, which are meaningful also to a Bayesian. This monograph presents, for the first time in book form, an up-to-date and comprehensive account of probability matching priors addressing the problems of both estimation and prediction. Apart from being useful to researchers, it can be the core of a one-semester graduate course in Bayesian asymptotics.Gauri Sankar Datta is a professor of statistics at the University of Georgia. He has published extensively in the fields of Bayesian analysis, likelihood inference, survey sampling, and multivariate analysis.Rahul Mukerjee is a professor of statistics at the Indian Institute of Management Calcutta. He co-authored three other research monographs, including "A Calculus for Factorial Arrangements" in this series. A fellow of the Institute of Mathematical Statistics, Dr. Mukerjee is on the editorial boards of several international journals.

Space, structure, and randomness: these are the three key concepts underlying Georges Matheron’s scientific work. He first encountered them at the beginning of his career when working as a mining engineer, and then they resurfaced in fields ranging from meteorology to microscopy. What could these radically different types of applications possibly have in common? First, in each one only a single realisation of the phenomenon is available for study, but its features repeat themselves in space; second, the sampling pattern is rarely regular, and finally there are problems of change of scale.This volume is divided in three sections on random sets, geostatistics and mathematical morphology. They reflect his professional interests and his search for underlying unity. Some readers may be surprised to find theoretical chapters mixed with applied ones. We have done this deliberately. GM always considered that the distinction between the theory and practice was purely academic.When GM tackled practical problems, he used his skill as a physicist to extract the salient features and to select variables which could be measured meaningfully and whose values could be estimated from the available data. Then he used his outstanding ability as a mathematician to solve the problems neatly and efficiently. It was his capacity to combine a physicist’s intuition with a mathematician’s analytical skills that allowed him to produce new and innovative solutions to difficult problems. The book should appeal to graduate students and researchers working in mathematics, probability, statistics, physics, spatial data analysis, and image analysis. In addition it will be of interest to those who enjoy discovering links between scientific disciplines that seem unrelated at first glance. In writing the book the contributors have tried to put GM’s ideas into perspective. During his working life, GM was a genuinely creative scientist. He developed innovativeconcepts whose usefulness goes far beyond the confines of the discipline for which they were originally designed. This is why his work remains as pertinent today as it was when it was first written.

These conferences are held every five years. This book contains papers by several prominent biostatisticians, which were inspired by the Second Seattle Symposium in Biostatistics.

The change-point problem has attracted many statistical researchers and practitioners during the last few decades. Here, we only concentrate on the sequential change-point problem. Starting from the Shewhart chart with app- cations to quality control [see Shewhart (1931)], several monitoring procedures have been developed for a quick detection of change. The three most studied monitoring procedures are the CUSUM procedure [Page (1954)], the EWMA procedure [Roberts (1959)] and the Shiryayev?Roberts procedure [Shiryayev (1963) and Roberts (1966)]. Extensive studies have been conducted on the p- formancesofthesemonitoringproceduresandcomparisonsintermsofthedelay detection time. Lai (1995) made a review on the state of the art on these charts and proposed several possible generalizations in order to detect a change in the case of the unknown post-change parameter case. In particular, a wind- limited version of the generalized likelihood ratio testing procedure studied by Siegmund and Venkatraman (1993) is proposed for a more practical treatment even when the observations are correlated. In this work, our main emphasis is on the inference problem for the chan- point and the post-change parameters after a signal of change is made. More speci?cally, due to its convenient form and statistical properties, most d- cussions are concentrated on the CUSUM procedure. Our goal is to provide some quantitative evaluations on the statistical properties of estimators for the change-point and the post-change parameters.

