MONOGRAPHS ON

STATISTICS AND APPLIED PROBABILITY

General Editors
D.R. Cox, D.V. Hinkley, N. Keiding, N. Reid,
D.B. Rubin and B. W. Silverman
Stochastic Population Models in Ecology and Epidemiology
M.S. Bartlett (1960)
2 Queues D.R. Cox and W.L. Smith (1961)
3 Monte Carlo Methods J.M. Hammersley and D.C. Handscomb (1964)
4 The Statistical Analysis of Series of Events D.R. Cox and P.A. W. Lewis (1966)
5 Population Genetics W.J. Ewens (1969)
6 Probability, Statistics and Time M.S. Bartlett (1975)
7 Statistical Inference S.D. Silvey (1975)
8 The Analysis of Contingency Tables B.S. Everitt (1977)
9 Multivariate Analysis in Behavioural Research A.E. Maxwell (1977)
10 Stochastic Abundance Models S. Engen (1978)
11 Some Basic Theory for Statistical Inference E.J.G. Pitman (1979)
12 Point Processes D.R. Cox and V. Isham (1980)
13 Identification of Outliers D.M. Hawkins (1980)
14 Optimal Design S.D. Silvey (1980)
15 Finite Mixture Distributions B.S. Everitt and D.J. Hand (1981)
16 Classification A.D. Gordon (1981)
17 Distribution-free Statistical Methods J.S. Mariz (1995)
18 Residuals and Influence in Regression R.D. Cook and S. Weisberg (1982)
19 Applications of Queueing Theory, 2nd edition G.F. Newell (1982)
20 Risk Theory, 3rd edition R.E. Beard, T. Pentikainen and E. Pesonen (1984)
21 Analysis of Survival Data D.R. Cox and D. Oakes (1984)
22 An Introduction to Latent Variable Models B.S. Everitt (1984)
23 Bandit Problems D.A. Berry and B. Fristedt (1985)
24 Stochastic Modelling and Control M.H.A. Davis and R. Vinter (1985)
25 The Statistical Analysis of Compositional Data J. Aitchison (1986)
26 Density Estimation for Statistical and Data Analysis B. W. Silverman
27 Regression Analysis with Applications G.B. Wetherill (1986)
28 Sequential Methods in Statistics, 3rd edition G.B. Wetherill and
K.D. Glazebrook (1986)
29 Tensor Methods in Statistics P. McCullagh (1987)
30 Transformation and Weighting in Regression R.J. Carroll and D. Ruppert (1988)
31 Asymptotic Techniques for Use in Statistics O.E. Barndoff-Nielson and
D.R. Cox (1989)
32 Analysis of Binary Data, 2nd edition D.R. Cox and E.J. Snell (1989)
 33 Analysis of Infectious Disease Data N. G. Becker ( 1989)
34 Design and Analysis of Cross-Over Trials B. Jones and M.G. Kenward (1989)
35 Empirical Bayes Method, 2nd edition J.S. Maritz and T. Lwin (1989)
36 Symmetric Multivariate and Related Distributions K.-T. Fang, S. Katz and
K. W. Ng (1990)
37 Generalized Linear Models, 2nd edition P. McCullagh and J.A. Neider (1989)
38 Cyclic and Computer Generated Designs, 2nd edition J.A. John and
E.R. Williams (1995)
39 Analog Estimation Methods in Econometrics C.F. Manski (1988)
40 Subset Selection in Regression A.J. Miller (1990)
41 Analysis of Repeated Measures M.J. Crowder and D.J. Hand (1990)
42 Statistical Reasoning with Imprecise Probabilities P. Walley (1991)
43 Generalized Additive Models T.J. Hastie and R.J. 1ibshirani (1990)
44 Inspection Errors for Attributes in Quality Control N.L. Johnson, S. Katz and
X. Wu (1991)
45 The Analysis of Contingency Tables, 2nd edition B.S. Everitt ( 1992)
46 The Analysis of Quantal Response Data B.J. T. Morgan ( 1993)
47 Longitudinal Data with Serial Correlation: A State-Space Approach
R.H. Jones (1993)
48 Differential Geometry and Statistics M.K. Murray and J. W. Rice (1993)
49 Markov Models and Optimization M.H.A. Davies (1993)
50 Networks and Chao-Statistical and Probabilistic Aspects
O.E. Bamdorff-Nielsen, J.L Jensen and W.S. Kendall (1993)
51 Number Theoretic Methods in Statistics K.- T. Fang and Y. Wang ( 1994)
52 Inference and Asymptotics O.E. Barndorff-Nielsen and D.R. Cox (1994)
53 Practical Risk Theory for Actuaries C.D. Daykin, T. Pentikainen and
M. Pesonen (1994)
54 Biplots J.C. Gower and D.J. Hand (1996)
55 Predictive Inference: An IntroductionS. Geisser (1993)
56 Model-Free Curve Estimation M.E. Tarter and M.D. Lock (1993)
57 An Introduction to the Bootstrap B. Efron and R.J. 1ibshirani (1993)
58 Nonparametric Regression and Generalized Linear Models P.J. Green and
B.W. Silverman (1994)
59 Multidimensional Scaling T.F. Cox and M.A.A. Cox (1994)
60 Kernel Smoothing M.P. Wand and M.C. Jones (1995)
61 Statistics for Long-Memory Processes J. Beran (1995)
62 Nonlinear Models for Repeated Measurement Data M. Davidian and
D.M. Giltinan (1995)
63 Measurement Error in Non-Linear Models R.J. Carroll, D. Ruppert and
LA. Stefanski ( 1995)
64 Analyzing and Modeling Rank Data John I. Marden (1995)

(Full details concerning this series are available from the Publishers.)
 Analyzing and
Modeling
Rank Data

JOHN I. MARDEN
Department of Statistics,
University of Illinois at Urbana-Champaign, USA

Boca Raton London New York

CRC Press is an imprint of the

Taylor & Francis Group, an informa business
A CHAPMAN & HALL BOOK
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 1995 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

This book contains information obtained from authentic and highly regarded sources. Reason-
able efforts have been made to publish reliable data and information, but the author and publisher
cannot assume responsibility for the validity of all materials or the consequences of their use. The
authors and publishers have attempted to trace the copyright holders of all material reproduced in
this publication and apologize to copyright holders if permission to publish in this form has not
been obtained. If any copyright material has not been acknowledged please write and let us know so
we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.
copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organiza-
tion that provides licenses and registration for a variety of users. For organizations that have been
granted a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
 Contents

