,Md/ Compur. MotMin~, Vol. 12. No. 415, pp. 397-404, 1989 0x95-7177,x9 $3.00 + 0.

00
Printed in Great Britain. All nghts reserved Copyright I! 1989 Pergamon Press plc

FORMAL THEORIES OF POLITICS:

THE SCOPE OF MATHEMATICAL MODELLING IN
POLITICAL SCIENCE

PAUL E. JOHNSON
Department of Political Science, University of Kansas, Lawrence, KS 66045, U.S.A

In a very brief period of time, three or four decades, mathematical political science has covered
a great deal of territory. This special issue presents a broad array of the mathematical approaches
which are used in political science. The intention of the issue is to convey a sense of the questions
and methods which govern the political science research agenda. The authors have been encouraged
to draft their presentations for the general scientific audience, rather than specialists in political
science.?
A BRIEF HISTORY OF MODELS IN POLITICAL SCIENCE

Mathematical models first became popular in political science in the 195Os, when inferential
statistics came into common usage. The 1950s and 1960s are referred to as the behavioral era,
because most effort was focused on the detection of empirical patterns in voting behavior and
public opinion data. A considerable infrastructure has developed to administer and make available
national surveys on a regular basis. Also during the 195Os, however, the roots of modern
mathematical modelling were established. Though there is still a great deal of interest in empirical
results, there has been a substantial growth in emphasis on mathematical theory as a technique
for the derivation of testable hypotheses. There are two major mathematical approaches, called
political economy and systems modelling.
The fundamental axioms of the political economy approach are that individuals in a political
system are rational actors (they have well-defined preferences and behave in accordance with them)
and that social outcomes (equilibria) result from the interaction of these individuals within
the constraints imposed by social institutions. Models in this tradition have three technical
components: individual behavioral principles, institutional structures and an equilibrium (or, more
generally, solution) concept. An equilibrium is a situation which exists when none of the actors
with power to effectuate change choose to do so. Observably stable or repetitive political events
are explained by equilibrium concepts.
This approach, which is also called public choice or rational choice, has thrived on the healthy
interaction of political scientists and economists. Though the early research mainly concerned
legislatures and elections, the field speaks to a much larger set of topics, including
bureaucratic-legislative relations, international alliances and war, interest groups and presidential
power. Game theory is the mode of investigation which unifies these studies.
The second approach, which I call systems modelling, for lack of a more appropriate term,
examines properties of systems and their changes in time. In contrast to the emphasis on individual
behavior in the political economy tradition, these approaches theorize about aggregates or systemic
variables. There are three branches of inquiry in this area. Dynamic modelling is the name used
to refer to a genre of studies in which a systems states are hypothesized to follow a set of differential
equations. The existence and stability of equilibria are examined. This field has benefitted
substantially from interaction with biologists and other natural scientists. The second systems
approach may be called probability modelling. Standard techniques for the analysis of random or
stochastic processes are applied to political processes. The third, and newest, field, is artificial
intelligence. The range of this research is quite broad and, in many ways. it offers a potential bridge

tThe references in the articles are presented in the style of the American Political Science Review, a leading journal in the
field. We refer to articles by the authors last name, the date and the page, e.g. (Arrow 1963, p. 19), and the references
appear alphabetically at the end of each article.
 398 PAUL E. JOHNSON

between political economy and systems models, in that systems too complex to manage analytically
might yield to computer analysis.

ABOUT THE ARTICLES

Each of the major research areas is broken into three areas of specialization for the purposes
of this issue. Even active political scientists will find subjects with which they are unfamiliar. In
order to broaden the appeal of their articles to nonspecialists (in and outside of political science),
authors have been asked to clearly state their research questions and methodology. As a result,
the articles are generally self-contained, though a brief explanation of their orientation will
probably help clarify matters.