A fundamental issue in statistical analysis is testing the fit of a particular probability model to a set of observed data. Monte Carlo approximation to the null distribution of the test provides a convenient and powerful means of testing model fit. Nonparametric Monte Carlo Tests and Their Applications proposes a new Monte Carlo-based methodology to construct this type of approximation when the model is semistructured. When there are no nuisance parameters to be estimated, the nonparametric Monte Carlo test can exactly maintain the significance level, and when nuisance parameters exist, this method can allow the test to asymptotically maintain the level. The author addresses both applied and theoretical aspects of nonparametric Monte Carlo tests. The new methodology has been used for model checking in many fields of statistics, such as multivariate distribution theory, parametric and semiparametric regression models, multivariate regression models, varying-coefficient models with longitudinal data, heteroscedasticity, and homogeneity of covariance matrices. This book will be of interest to both practitioners and researchers investigating goodness-of-fit tests and resampling approximations.Every chapter of the book includes algorithms, simulations, and theoretical deductions. The prerequisites for a full appreciation of the book are a modest knowledge of mathematical statistics and limit theorems in probability/empirical process theory. The less mathematically sophisticated reader will find Chapters 1, 2 and 6 to be a comprehensible introduction on how and where the new method can apply and the rest of the book to be a valuable reference for Monte Carlo test approximation and goodness-of-fit tests.Lixing Zhu is Associate Professor of Statistics at the University of Hong Kong. He is a winner of the Humboldt Research Award at Alexander-von Humboldt Foundation of Germany and an elected Fellow of the Institute of Mathematical Statistics.From the reviews:"These lecture notes discuss several topics in goodness-of-fit testing, a classical area in statistical analysis. … The mathematical part contains detailed proofs of the theoretical results. Simulation studies illustrate the quality of the Monte Carlo approximation. … this book constitutes a recommendable contribution to an active area of current research." Winfried Stute for Mathematical Reviews, Issue 2006"...Overall, this is an interesting book, which gives a nice introduction to this new and specific field of resampling methods." Dongsheng Tu for Biometrics, September 2006

Point process statistics is successfully used in fields such as material science, human epidemiology, social sciences, animal epidemiology, biology, and seismology. Its further application depends greatly on good software and instructive case studies that show the way to successful work. This book satisfies this need by a presentation of the spatstat package and many statistical examples.Researchers, spatial statisticians and scientists from biology, geosciences, materials sciences and other fields will use this book as a helpful guide to the application of point process statistics. No other book presents so many well-founded point process case studies.From the reviews:"For those interested in analyzing their spatial data, the wide variatey of examples and approaches here give a good idea of the possibilities and suggest reasonable paths to explore." Michael Sherman for the Journal of the American Statistical Association, December 2006

In modern economies, time series play a crucial role at all levels of activity. They are used by decision makers to plan for a better future, by governments to promote prosperity, by central banks to control inflation, by unions to bargain for higher wages, by hospital, school boards, manufacturers, builders, transportation companies, and by consumers in general.A common misconception is that time series data originate from the direct and straightforward compilations of survey data, censuses, and administrative records. On the contrary, before publication time series are subject to statistical adjustments intended to facilitate analysis, increase efficiency, reduce bias, replace missing values, correct errors, and satisfy cross-sectional additivity constraints. Some of the most common adjustments are benchmarking, interpolation, temporal distribution, calendarization, and reconciliation.This book discusses the statistical methods most often applied for such adjustments, ranging from ad hoc procedures to regression-based models. The latter are emphasized, because of their clarity, ease of application, and superior results. Each topic is illustrated with many real case examples. In order to facilitate understanding of their properties and limitations of the methods discussed, a real data example, the Canada Total Retail Trade Series, is followed throughout the book.This book brings together the scattered literature on these topics and presents them using a consistent notation and a unifying view. The book will promote better procedures by large producers of time series, e.g. statistical agencies and central banks. Furthermore, knowing what adjustments are made to the data and what technique is used and how they affect the trend, the business cycles and seasonality of the series, will enable users to perform better modeling, prediction, analysis and planning.This book will prove useful to graduate students and final year undergraduate students of time series and econometrics, as well as researchers and practitioners in government institutions and business. From the reviews:"It is an excellent reference book for people working in this area." B. Abraham for Short Book Reviews of the ISI, December 2006

This book was originally compiled for a course I taught at the University of Rochester in the fall of 1991, and is intended to give advanced graduate students in statistics an introduction to Edgeworth and saddlepoint approximations, and related techniques. Many other authors have also written monographs on this s- ject, and so this work is narrowly focused on two areas not recently discussed in theoretical text books. These areas are, ?rst, a rigorous consideration of Edgeworth and saddlepoint expansion limit theorems, and second, a survey of the more recent developments in the ?eld. In presenting expansion limit theorems I have drawn heavily on notation of McCullagh (1987) and on the theorems presented by Feller (1971) on Edgeworth expansions. For saddlepoint notation and results I relied most heavily on the many papers of Daniels, and a review paper by Reid (1988). Throughout this book I have tried to maintain consistent notation and to present theorems in such a way as to make a few theoretical results useful in as many contexts as possible. This was not only in order to present as many results with as few proofs as possible, but more importantly to show the interconnections between the various facets of asymptotic theory. Special attention is paid to regularity conditions. The reasons they are needed and the parts they play in the proofs are both highlighted.