Preface xi

1 Introduction 1
1.1 Rank data 1

2 Looking at Data 5
2.1 Introduction: Permutation polytopes 5
2.2 Projections of polytopes 12
2.3 Marginals 17
2.4 Pairs 19
2.5 Center, spread, and distance 21
2.5.1 Some useful distances 23
2.5.2 Estimating the center 28
2.5.3 Estimating spread, location known 30
2.5.4 Estimating spread, location unknown 31
2.5.5 Clustering: £-centers 33
2.6 Linear su bspaces 39
2.6.1 Spectral decomposition 45
2.6.2 Inversions 52
2.7 Exercises 54

3 Formal Tests of Uniformity 56

3.1 Introduction 56
3.2 Summary statistics 57
3.2.1 Projections 58
3.2.2 Probabilities 60
3.2.3 Marginals 61
3.2.4 Means 63
3.2.5 Pairs 64
3.3 lnvariance and monotonicity of distances 68
3.4 Distance from a fixed vector 73
3.4.1 Means and variances 75
vi CONTENTS
3.4.2 Exact distributions 77
3.4.3 Asymptotics as m - oo 82
3.5 One-sample diversity and concordance 86
3.6 Summary 92
3.7 Exercises 93

4 Comparing Populations of Judges 97

4.1 Introduction 97
4.2 The basic statistics:
Multivariate analysis of variance tests 98
4.3 Distances from a modal ranking 101
4.4 Concordance and diversity 104
4.5 Exercises 110

5 Overview of Models 111

5.1 Introduction 111

=
5.2 Probability models - General
5.3 Thurstonian Order statistic models
5.4 Distance-based models
112
114
115
5.5 Paired comparison models - Babington Smith 115
5.5.1 Bradley-Terry /Mallows 117
5.5.2 Mallows' models 118
5.6 Multistage models 118
5.6.1 Plackett-Luce 118
5.6.2 Free and ¢-component models 120
5. 7 Sufficient statistic models 122
5.8 Loglinear models 123
5.9 ANOVA-like models 124
5.10 Nested orthogonal contrast models 126
5.10.1 The free model 129
5.10.2 The ¢ model 129
5.10.3 Contingency table models 130
5.11 Unfolding models 132
5.12 Generalizing the models 133
5.13 Some axiomatics 134
5.13.1 Luce's choice axiom 134
5.13.2 Unidimensionality, unimodality, and consensus 136
5.14 Likelihood methods and exponential families 140
5.14.1 The likelihood function and
Fisher information 141
5.14.2 Maximum likelihood estimation 143
CONTENTS vii

5.14.3 Likelihood ratio tests 143

5.14.4 Exponential families 145
5.15 Exercises 147

6 Distance-Based Models 149

6.1 Introduction 149
6.2 Fitting the models: Known mode 149
6.2.1 Kendall 151
6.2.2 Cayley 151
6.2.3 Hamming 152
6.3 Unknown mode - Likelihood 154
6.4 Unknown mode - Bayesian 156
6.5 Asymptotics as m --+ oo 158
6.5.1 Kendall 159
6.5.2 Cayley 160
6.5.3 Hamming 161
6.5.4 Maximum 161
6.6 Assessing fit 162
6.6.1 Kendall 163
6.6.2 Hamming and Cayley 165
6.7 Exercises 168

7 Babington Smith, Phi-Models, and Inversions 170

7.1 Introduction 170
7.2 Babington Smith 172
7.3 Contrast 4J models 175
7.4 Bradley-Terry/Mallows and Spearman's
distance models 178
7.4.1 Bradley-Terry /Mallows 179
7.4.2 Submodels of Bradley-Terry /Mallows,
including Spearman 180
7.4.3 Between Babington Smith and
Bradley-Terry /Mallows 180
7.5 Basic results for orthogonal contrast 4J models 182
7.6 Details for the orthogonal contrast 4J models 184
7.6.1 Preliminaries 185
7.6.2 Null distribution for Kendall 187
7.6.3 Non-null distribution for Phi, Mallows,
and Kendall 189
7.6.4 Decomposing Mallows' 4J model;
Proofs of Theorems 6.3 and 6.5 191
viii CONTENTS

7.7 Examples of orthogonal contrast models 193

7.8 Mixed orthogonal contrastjBabington Smith
models 198
7.9 Multistage models and patterns of ties 200
7.9.1 Patterns of ties 200
7.9.2 Mallows' ~ model for tied ran kings 202
7.9.3 Multistage models: ~-component 203
7.10 Free models and contingency tables 204
7.11 Inversion models 209
7.12 Exercises 211

8 Plackett-Luce, Logistic, and Vase Models 213

8.1 The first-order model 213
8.2 Extended vase models: No interaction 217
8.3 Extended Plackett models 219
8.4 q-permutations 226

9 Marginal and ANOVA-Type Loglinear Models 228

9.1 Introduction 228
9.2 Submodels of the Marginals model 228
9.3 Extended Marginals models 232
9.4 ANOVA-type loglinear models 234
9.5 Paired and multisample models 236

10 Latent Class and Unfolding Models 243

10.1 Introduction 243
10.2 Latent class models 244
10.3 The EM algorithm for latent class models 247
10.4 Unidimensional unfolding models 251
10.5 Exercise 258

11 Tied, Partial, and Incomplete Rankings 259

11.1 Introduction 259
11.1.1 Tied rankings 261
11.1.2 Rankings of subsets:
Balanced incomplete block designs 261
11.2 Descriptive statistics 263
11.3 Tests of uniformity 265
11.4 Tied rankings 269
11.4.1 Marginals and Means 270
11.4.2 Pairs 271
CONTENTS ix

11.4.3 Distances 273

11.5 Rankings of subsets:
Balanced incomplete block designs 275
11.5.1 Marginals and Means;
Hamming's and Spearman's distances 275
11.5.2 Pairs and Kendall's distance 278
11.6 Comparing samples 280
11.7 Models 281
11.7.1 Thurstonian models 283
11.7.2 Plackett-Luce and vase models 284
11.7.3 Babington Smith 284
11.7.4 Orthogonal contrast ¢models:
Tied rankings 286
11.7.5 Orthogonal contrast ¢models: Subsets 287
11.7 .6 Exponential family models 288
11.7.7 Coefficients and comparing populations 289
11.8 Exercises 291