1. Multidimensional Models of Legislative Decision

The first two sections of the issue, which address legislatures and elections, respectively, draw
their substantive motivation from the theory of social choice. The basic assumption is that
democracy should be studied as an aggregation of individual tastes. Voter/participants are
hypothesized to be rational, in the sense that they have transitive preference orderings over the
alternatives from which they choose. That is, if P, is a binary preferred to relation for a voter
iEN(whereN={l,2,..., H}) and the alternatives include x, y and Z, then transitivity means that
xP,y and yP,z *xP,z. Majority rule, the most frequently investigated method of social choice,
generates a binary social preference relation, M. This relation? is xMy iff 1{i EN: xP,y} / > n/2.
The main part of the research agenda in this field concerns the transitivity of group choice and
the existence of an equilibrium in a voting scheme. In brief, though Vi, P, is transitive, M may not
be. The argument is inspired, in part, by a widely noted paradox of voting:1 given transitive
preferences for three voters,

xP,y and yP,z yP,z and zP,x zP,x and xP,y,

under majority rule, x beats y, z beats x, but, surprisingly, y beats z. When the social order is
intransitive (there is a cycle yMzMxMy), the social decision is indeterminate-there is no
meaningful way to infer the will of the people from the voting process because every proposal
can be defeated by another one. Kenneth Arrow, a Nobel prize winner, presented the famous
Impossibility Theorem which showed that intransitivity is generic to all nondictatorial social
choice rules which meet certain minimal criteria of democratic government (Arrow 1951, 1963).
Another way of putting this is that there is always a set of rationally structured individual
preferences which causes a binary social preference relation to be irrational or incoherent. Excellent
introductory discussions of the theorem can be found in Ordeshook (1986) and Riker (1982).
For some time it was believed that the intransitivity problem could be minimized by giving the
policy alternatives some geometric structure. Duncan Black (1948, 1958) presented what is now
called the spatial political model. The typical spatial model begins with the assumption that the
set of alternatives is a convex subset of Euclidean n-space called X. The desirability of a point x E X
to voter i depends on the distance between x and the voters most-preferred point, XT E X, according
to a function U,(x; x,*). A voters preferences are said to be single-peaked if Uj is quasi-concave
in X. Blacks solution to the paradox of voting is called the Median Voter Theorem: if the choice
space is one-dimensional and voter preferences are single-peaked, then under majority rule the
median of the voters ideal points is an equilibrium (it defeats all other proposals) and, furthermore,
majority rule generates a transitive binary social preference relation. The single-peakedness
requirement is thought to be resaonable-people generally have a favorite policy alternative and
movements along a line going away from the favorite policy are not preferred.
The main limitation on the Median Voter Theorem is the dimensionality of the social choice
space. Most voting processes in reality seem to range over spaces of considerably higher
dimensionality. The literature appearing in the 1970s and 1980s on cycles and multidimensional
models is immense and mathematically sophisticated. Cycling in higher dimensions usually cannot

*If voter indifference and abstention are taken into account, this definition must be modified, but not in a substantial way.
:This paradox is attributed to the eighteenth century writings of the Marquis de Condorcet (see Arrow 1963, p. 93).
 Mathematical modelling in political science 399

be avoided unless much stronger assumptions than single-peakedness are imposed (see Plott 1967;
Kramer 1973). Since these much stronger conditions are not thought to prevail in reality, the result
implies that majority rule decisions are generally intransitive. More surprisingly, if preferences are
single-peaked (quasi-concave) in the multidimensional space, the so-called chaos theorems
indicate that, if no point exists which is majority preferred to all other points, then a cycle exists
between any two points in the space. As McKelvey (1976, p. 475) put it, when transitivity breaks
down, it completely breaks down, engulfing the whole space in a single cycle set. It is now known
that, in spaces of sufficiently high dimension, a continuous trajectory of proposals exists leading
from one point to any other, along which each point defeats the immediately preceding point
(Schofield 1978). Later research convincingly demonstrates that equilibria generically (in the
measure theoretic sense: almost always) do not exist (McKelvey and Schofield 1986, 1987). To
summarize, it has been shown that the Median Voter Theorem cannot be extended to a
multidimensional setting.
The pervasiveness and severity of voting cycles is considerd to be one of the most important
problems in the field. In this volume, Schofield, one of the important contributors to this line of
research, summarizes the results on cycles and presents a variety of results which link the extent
of cycling to the dimensionality of the social choice space. One implication of these theorems is
that competing parties in parliamentary systems of government have incentives to attempt to alter
the dimensionality of the choice space in order to enhance their position. The article by Feld,
Grofman and Miller emphasizes the stabilizing impact of political institutions-congressional
procedures and voting agendas. This research lies in a tradition called the new institutionalism,
which is associated with an important article by Shepsle (1979). The viewpoint is that emphasis
on the institution-free properties of a majority preference relation, as in the chaos literature,
overlooks the most important factors which affect social choice.