The purpose of this book is to give a detailed account of some recent devel- ments in the ?eld of probability and statistics for dependent data. It covers a wide range of topics from Markov chains theory, weak dependence, dynamical system to strong dependence and their applications. The title of this book has been somehow borrowed from the book ”Dependence in Probability and Statistics: a Survey of Recent Result” edited by Ernst Eberlein and Murad S. Taqqu, Birkh¨ auser (1986), which could serve as an excellent prerequisite for reading this book. We hope that the reader will ?nd it as useful and stimulating as the previous one. This book was planned during a conference, entitled “STATDEP2005: Statistics for dependent data”, organized by the Statistical Laboratory of the CREST (Research Center in Economy and Statistics), in Paris/Malako?, under the auspices of the French State Statistical Institute, INSEE. See http://www.crest.fr/pageperso/statdep2005/home.htm for some r- rospective informations. However this book is not a conference proceeding. This conference has witnessed the rapid growth of contributions on dep- dent data in the probabilistic and statistical literature and the need for a book covering recent developments scattered in various probability and s- tistical journals. To achieve such a goal, we have solicited some participants of the conferences as well as other specialists of the ?eld.

This monograph is addressed to anyone interested in the subject of restrict- parameter-space estimation, and in particular to those who want to learn, or bring their knowledge up to date, about (in)admissibility and minimaxity problems for such parameter spaces. The coverage starts in the early 1950s when the subject of inference for - stricted parameter spaces began to be studied and ends around the middle of 2004. It presents known, and also some new, results on (in)admissibility and minimaxity for nonsequential point estimation problems in restricted ?ni- dimensional parameter spaces. Relationships between various results are d- cussed and open problems are pointed out. Few complete proofs are given, but outlines of proofs are often supplied. The reader is always referred to the published papers and often results are clari?ed by presenting examples of the kind of problems an author solves, or of problems that cannot be solved by a particular result. The monograph does not touch on the subject of testing hypotheses in - stricted parameter spaces. The latest books on that subject are by Robertson, Wright and Dykstra (1988) and Akkerboom (1990), but many new results in that area have been obtained since. The monograph does have a chapter in which questions about the existence of maximum likelihood estimators are discussed. Some of their properties are also given there as well as some algorithms for computing them. Most of these results cannot be found in the Robertson, Wright, Dykstra book.

Our motivation for writing this book was a dissatisfaction with the many books with titles hke Foundations of Statistics. These books provide a needed description of the subject with examples of various statistical methods; but they do not satisfy the discerning reader because they do not explain why certain conclusions may be drawn from certain data and they do not discuss how statistics, the subject, meshes with the scientific process. We naively set out to fill in these gaps, but the situation is not so simple. What is desired is a tool—the one true statistics—which can be applied to data with certainty to justify conclusions, but what is available, in fact, are various competing theories of statistics. The careful reader may correctly point out the relation of this manuscript to po- modem epistemology, a philosophy which emphasizes that all human knowing—in fields as diverse as religion and science—is culture dependent and that therefore "truth" is not absolute. Take, for example, the "truths" of Christianity and Islam— wars have been fought over their differences. In the field of statistics there are, for example, Bayesians and sampling theorists. Carson (2003) states, "Surely it is better, postmodems tell us, to encourage insights that flow from many different perspectives. . . " In statistics, postmodernism is manifest at two levels. First, the findings of statistical applications are not about the true state of nature but about models of experiments on nature.

This monograph is the first book-length exposition of ranked set sampling. But, the subject matter is by no means new. The original notion of ranked set sampling, though not the technical term, was proposed by McIntyre in 1952. It was buried in the literature for quite a while. Its value was only re-discovered in recent years because of its cost-effective nature. Now, ranked set sampling has attracted practical interest in application areas such as agriculture, forestry, ecological and environmental science, and medical studies etc .. The theoretical foundations of the method has also been developed considerably, particularly during the last 15 years or so. A systematic exposition of the subject becomes necessary. This book covers every development of RSS since the birth of the original idea. Statistical inferences based on RSS are investigated from the originally intended estimation of a population mean to many more complicated pro­ cedures such as the inferences on smooth-function-of-means, quantiles and density functions, the distribution-free tests and regression analyses. Various variants of the original RSS scheme are explored, including RSS with imper­ fect judgment ranking or ranking by concomitant variables, adaptive RSS, multi-layer RSS and a variety of unbalanced RSS with certain optimalities.