12 Appendix 294
12.1 Some linear algebra 294
12.1.1 Eigenvalues and eigenvectors 295
12.2 Means and covariances for vectors and matrices 296
12.2.1 Definitions 296
12.2.2 Kronecker products 297
12.3 Normality and chi-squares 299
12.4 Some asymptotics 300
12.4.1 Central limit theorem 300
12.4.2 Convergence in probability 301

Bibliography 302

Author Index 311

Subject Index 314

 Preface

Ranking has always been an integral part of statistics, both in

nonparametric analysis and in the analysis of judges' rankings of
objects. Myriad data-analytic techniques and probability mod-
els have been developed over the years. The present volume at-
tempts to summarize and unify much of the statistical work on
rank data. The state of the art circa 1950 is summarized in the
"Symposium on Ranking Methods" conducted by the Royal Sta-
tistical Society. (See Journal of the Royal Statistical Society B,
1950, Volume 12, pages 153-191.) The papers and discussions
address the fundamental the questions: How best to elicit rank-
ings? What goes on in a ranker's mind? How does a statistician
model and analyze such highly structured data? Since then, high
points include Luce's work on choice (Individual Choice Behavior,
1959), Arrow's on combining rankings (Social Choice and Individ-
ual Values, 1963), and Coombs' on unfolding models (A Theory
of Data, 1964). Critchlow's book, Metric Methods for Analyzing
Partially Ranked Data (1985), unifies an important aspect ofrank-
ing. The book by Diaconis, Group Representations in Probability
and Statistics (1988), has an amazing wealth of information and
insights, on ranking and many other subjects. Most of the proba-
bility models and statistical analyses resulting from these years of
activity gathered (with their creators) in Amherst, Massachusetts
in 1990. An AMS-IMS-SIAM Research Conference at the Univer-
sity of Massachusetts turned the spotlight on the broad range of
ranking activities. Michael Fligner and Joseph Verducci, in addi-
tion to arranging the conference, edited the collection of papers
presented in the book Probability Models and Statistical Analyses
for Ranking Data (1993). I think all participants were surprised
and delighted at the variety of work in the area.
The literature is rich with methods for analysis and modeling of
rank data. People already in the field know the basic models and
methods, but the scientific community at large has no single source
for this information. The main goal of this book is to collect and
xii PREFACE
present systematically many of the popular and useful methods. I
try to reflect the spirit of the 1950 symposium and 1990 conference.
Models and analyses should be realistic enough to make sense, but
practical, too. The objective is to gain insight into the data and
the mechanism that produced them. Beautiful models that do not
make sense and cannot be calculated with current technology are
not helpful.
The material should be accessible to anyone with a basic course
in mathematical statistics, and familiarity with linear algebra and
the basic elements of multivariate statistical analysis. The texts
by Hogg and Craig (1978) and Bickel and Doksum (1977) are suf-
ficient to understand and apply most of the results, although a
few of the proofs need more advanced material such as Lehmann
(1983), Rao (1973) and Serfling (1980). Mardia, Kent and Bibby
(1979) is an excellent introduction to multivariate analysis. To ap-
preciate the models in the book, one should have familiarity with
linear regression and analysis of variance (e.g., Weisberg, 1985) and
contingency table models (e.g., Agresti, 1990, or Bishop, Fienberg,
and Holland, 1975).
The book is quite dense. To use it as a text in a one-semester
graduate-level course, one may wish to cover about half the book.
I recommend choosing something like the following material: all of
Chapter 1; all of Chapter 2 through Section 2.5, and the beginning
of Section 2.6, leaving out Sections 2.6.1 and 2.6.2; Sections 3.1 and
3.2, Section 3.4 through 3.4.1, and Sections 3.5 and 3.6; Chapter
5 through Section 5.11, plus Section 5.14; Sections 6.1 and 6.2;
Chapter 7 through Section 7.7, but skipping Sections 7.4.2, 7.4.3,
and 7.6; Section 8.1; Sections 9.1, 9.2, and 9.4.
Many people have spurred my interest in ranking, most notably
Douglas Critchlow, Persi Diaconis, Michael Fligner, Georgia Lee
Thompson, and Joseph Verducci. Robert Bohrer, who recently
passed away, had been terrifically inspiring. I especially want to
thank Lyinn Chung for her fine work in this area, and Kyungmee
Choi, Michael Walker, Colin Mallows, and Louis Roussos for pro-
viding fascinating data. The University of Illinois deserves thanks
for granting me a sabbatical leave, as does the National Science
Foundation for support during my career. The reviewers made
many valuable comments, and 'fEX and Dra'fEX (Gurari, 1994)
helped create the typesetting and figures.
PREFACE xiii

This book is dedicated to my mother, Ann, and Phil.

John Iglehart Marden

Urbana, March 1995
 CHAPTER 1

Introduction

1.1 Rank data

Humans seem to be unable to avoid ranking things. Top Five/
Ten/Twenty/Hundred lists abound: Best (Worst) Dressed, Best
Suburbs, Most Watched TV Shows, Movies, Videos, Best Foot-
ball Teams, and so on. Rankings have very serious uses as well.
Companies need to know what products consumers prefer; social
and political leaders need to know what the society values; elec-
tions need to be conducted; young people need to be drafted (?);
prospective students need to be assessed. Data consisting of rank-
ings appear in Psychology, Animal Science, Educational Testing,
Sociology, Economics, and Biology. In fact, for almost any situa-
tion where there are data it can be helpful to transform the data
into ranks. Nonparametric statistical analysis was initially based'
mainly on ranks.
This book is aimed at preference rankings. The basic unit of
analysis consists of n judges ranking a set of m objects, the set
of objects denoted by 0 = {OI.02,····0m}. The objects can be
almost anything: words, people, sports, soft drinks, ethnic groups,
styles of literary criticism, crackers, political goals, occupations.
Judges are usually people, but they can be animals or even inani-
mate sensing devices.
A full ranking of the objects assigns a complete ordering to the
objects: There is a best object, second best, ... , and finally a worst
one. The two most useful representations of a ranking are the rank
vector and the order vector. A rank vector lists the ranks given
to the objects, where "1" denotes best and "m" denotes worst. It
presumes the objects are listed in a prespecified order. An order
vector lists the objects themselves, in order from best to worst.
For example, suppose the objects are m = 4 popular soft drinks,
0 = {Coke, Pepsi, 7- Up , Sprite}, and a judge likes 7- Up best,
then Sprite, then Pepsi, then Coke. The ordering is ( 7- Up, Sprite,
 2 INTRODUCTION

Table 1.1 Two representations

of the same ranking.