2. Spatial Competition Among Candidates for Public Ofice

The major difference between the research on legislatures and elections is that legislatures allow
the members to propose alternatives on an agenda, while election models constrain voters to choose
between the offerings of the candidates. Studies of electoral behavior following Downs (1957) An
Economic Theory of Democracy, examine competitive position-taking by candidates. In a one-
dimensional model of a two-party competition for votes, Downs found that an equilibrium exists
in which both parties announce positions which are equal to the median of the voters ideal points.
Hence, Black and Downs arrived at the same observation: the median is a solution to electoral
competition in one dimension.
The research questions which arise in the subsequent studies concern electoral equilibrium and
candidate convergence. On the one hand, observed candidate behavior is substantially more
regular than the chaos theorems discussed in the previous section imply. On the basis of those
results, we would expect to see candidates frantically changing their positions to defeat each other.
On the other hand, since Presidential candidates are not identical (there is not convergence to the
median voters ideal point), it must be that the median voter result is in need of revision. For the
past 20 years or so, research has been concentrated on examination of variations in the candidates
and voters incentives, as well as the institutional structures which govern elections.
Articles on candidate position-taking in this issue examine a number of factors. Aldrich and
McGinnis examine the restrictions on candidate behavior which result from the introduction of
party membership as a factor in multidimensional position-taking. Their article shows that, by
taking into account the relationship between candidates and their partisan followers, the stability
of nonmedian, nonconvergent candidate positions can be explained. Until recently, these models
of candidate competition have supposed that the voters accept the candidates declarations of
policy at face value, without examining the incumbents record. Ingbermans article explores the
importance of incumbency and reputation in restricting the movement of candidates. Glazer,
Grofman and Owen introduce a model in which candidates do not know voter preferences and
are thus forced to compete under conditions of incomplete information. On the voters side of the
electoral model, Enelow and Hinich provide a probabilistic model of voter behavior and
demonstrate the implications with a sophisticated analysis of data taken from national voter
surveys during the Presidential election of 1980. One shortcoming of these approaches is that only
400 PAUL E. JOHNSON

majority rule type electoral institutions are discussed. In a unique paper, Cox shows that we can
make meaningful statements about candidate strategies for a variety of electoral systems. It is
agreed that candidates choose from the set undominated strategies when taking electoral positions.
Cox shows bounds on the undominated set which depend on the voting rule.