Time series and random ?elds are main topics in modern statistical techniques. They are essential for applications where randomness plays an important role. Indeed, physical constraints mean that serious modelling cannot be done - ing only independent sequences. This is a real problem because asymptotic properties are not always known in this case. Thepresentworkisdevotedtoprovidingaframeworkforthecommonlyused time series. In order to validate the main statistics, one needs rigorous limit theorems. In the ?eld of probability theory, asymptotic behavior of sums may or may not be analogous to those of independent sequences. We are involved with this ?rst case in this book. Very sharp results have been proved for mixing processes, a notion int- duced by Murray Rosenblatt [166]. Extensive discussions of this topic may be found in his Dependence in Probability and Statistics (a monograph published by Birkhau ¨ser in 1986) and in Doukhan (1994) [61], and the sharpest results may be found in Rio (2000)[161]. However, a counterexample of a really simple non-mixing process was exhibited by Andrews (1984) [2]. The notion of weak dependence discussed here takes real account of the available models, which are discussed extensively. Our idea is that robustness of the limit theorems with respect to the model should be taken into account. In real applications, nobody may assert, for example, the existence of a density for the inputs in a certain model, while such assumptions are always needed when dealing with mixing concepts.

Two important types of models for reliability and genetics, the semi-Markov and hidden semi-Markov models are presented in this book. Their importance relies on the fact that they generalize several previous models and provide new possibilities to handle real problems. For reliability scientists and engineers, this book comes with a new method for studying systems reliability and it offers modelling and estimation tools. As for the biologists, this work offers them a more adapted model for DNA analysis, namely the hidden semi-Markov model, which is more flexible than the hidden Markov models, extensively used in this field. Moreover, the hidden semi-Markov framework can be used in many other applications, such as reliability, signal treatment, speech or image processing, among others.

Random Effect and Latent Variable Model Selection In recent years, there has been a dramatic increase in the collection of multivariate and correlated data in a wide variety of ?elds. For example, it is now standard pr- tice to routinely collect many response variables on each individual in a study. The different variables may correspond to repeated measurements over time, to a battery of surrogates for one or more latent traits, or to multiple types of outcomes having an unknown dependence structure. Hierarchical models that incorporate subje- speci?c parameters are one of the most widely-used tools for analyzing multivariate and correlated data. Such subject-speci?c parameters are commonly referred to as random effects, latent variables or frailties. There are two modeling frameworks that have been particularly widely used as hierarchical generalizations of linear regression models. The ?rst is the linear mixed effects model (Laird and Ware , 1982) and the second is the structural equation model (Bollen , 1989). Linear mixed effects (LME) models extend linear regr- sion to incorporate two components, with the ?rst corresponding to ?xed effects describing the impact of predictors on the mean and the second to random effects characterizing the impact on the covariance. LMEs have also been increasingly used for function estimation. In implementing LME analyses, model selection problems are unavoidable. For example, there may be interest in comparing models with and without a predictor in the ?xed and/or random effects component.

Permutation testing for multivariate stochastic ordering and ANOVA designs is a fundamental issue in many scientific fields such as medicine, biology, pharmaceutical studies, engineering, economics, psychology, and social sciences. This book presents new advanced methods and related R codes to perform complex multivariate analyses. The prerequisites are a standard course in statistics and some background in multivariate analysis and R software.

From the reviews: "This collection of essays surveys the most important of Fisher's papers in various areas of statistics. ... ... the monograph will be a useful source of reference to most of Fisher's major papers; it will certainly provide background material for much vigorous discussion." #Australian Journal of Statistics#1