The Ranking The Ordering

Coke 4 7-Up 1
Pepsi 3 Sprite 2
7-Up 1 Pepsi 3
Sprite 2 Coke 4

Pepsi, Coke). What is the rank vector? It depends on the listed

order of the drinks. If this list is taken as in the first column
of Table 1.1, i.e., Coke, Pepsi, 7-Up, Sprite, then the ranking is
(4, 3, 1, 2). (We will sometimes write it 4312.) As in the last two
columns of Table 1.1, the ordering is the listing of the objects for
which the rankings are in the order 1, 2, ... , m. Whichever way is
chosen, the basic idea is to match each integer from 1, ... , m with
an object.
Abstractly, the objects themselves will be identified with inte-
gers, so that both rankings and orderings will be permutations of
the first m integers. Through bitter experience, I know how easy
it is to confuse rankings and orderings. To alleviate some of the
confusion, rankings will generally be called "y" (or maybe ".:z:" or
"z"), and will reside in the space

Sm ={Permutations of the Ranks {1, 2, ... , m} }, {1.1)

whereas orderings will generally be called "w" and reside in the

space

Tm ={Permutations of the Objects {1, 2, ... , m} }. (1.2)

For either, there needs to be an a priori identification of the objects

with the integers.
We will typically be dealing with a sample of n rank vectors:

Y(1) , •.. ,y(n) E S m• {1.3)

 1.1 RANK DATA 3

What does one do with such rank data? Chapter 2 takes a data-
analytic approach. Rank data are multivariate data, the objects
representing the variables; hence any multivariate method can be
applied to the rank data. Means, standard deviations, histograms,
box plots, cluster analyses, multidimensional scaling, factor anal-
ysis, etc., all have potential to provide insight into the data. For
example, simple statistics such as the number of judges who rank
a particular object as first, or the number who prefer one object
to another, arise naturally.
Since rank data are so highly structured, methods that respect
the peculiarities of rank data can be especially useful. Our funda-
mental structure is the "permutation polytope" created by plotting
the rank vectors in Euclidean space, then connecting the points.
Thompson (1993a,b) represents a set of ranking data on the poly-
tope by placing a ball at each ranking, where the radius of the ball
is indicative of the frequency of that ranking in the sample. Much
of Chapter 2 is involved with trying to visualize (or numericize)
these multidimensional plots. Distances defined on Sm or Tm allow
defining the center and spread of a data set, and finding clusters
of judges. In the final section of Chapter 2, we consider functions
on Sm as vectors in m!-dimensional space, and try to decompose
them by projecting onto interesting subspaces. The popular de-
scriptive statistics fit into this framework. Diaconis (1988, 1989)
develops a spectral analysis along these lines, decomposing frequen-
cies into components based on the number of rankings with each
object attaining each rank, or each pair of objects attaining each
pair ofranks, etc. McCullagh (1993a) introduces inversions, which
similarly decompose the data, but into factors based on paired
comparisons of objects, triples of objects, etc.
Along with descriptions, a good statistician wants to be careful
not to see structure in the data that is basically just random varia-
tion. Chapter 3 provides some tools, mainly for testing whether the
uniform distribution is tenable. (Under the uniform distribution,
all rankings have the same probability.) The statistics in Chapter
2 are turned into formal test statistics. In addition, concordances
based on distance measures prove to be very useful. Chapter 4
continues by comparing groups of judges.
Modeling begins in Chapter 5. Some models arise from theoret-
ical constructs, some from experimental methods, and others from
attempts to find a simple description of the population of rankers.
The basic models are summarized, including those named after
 4 INTRODUCTION

Thurstone, Babington Smith, Bradley, Terry, Mallows, Plackett,

and Luce. Many of the models are built, either on purpose or by
happenstance, on the statistics in Chapters 2 and 3. Chapters 6
through 10 plunge more deeply into the models, finding specializa-
tions and generalizations, and methods for assessing fit.
Partial rankings, incomplete rankings, and tied rankings chal-
lenge the material up to Chapter 11. Rank data commonly arrive
in form different than n complete rankings of m objects. It may
be impractical to have every judge assess every object, so that
the experiment will present different judges with different subsets
of objects to rank. Often, people are asked to list only their fa-
vorite q out of m rankings. (Such partial rankings are called "q-
permutations" by Silverberg, 1980.) For example, in ranking col-
lege football teams, the judges are not expected to list all hundreds
or thousands of college teams, but just their top 25, or a similar
number. Judges are also wont to not provide full ranks even when
asked. For example, when the American Psychological Association
voted for President in 1980, the ballots asked for a complete rank-
ing of the candidates, but most people provided only their top one,
two, or three candidates. See Diaconis (1988, 1989). When ana-
lyzing such data, some models adapt well to some types of partial
rankings, but none adapt well to all types, and some to no types.
Rank data have dose connections with many other areas of
statistics. Nonparametric analysis makes heavy use of ranks. See
Gibbons and Chakraborti (1992). Paired-comparison analysis uses
data in the form of pairwise comparisons of several objects or treat-
ments, rather than complete rankings. See David (1988). Ordinal
analysis deals with variables whose values are given only on an
ordered scale. For example, a question on a survey may be "How
important is it for the government to fund education?" The answers
are chosen among the ratings very important, somewhat important,
slightly important, not important at all. There is a definite order to
the possible answers, but no obvious numerical scale. See Agresti
(1984).
The models and exploratory techniques covered in this book are
meant to be building blocks. Rankings may be only a part of the
data, or may be collected in complicated designs. The hope is that
students and researchers will use this book as a starting point for
further explorations.
 Bibliography

Abramowitl, M. and Stegun, I. A. {1972) Handbook of Mathematical

Functions. U.S. Government Printing Office: Washington, D.C.
Agresti, A. {1984) Analysis of Ordinal Categorical Data. Wiley: New
York.
Agresti, A. {1990) Categorical Data Analysis. Wiley: New York.
Alvo, M. and Cabilio, P. {1991) On the balanced incomplete block design
for rankings. Annals of Statistics, 19, 1597 - 1613.
Alvo, M. and Cabilio, P. {1995) Rank correlation methods for missing
data. To appear in Canadian Journal of Statistics.
Alvo, M. and Cabilio, P. {1995a) A note on the balanced incomplete block
design for rankings. Technical Report No. 200, Laboratory for Re-
search in Statistics and Probability, University of Ottawa and Carleton
University.
Alvo, M. and Cabilio, P. {1995b) Testing ordered alternatives in the pres-
ence of incomplete data. Journal of the American Statistical Associa-
tion, 80. To appear.
Alvo, M., Cabilio, P. and Feigin, P. D. {1982) Asymptotic theory for mea-
sures of concordance with special reference to average Kendall tau. An-
nals of Statistics, 10, 1269 - 1276.
Alwin, D. F. and Jackson, D. J. {1982) Adult values for children: an ap-
plication of factor analysis to ranked preference data. Social Structure
and Behavior, 311 - 329. Hauser, R. M., Mechanic, D., Haller, A. 0.
and Hauser, T. S., eds. Academic Press: New York.
Anderson, R. L. {1959) Use of contingency tables in the analysis of con-
sumer preference studies. Biometrics, U, 582 - 590.
Arrow, K. J. {1963) Social Choice and Individual Values. Wiley: New
York.
Baer, R. M. and Brock, P. (1968) Naturalsortingoverpermutationspaces.
Mathematics of Computation, 22, 385 - 410.
Barnes, S. H. and Kaase, M. {1979) Political Action: Mass Participation
in Five Western Countries. Sage: Beverly Hills.
BIBLIOGRAPHY 303

Barton, D. E., David, F. N. and Mallows, C. L. (1958) Non-randomness

in a sequence of two alternatives: I. Wilcoxon and allied test statistics.
Biometrika, 45, 166- 177.
Beckett, L. A. (1993) A censored ranking problem. Probability Models
and Statistical Analyses for Ranking Data, 92 - 107. Fligner, M. A.
and Verducci, J. S., eds. Springer-Verlag: New York.
Beran, R. and Walker, M. E. (1990) A .scaling approach to .social distance.
Presented at the Sixty-second Annual Meeting, Midwestern Psycholog-
ical Association, Chicago. May, 1990.
Bickel, P. J. and Doksum, K. A. (1977) Mathematical Statistics. Holden-
Day: New York.
Bishop, Y. M. M., Fienberg, S. E. and Holland, P. W. (1975) Discrete
Multivariate Analysis. MIT Press: Cambridge, Massachusetts.
Block, H. W., Chhetry, D., Fang, Z. and Sampson, A. R. (1990) Partial
orders on permutations and dependence orderings on bivariate empirical
distributions. Annals of Statistics, 18, 1840 - 1850.
Block, H. W., Chhetry, D., Fang, z. and Sampson, A. R. (1991) Partial
orderings on permutations. Topics in Statistical Dependence, 45 - 55.
Block, H. W., Sampson, A. R. and Savits, T. H. (eds.), IMS: Hayward,
California.
Block, H. W., Chhetry, D., Fang, Z. and Sampson, A. R. (1992) Some new
partial orderings on Sn and Sn x Sn. Journal of Statistical Planning
and Inference, ao, 351 - 357.
BOcltenholt, U. (1992) Thurstonianrepresentationfor partial ranking data.
British Journal of Mathematical and Statistical Psychology, 45, 31 -
49.
BOckenholt, U. (1993) Applications of Thurstonian models to ranking
data. Probability Models and Statistical Analyses for Ranking Data,
157- 172. Fligner, M. A. and Verducci, J. S., eds. Springer-Verlag:
New York.
Bockenholt, U. and Gaul, W. (1986) Analysis of choice behavior via prob-
abilistic ideal point and vector models. Applied Stochastic Models and
Data Analysis, 2, 209 - 226.
Box, G. E. P. and Tiao, G. C. (1973) Bayesian Inference in Statistical
Analysis. Addison-Wesley: Massachusetts.
Bradley, R. A. and Terry, M. A. (1952) Rank analysis of incomplete block
designs. I. Biometrika, II, 324 - 345.
Brady, H. (1989) Factor and ideal point analysis for interpersonally in-
comparable data. Psychometrih, 54, 181- 202.
Brook, D. and Upton, G. J. G. (1974) Biases in local government elections
due to position on the ballot paper. Applied Statistics, 21, 414 - 419.
Burr, E. J. (1960) The distribution of Kendall's scoreS for a pair of tied
rankings. Biometrika, 47, 151- 171.
304 BIBLIOGRAPHY