Since 1972 the Institute of Mathematics and the Committee of Mathematics of the Polish Academy of Sciences organize annually con­ ferences on mathematical statistics in Wisla. The 1978 conference, supported also by the University of Wroclaw,was held in Wisla from December 7 to December 13 and attended by around 100 participants from 11 countries. K. Urbanik, Rector of the University of Wroclaw, was the honorary chairman of the conference. Traditionally at these conferences there are presented results on mathematical statistics and related fields obtained in Poland during the year of the conference as well as results presented by invited scholars from other countries. In 1978 invitations to present talks were accepted by 20 e~inent statisticians and probabilists. The topics of the invited lectures and contributed papers included theoretical statistics with a broad cover of the theory of linear models, inferences from stochastic processes, probability theory and applications to biology and medicine. In these notes there appear papers submitted by 30 participants of the conference. During the conference, on December 9, there was held a special session of the Polish Mathematical Society on the occasion of elect­ ing Professor Jerzy Neyman the honorary member of the Polish Mathematical Society. At this session W. Orlicz, president of the Polish Mathematical Society, K.Krickeberg,president of the Bernoulli Society. R. Bartoszynski and K. Doksum gave talks on Neyman IS con­ tribution to statistics, his organizational achievements in the U.S.

This monograph treats the question of determining how much to spend for the collection and analysis of public data. This difficult problem for government statisticians and policy-makers is likely to become even more pressing in the near future. The approach taken here is to estimate and compare the benefits and costs of alternative data programs. Since data are used in many ways, the benefits are hard to measure. The strategy I have adopted focuses on use of data to determine fund allocations, particularly in the General Revenue Sharing program. General Revenue Sharing is one of the largest allocation programs in the United States. That errors in population counts and other data cause sizable errors in allocation has been much publicized. Here we analyze whether the accuracy of the 1970 census of population and other data used by General Revenue Sharing should be improved. Of course it is too late to change the 1970 census program, but the method and techniques of analysis will apply to future data programs. In partic­ ular, benefit-cost analyses such as this are necessary for informed decisions about whether the expense of statistical programs is justi­ fied or not. For example, although a law authorizing a mid-decade census was enacted in 1976, there exists great doubt whether funds will be provided so a census can take place in 1985. (The President's Budget for 1981 allows no money for the mid-decade census, despite the Census Bureau's request for $1. 9 million for planning purposes.

A stochastic process {X(t): 0 S t < =} with discrete state space S c ~ is said to be stochastically increasing (decreasing) on an interval T if the probabilities Pr{X(t) > i}, i E S, are increasing (decreasing) with t on T. Stochastic monotonicity is a basic structural property for process behaviour. It gives rise to meaningful bounds for various quantities such as the moments of the process, and provides the mathematical groundwork for approximation algorithms. Obviously, stochastic monotonicity becomes a more tractable subject for analysis if the processes under consideration are such that stochastic mono tonicity on an inter­ val 0 < t < E implies stochastic monotonicity on the entire time axis. DALEY (1968) was the first to discuss a similar property in the context of discrete time Markov chains. Unfortunately, he called this property "stochastic monotonicity", it is more appropriate, however, to speak of processes with monotone transition operators. KEILSON and KESTER (1977) have demonstrated the prevalence of this phenomenon in discrete and continuous time Markov processes. They (and others) have also given a necessary and sufficient condition for a (temporally homogeneous) Markov process to have monotone transition operators. Whether or not such processes will be stochas­ tically monotone as defined above, now depends on the initial state distribution. Conditions on this distribution for stochastic mono tonicity on the entire time axis to prevail were given too by KEILSON and KESTER (1977).

The theory and practice of decision making involves infinite or finite number of actions. The decision rules with a finite number of elements in the action space are the so-called multiple decision procedures. Several approaches to problems of multi­ ple decisions have been developed; in particular, the last decade has witnessed a phenomenal growth of this field. An important aspect of the recent contributions is the attempt by several authors to formalize these problems more in the framework of general decision theory. In this work, we have applied general decision theory to develop some modified principles which are reasonable for problems in this field. Our comments and contributions have been written in a positive spirt and, hopefully, these will an impact on the future direction of research in this field. Using the various viewpoints and frameworks, we have emphasized recent developments in the theory of selection and ranking ~Ihich, in our opinion, provides one of the main tools in this field. The growth of the theory of selection and ranking has kept apace with great vigor as is evidenced by the publication of two recent books, one by Gibbons, Olkin and Sobel (1977), and the other by Gupta and Panchapakesan (1979). An earlier monograph by Bechhofer, Kiefer and Sobel (1968) had also provided some very interest­ ing work in this field.