Chung, L. (1989) The Use of NonnuH Models lor Rank Data in Nonpara-
metric Statistics. Ph. D. Thesis, Statistics, University of lllinois at
Urbana-Champaign.
Chung, L. and Marden, J. I. (1990) Finding the marginal distribution of
ranks in some nonnull ranking models, with applications to goodness-
of-fit statistics. Department of Statistics, University of lllinois at Ur-
bana-Champaign.
Chung, L. and Marden, J. I. (1991) Use of nonnull models for rank statis-
tics in bivariate, two-sample, and analysis-of-variance problems. Jouz-
nal of the American Statistical Association, 88, 188 - 200.
Chung, L. and Marden, J. I. (1993) Extensions of Mallows'¢> model. Prob-
ability Models and Statistical Analyses lor Ranking Data, 108 - 139.
Fligner, M. A. and Verducci, J. S., eds. Springer-Verlag: New York.
Cleveland, W. S. (1985) The Elements of Graphing Data. Wadsworth:
Monterey.
Cochran, W. G. and Cox, G. M. (1957) Experimental Designs. Wiley:
New York.
Cohen, A. (1982) Analysis of large sets of ranking data. Communications
in Statistics- Theory and Methods, 11, 235 - 256.
Cohen, A. and Mallows, C. L. (1980) Analysis of ranking data. Technical
Report, Bell Laboratories, Murray Hill, New Jersey.
Cohen, A. and Mallows, C. L. {1983) Assessing goodness of fit of ranking
models to data. The Statistician, n, 361 - 373.
Coombs, C. H. (1964) A Theory of Data. Wiley: New York.
Critchlow, D. E. (1985) Metric Methods for Analyzing Partially Ranked
Data. Springer-Verlag: New York.
Critchlow, D. E. (1986) A unified approach to constructing nonparametric
rank tests. Technical Report #86-15, Department of Statistics, Purdue
University.
Critchlow, D. E. and Verducci, J. S. (1992) Detecting a trend in paired
rankings. Applied Statistics, 41, 17- 29.
Critchlow, D. E., Fligner, M. A. and Verducci, J. S. (1991) Probability
models on rankings. Jouznal of Mathematical Psychology, 32, 294-
373.
Critchlow, D. E. and Fligner, M. A. (1993) Ranking models with item
covariates. Probability Models and Statistical Analyses for Ranking
Data, 1 - 19. Fligner, M. A. and Verducci, J. S., eds. Springer-Verlag:
New York.
Croon, M. A. (1989a) The analysis of partial rankings by log-linear and
latent-class models. Multiway Data Analysis, 497- 506. North-Hol-
land: Amsterdam.
Croon, M. A. (1989b) Latent class models for the analysis of rankings.
New Developments in Psychological Choice Modeling, 99 - 121. Feger,
Klauer and de Soete, eds. North-Holland: Amsterdam.
BIBLIOGRAPHY 305

Croon, M. A. and Luijh, R. (1993) Latent structure models for ranking

data. Probability Models and Statistical Analyses for Ranl:ing Data,
53-74. Fligner, M.A. and Verducci, J. S., eds. Springer-Verlag: New
York.
Daniels, H. E. and Kendall, M. G. (1947) The significance of rank corre-
lations where parental correlation exists. Biometrih, S4, 197 - 208.
Daniels, H. E. (1950a) Rank correlation and population methods. Bio-
metrih, SS, 129 - 135.
Daniels, H. E. (1950b) Rank correlation and population models. Journal
oftlle Royal Statistical Society B, 13, 171- 181.
Darroch, J. N. and Ratclift", D. (1972) Generalised iterative scaling for
log-lineu models. Annals of Matllematical Statistics, 4S, 1470- 1480.
David, H. A. (1988) Tlle Metllod of Paired Comparisons. Oxford Univer-
sity Press: New York.
Davidson, R. R. (1970) On extending the Bradley-Terry model to accom-
modate ties in paired comparison experiments. Journal of tlle American
Statistical Association, 15, 317- 328.
DeSoete, G., Carroll, J. D. and DeSarbo, W. S. (1986) The wandering
ideal point model: a probabilistic multidimensional unfolding model
for paired comparison data. Journal of Matllematical Psychology, SO,
28-41.
Dempster, A. P., Laird, N. M. and Rubin, D. B. (1977) Maximum like-
lihood from incomplete data via the EM algorithm (with discussion).
Journal of tlle Royal Statistical Society B, st, 1 - 38.
Diaconis, P. (1988) Group Representations in Probability and Statistics.
Institute of Mathematical Statistics: Hayward, California.
Diaconis, P. (1989) A generalisation of spectral analysis with application
to ranked data. Annals of Statistics, 1 '1, 949- 979.
Diaconis, P. and Graham, R. L. (1977) Spearman's footrule as a measure
of disarray. Journal of tlle Royal Statistical Society B, st, 262 - 268.
Duncan, 0. D. and Brody, C. (1982) Analysing n ranltings of three items.
Social Structure and Bellavior, 269- 310. Hauser, R. M., Mechanic,
D., Haller, A. 0. and Hauser, T. S., eds. Academic Press: New York.
Durbin, J. (1951) Incomplete blocks in ranking experiments. Britisll Jour-
nal of Psychology, 4, 85 - 90.
Durbin, J. and Stuart, A. (1951) Inversions and rank correlations. Journal
of tlle Royal Statistical Society B, lS, 303 - 309.
Efron, B. (1982) Tlle Jacllcnife, tlle Bootstrap and Otller Resampling
Plans. SIAM: Philadelphia.
Ehrenberg, A. C. S. (1952) On sampling from a population of rankers.
Biometrih, Sl, 82 - 87.
Everitt, B.S. and Hand, D. J. (1981) Finite Mixture Distributions. Chap-
man and Hall: London.
Feigin, P. and Cohen, A. (1978) On a model of concordance between
judges. Journal of tlle Royal Statistical Association B, 40, 203 - 213.
308 BIBLIOGRAPHY

Fei.pn, P. D. (1993) Modelling and analysing paired ranking data. Prob-

abili'y Model. and S'atistical Analyses for Ranling Data, 75 - 91.
Fligner, M.A. and Verducci, J. S., eds. Springer-Verlag: New York.
Feigin, P. D. and Alvo, M. (1986) Intergroup diversity and concordance
for ranking data: an approach via metrics £or permutations. Annals of
S'atistics, 14, 691 - 707.
Feller, W. (1968) An Introduction to Probability Theory and Its Applica-
tions. Wiley: New York.
Fienberg, S. (1971) Randomization and social affairs: the 1970 draft lot-
tery. Science, 171, 255 - 261.
Fishburn, P. C. and Brams, S. J. (1983) Paradoxes o£ preferential voting.
Mathematics Maguine, 58, 207- 214.
Fligner, M. A. and Verducci, J. S. (1990) Posterior probabilities £or a
consensus ordering. Psychometrih, 65, 53 - 63.
Fligner, M.A., and Verducci, J. S. (1986) Distance based ranking models.
Journal of the Royal Statistical Society B, 48, 359 - 369.
Fligner, M. A., and Verducci, J. S. (1988) Multistage ranking models.
Journal of the American Statistical Association, 8S, 892 - 901.
Fligner, M. A., and Verducci, J. S., eds. (1993) Probability Models and
S'atistical Analyses for Ranking Data. Springer-Verlag: New York.
Friedman, M. (1937) The use o£ranks to avoid the assumption of normality
implicit in the analysis o£ variance. Journal of the American Statistical
Association, S2, 675 - 701.
Gabriel, K. R. (1982) Biplot. Encyclopedia of Statistical Science, 1, 263
-271.
Gibbons, J. D. and Chakraborti, S. (1992) Nonparametric Statistical In-
ference. Deller, New York.
Goldberg, A. I. (1976) The relevance o£ cosmopolitan/local orientations to
professional values and behavior. Sociology of Worlc and Occupation,
S, 331- 356.
Goodman, L. A. (1969) How to ransack social mobility tables and other
kinds o£ cross-classification tables. The American Journal of Sociology,
n, 1-39.
Gordon, A. D. (1979) A measure o£ the agreement between rankings. Bio-
metrih, 88, 7 - 15.
Gurari, E. M. (1994) TeX and LATeX: Drawing and Literate Program-
ming. McGraw Hill: New York.
Hartigan, J. A. (1975) Clustering Algorithms. Wiley: New York.
Hays, W. L. (1960) A note on average tau as a measure o£ concordance.
Journal of the American Statistical Association, 65, 331 - 341.
Henery, R. J. (1981) Permutation probabilities as models £or horse races.
Journal of the Royal Statistical Society B, 4S, 86 - 91.
Henery,R. J. (1983} Permutation probabilities £or gamma random vari-
ables. Applied Probability, 20, 822- 834.
Herstein, I. N. (1964) Topics in Algebra. Blaisdell: Massachusetts.
BIBLIOGRAPHY 307

Hoefl'ding, W. (1951) A combinatorial central limit theorem. Annals of

Mathematical Statistics, 22, 558 - 566.
Hogg, R. V. and Craig, A. T. (1978) Introduction to Mathematical Statis-
tics. MacMillan: New York.
Hollander, M. and Sethuraman, J. (1978) Testing Cor agreement between
groups of judges. Biometrika, 15, 403 - 411.
Huber, P. J. (1985) Projection pursuit. Annals of Statistics, 11, 435 -
475.
Joe, H. and Verducci, J. S. (1993) Babington Smith models. Probability
Models and Statistical Analyses for Ranking Data, 37- 52. Fligner, M.
A. and Verducci, J. S., eds. Springer-Verlag: New York.
Kendall, M. G. and Smith, B. Babington (1940) On the method of paired
comparisons. Biometrika, 11, 324 - 345.
Kendall, M. G., and Gibbons, J. D. (1990) Rank Correlation Methods.
Edward Arnold: London.
Kerov, S. and Vershiclt, A. (1985) The characters of the infinite symmet-
ric group and probability properties of the Robinson-Schensted-Knuth
algorithm. SIAM Journal on Algebraic and Discrete Methods, 7, 116 -
124.
Kruskal, W. H. (1958) Ordinal measures of association. Journal of the
American Statistical Association, 51, 814 - 861.
Lantermann, E. D. and Feger, H., eds. (1980) Similarity and Choice:
Papers in Honour of Clyde Coombs. Huber: Bern.
Lehmann, E. L. (1975) Statistical Methods Based on Ranlcs. Holden-Day:
San Francisco.
Lehmann, E. L. (1983) Theory of Point Estimation. Wiley: New York.
Li, L. and Schucany, W. R. (1975) Some properties of a test Cor concor-
dance of two groups of rankings. Biometrika, 12, 417- 423.
Lindsay, B. G. (1983a) The geometry of mixture likelihoods: a general
theory. Annals of Statistics, 11, 86- 94.
Lindsay, B. G. (1983b) The geometry of mixture likelihoods, Part II: the
exponential family. Annals of Statistics, 11, 783 - 792.
Luce, R. D. (1959) Individual Choice Behavior. Wiley: New York.
Luce, R. D. (1977) Thurstone's discriminal processes fifty years later. Psy-
cllometrib, 43, 461 - 489.
Mallows, C. L. (1957) Non-null ranking models: I. Biometrika, 44, 114-
130.
Mann, H. B. (1945) Non-parametric tests against trends. Econometrica,
n, 245-259.
Marden, J. I. (1992) Use of nested orthogonal contrasts in analyzing rank
data. Journal of the American Statistical Association, 87, 307 - 318.
Marden, J. I. (1994) Comparing distances for rank vectors. Department
of Statistics, University of Dlinois at Urbana-Champaign.
Mardia, K. V., Kent, J. T. and Bibby, J. M. (1979) Multivariate Analysis.
Academic Press: London.
308 BIBLIOGRAPHY

Marley, A. A. J. (1993) Aggregation theorems and the combination ol

probabilistic rank orders. Probability Models and Statistical Analyses
for Ranlcing Data, 216- 240. Fligner, M. A. and Verducci, J. S., eds.
Springer-Verlag: New York.
Marshall, A. W. and Olltin, I. (1979) Inequalities: Theory of Majorisation
and it. Applications. Academic Press: New York.
McCullagh, P. (1993a) Permutations and regression models. Probability
Models and Statistical Analyses for Ranlcing Data, 196- 215. Fligner,
M.A. and Verducci, J. S., eds. Springer-Verlag: New York.
McCullagh, P. (1993b) Models on spheres and models lor permutations.
Probability Models and Statistical Analyses for Ranlcing Data, 278 -
283. Fligner, M. A. and Verducci, J. S., eds. Springer-Verlag: New
York.
McCullagh, P. and Nelder, J. A. ( 1989) Generalised Linear Models. Chap-
man· and Hall: London.
McLachlan, G. J. and Basford, K. E. (1988) Mixture Models. Marcel
Deller: New York.
Moran, P. A. P. (1950) Recent developments in ranking theory. Journal
of the Royal Statistical Society B, 12, 153 - 162.
Nombekela, S. W., Murphy, M. R., Gonyou, H. W. and Marden, J. I.
(1993) Dietary preferences in early lactation cows as affected by primary
tastes and some common feed flavors. Journal of D&Uy Science, 77,
2393- 2399.
Pendergrass, R. N. and Bradley, R. A. (1960) Ranking in triple compar-
isons. Contributions to Probability and Statistics, 331 - 351. Olltin,
0., Ghurye, S. G., Hoeffding, W., Madow, W. G. and Mann, H. B., eds.
Stanford University Pres.
Plackett, R. L. (1975) The analysis o£ permutations. Applied Statistics,
24, 193 - 202.
Press, W. H., Teultolsky, S. A., Vetterling, W. T. and Flannery, B. P.
(1992) Numerical Recipes in C. Cambridge University Press.
Rao, C. R. (1973) Linear Statistical Inference and its Applications. Wiley:
New York.
Rao, C. R. (1982) Diversity and dissimilarity coefficients: a unified ap-
proach. Journal of Theoretical Population Biology, 21, 24- 43.
Rao, P. V. and Kupper, L. L. (1967) Ties in paired-comparison exper-
iments: a generalization of the Bradley-Terry model. Journal of the
American Statistical Association, 82, 194 - 204.
Rodrigues, 0. (1839) Lowville Journal of Mathematics, 4, 236.
Rogers, T. (1991) Students as literary critics: the interpretive experiences,
beliefs, and processes o£ ninth-grade students. Journal of Reading Be-
havior, 2S, 391 - 423.
Santner, T. J. and Duffy, D. E. (1989) The Statistical Analysis of Discrete
Data. Springer-Verlag: New York.
BIBLIOGRAPHY 309