In this set of notes we study a notion of a random process assoc- ted with a point process. The presented theory was inSpired by q- ueing problems. However it seems to be of interest in other branches of applied probability, as for example reliability or dam theory. Using developed tools, we work out known, aswell as new results from queueing or dam theory. Particularly queues which cannot be treated by standard techniques serve as illustrations of the theory. In Chapter 1 the preliminaries are given. We acquaint the reader with the main ideas of these notes, introduce some useful notations, concepts and abbreviations. He also recall basic facts from ergodic theory, an important mathematical tool employed in these notes. Finally some basic notions from queues are reviewed. Chapter 2 deals with discrete time theory. It serves two purposes. The first one is to let the reader get acquainted with the main lines of the theory needed in continuous time without being bothered by tech­ nical details. However the discrete time theory also seems to be of interest itself. There are examples which have no counte~ in continuous time. Chapter 3 deals with continuous time theory. It also contains many basic results from queueing or dam theory. Three applications of the continuous time theory are given in Chapter 4. We show how to use the theory in order to get some useful bounds for the stationary distribution of a random process.

This monograph is a collection of results recently obtained by the authors. Most of these have been published, while others are awaitlng publication. Our investigation has two main purposes. Firstly, we discuss higher order asymptotic efficiency of estimators in regular situa­ tions. In these situations it is known that the maximum likelihood estimator (MLE) is asymptotically efficient in some (not always specified) sense. However, there exists here a whole class of asymptotically efficient estimators which are thus asymptotically equivalent to the MLE. It is required to make finer distinctions among the estimators, by considering higher order terms in the expansions of their asymptotic distributions. Secondly, we discuss asymptotically efficient estimators in non­ regular situations. These are situations where the MLE or other estimators are not asymptotically normally distributed, or where l 2 their order of convergence (or consistency) is not n / , as in the regular cases. It is necessary to redefine the concept of asympto­ tic efficiency, together with the concept of the maximum order of consistency. Under the new definition as asymptotically efficient estimator may not always exist. We have not attempted to tell the whole story in a systematic way. The field of asymptotic theory in statistical estimation is relatively uncultivated. So, we have tried to focus attention on such aspects of our recent results which throw light on the area.

The first Pannonian Symposium on Mathematical Statistics was held at Bad Tatzmannsdorf (Burgenland/Austria) from September 16th to 21st, 1979. The aim of it was to furthe~ and intensify scientific cooperation in the Pannonian area, which, in a broad sense, can be understood to cover Hungary, the eastern part of Austria, Czechoslovakia, and parts of Poland, ¥ugoslavia and Romania. The location of centers of research in mathematical statistics and probability theory in this territory has been a good reason for the geographical limitation of this meeting. About 70 researchers attended this symposium, and 49 lectures were delivered; a considerable part of the presented papers is collected in this volume. Beside the lectures, vigorous informal discussions among the participants took place, so that many problems were raised and possible ways of solutions were attacked. We take the opportunity to thank Dr. U. Dieter (Graz), Dr. F. Konecny (Wien), Dr. W. Krieger (G8ttingen) and Dr. E. Neuwirth (Wien) for their valuable help in the refereeing work for this volume. The Pannonian Symposium could not have taken place without the support of several institutions: The Austrian Ministry for Research and Science, the State government of Burgenland, the Community Bad Tatzmannsdorf, the Kurbad Tatzmannsdorf AG, the Austrian Society for Information Science and Statistics, IBM Austria, Volksbank Oberwart, Erste Osterreichische Spar-Casse and Spielbanken AG Austria. The Austrian Academy of Sciences iv made possible the participation in the Symposium for several mathematicians. We express our gratitude to all these institutions for their generous help.

In 1978 the idea of studying the generalized inverse Gaussian distribution was proposed to me by Professor Ole Barndorff-Nielsen, who had come across the distribution in the study of the socalled hyperbolic distributions where it emerged in connection with the representation of the hyperbolic distributions as mixtures of normal distributions. The statistical properties of the generalized inverse Gaussian distribution were at that time virtually unde- veloped, but it turned out that the distribution has some nice properties, and models many sets of data satisfactorily. This work contains an account of the statistical properties of the distribu- tion as far as they are developed at present. The work was done at the Department of Theoretical Statistics, Aarhus University, mostly in 1979, and was partial fulfilment to- wards my M.Sc. degree. I wish to convey my warm thanks to Ole Barn- dorff-Nielsen and Preben BI~sild for their advice and for comments on earlier versions of the manuscript and to Jette Hamborg for her skilful typing.