Schucany, W. R. and Frawley, W. H. {1973) A rank test {or two group

concordance. Psychometrib, 18, 249 - 258.
Schulman, R. R. {1979) Ordinal Data: an alternative distribution. Psy-
chometrib, 4-&, 3 - 20.
Serfling, R. J. {1980) Approximation TheoremsofMathematicalStatistics.
Wiley: New York.
Silverberg, A. R. (1980) Statistical Models for q-permutations. Ph. D.
Thesis, Department o{ Statistics, Princeton University.
Silverberg, A. R. {1984) Statistical models for q-permutations. Proceed-
ings of the Biopharmaceutical Section, American Statistical Auocia-
tion., 107 - 112.
Smith, B. Babington {1950) Discussion o{ Professor Ross's paper. Journal
of the Royal Statistical Society B, 13, 53 - 56.
Stem, H. {1990) Models for distributions on permutations. Journal of the
American Statistical Association, 85, 558 - 564.
Stem, H. {1993) Probability models on rankings and the electoral process.
Probability Models and Statistical Analyses for Ranlcing Data, 173 -
195. Fligner, M. A. and Verducci, J. S., eds. Springer-Verlag: New
York.
Tallis, G. M. and Dansie, B. R. {1983) An alternative approach to the
analysis o{ permutations. Applied Statistics, 13, 110- 114.
Tanner, M. A. {1993) Tools for Statistical Inference. Springer-Verlag:
New York.
Terpstra, T. J. (1952) The asymptotic normality and consistency o{ Ken-
dall's test against trend, when ties are present in one ranking. Indaga-
tio.nes Mathematicae, 14, 327 - 333.
Thompson, G. L. (1993a) Generalised permutation polytopes and explor-
atory graphical methods for ranked data. Annals of Statistics, 21, 1401
- 1430.
Thompson, G. L. (1993b) Graphical techniques for ranked data. Prob-
ability Models and Statistical Analyses for Ranling Data, 294 - 298.
Fligner, M.A. and Verducci, J. S., eds. Springer-Verlag: New York.
Thurstone, L. L. (1927) A law o{ comparative judgement. Psychological
Reriews, 1-&, 273 - 286.
Tubbs, J. D. (1993) Assessing changes in repeated ranked lists. Depart-
ment o{ Mathematical Sciences, University o{ Arkansas.
Upton, G. J. G. and Brook, D. (1975) The determination of the optimum
position on a ballot paper. Applied Statistics, 2-&, 279- 287.
Ury, H. and Kleinecke, D. (1979) Tables o{ the distribution o{ Spearman's
footrule. Applied Statistics, 28, 271 - 275.
Vals, P. D. and Thompson, M. E. (1993) Exact inference for Kendall's S
and Spearman's p with extension to Fisher's exact test in r X c contin-
gency tables. Technical Report STAT-93-05, University o{ Waterloo.
Van Blokland-Vogelesang, A. W. (1989) Unfolding and consensus rank-
ing: a prestige ladder for technical occupations. New Developments in
310 BIBLIOGRAPHY

Psychological C.hoice Modeling, 237 - 258. De Soete, G., Feger, H. and

Klauer, K. C., eels. North-Holland: ~terdam.
Van Blokland-Vogelesang, A. W. (1993) A nonparametric distance model
for unidimensional unfolding. Probability Models and Statistical Anal-
yses for .Ranling Data, 241- 276. Fligner, M. A. and Verducci, J. S.,
eels. Springer-Verlag: New York.
Van Blokland-Vogelesang, A. W. and Van Blokland, P. J. (1990) UN-
FOLD: unidimensional unfolding of preference data. User's Manual. I.
E. C. ProGAMMA: Groningen, The Netherlands.
Verducci, J. S. (1982) Discriminating between two probabilities on t.he ba-
sis of ranled preferences. Ph.D. Thesis, Statistics, Stanford University.
Verducci, J. S. (1989) Minimum majorilation decomposition. Contribu-
tions to Probability and Statistics, 160- 173. Gieser, L. J., Perlman,
M.D., Press, S. J., and Sampson, A. R., eels. Springer-Verlag: New
York.
Walker, M. E. (1993) Social distance: analysis of race preferences. Un-
published manuscript, University of Illinois at Urbana-Champaign.
Weisberg, S. (1985) Applied Linear Regression. Wiley: New York.
Whitfield, J. W. (1950) Uses oCthe ranking method in psychology. Journal
of t.he Royal Statistical Society, 12, 163 - 170.
Wu, C. F. J. (1983) On the convergence properties of the EM algorithm.
Annals of Statistics, 11, 95 - 103.
Yao, K. G. (1994) Bayesian estimation o{ Thurstonian ranking models:
applications via the Gibbs sampling and the importance sampling al-
gorithms. Department o{ Psychology, University o£ Illinois at Urbana-
Champaign.
Ye, J. and McCullagh, P. (1993) Matched pairs and ranked data. Prob-
ability Models and Statistical Analyses for Ranking Data, 299 - 306.
Fligner, M.A. and Verducci, J. S., eels. Springer-Verlag: New York.
Yellott, J. (1977) The relationship between Luce's choice axiom, Thur-
stone's theory of comparative judgment, and the double exponential
distribution. Journal of Mathematical Psychology, U, 109- 144.
Yemelichev, V. A., Kovalev, M. M. and Kravtsov, M. K. (1984) Polytopes,
Grap.hs and Optimisation. Cambridge University Press.