3. Game and Decision Theory

Legislatures and elections have been the dominant fields of inquiry in political economy. The
methods of game and decision theory, however, can be applied more generally to situations that
involve strategy choice or multiperson decision making. Decision theory is the study of individual
choice behavior when the behavior of others is taken to be exogenous, while game theory is the
study of N decision makers whose payoffs are jointly determined by their choices.
The field of game theory began with the monumental accomplishment of Von Neumann and
Morgenstern (1944). This book made many lasting contributions, including results for zero sum
games, n-person cooperative games and simple games. In addition, they provided a utility theory
for decision making in risky situations (which are represented by lotteries?). The second edition
(1947) presents a proof of the expected utility theorem, which states that under some general
conditions, a persons utility for a lottery may be represented by the probability-weighted sum of
the utilities of the prices which are offered in the lottery. This is a result of the utmost importance
in the theory of decisions and games. A second landmark is the contribution to noncooperativef
game theory of Nash (1950). A Nash equilibrium is a set of player actions which are self-
reinforcing-no player will unilaterally alter his choice if the others remain in equilibrium. This
equilibrium concept offers an appealing description of stable political interaction. Nash showed
that existence of equilibria is equivalent to existence of fixed point in vector-valued best reply
mappings (i.e. each players equilibrium strategy is a best reply to the others equilibrium strategies).
He also noted that the fixed-point theorem of Kakutani can be used to prove existence of equilibria
in many noncooperative games. Game theorists have constantly attempted to generalize the
applicability of these basic concepts to situations of incomplete information (players do not know
each others preferences or strategies) and sequential interaction. The most widely read intro-
ductory text in game theory has been Lute and Raiffa (1957), though as it has become dated, a
number of helpful tests have become available (Friedman 1986; Owen 1982; Ordeshook 1986).
Although the studies of legislatures and elections previously described have game theoretic
features, they are distinguished by their spatial orientation and research questions. Game and
decision theory have recently been applied in a variety of other areas which deserve some attention.
A notable application of decision theory arose in attempts to resolve the paradox of not voting:
people vote, even though the impact of their votes on the electoral outcome is negligible and the
costs of voting are substantial (Riker and Ordeshook 1968; Ferejohn and Fiorina 1975). This is
still an important problem in incomplete information (Bayesian) game theory (Palfrey and
Rosenthal 1985). Decision theory has fruitfully been applied in a study of principal-agent
relationships in public bureaucracies (Bendor, Taylor and van Gaalen 1987). Bueno de Mesquitas
(198 1) precise accounting of war initiation by nation-states is an excellent application of decision
theory. Game theoretic analyses are used to study bureaucratic-legislative interaction (Niskanen
1971; Miller and Moe 1983) international crises (Brams and Kilgour 1987; Powell 1987) and the
evolution of cooperation (Axelrod 1984) among other things.
In this volume there is one decision theory model and two different game theory models.
Calverts article is a decision theoretic analysis of political learning. The important insight of the
paper pertains to the way rational actors make choices when information is costly and available
only in increments. The paper derives from the principles of full rationality, a result consistent
with observations of bounded (or poorly informed) informed rationality. The article by
Gradstein and Nitzan describes competition among interest groups for govenmental favors, which
are called rents (hence the jargon, rent-seeking). The authors use the Nash equilibrium concept
to describe configurations of political competition. The article contributes to our understanding

?A lottery is a list of possible prizes with the probability of each prize attached.
$Game theoretic models are differentiated by the amount of cooperation which is allowed between the players. A game
is called noncooperative if the players are not allowed to make binding agreements (this means, in effect, they cannot
be bound to comply with agreements which are not in their interest).
 Mathematical modelling in political science 401

of competition among a plurality of organizations for a plurality of prizes. Niou and Ordeshook
offer a cooperative game theoretic approach to international alliances. Their approach incorporates
the effect of geography on the ability of certain members in alliances to stabilize the international
balance of power.

4. Dynamic Models
This section begins the presentation of systems models. It is difficult to succinctly summarize the
research questions which govern this field, mainly because the authors share a perspective rather
than a research agenda. The unifying concepts are drawn from the methodology of dynamic
systems (see Cortes, Przeworski and Sprague 1974; Hanneman 1988). Research in this field typically
emphasizes development more than equilibrium and aggregates more than individuals. The
emphasis on aggregates is mainly due to the research questions which inspire these studies.
Proponents of the systems perspective generally believe that static (rational choice) models can not
explain change except with reference to exogenous variables. Systems modellers believe that change
is endogenous, a result of adjustment processes which should be explicitly represented. Static
models, which rely on comparative statics, are inherently incapable of explaining change.
The emphasis on aggregation is often seen as a rejection of individual level analysis (political
economy), though I do not believe this is necessarily so. The macro/micro controversy has not yet
been settled. Some scholars believe that individual behavior is not sufficiently well-understood or
regularized to be the primitive concept of social science research. In this view, aggregated analysis
is the only option. A second school of thought, to which I am partial, is ambivalent. Aggregate
patterns may, but need not, have a basis in individual behavior. In the same way that the physics
of large bodies can be studied apart from the movement of the atomic particles of which they are
composed, meaningful statements can be made about systems without regard to individual
behaviors. There is reason for hope that the micro/macro gap will be bridged. Notable accom-
plishments in this field are attributed to Schelling (1971, 1978) and Goodman (1957), who
demonstrated tractable methods by which to deal with aggregation prob1ems.t
Dynamic models are well-known in the field of international relations. Lewis Richardsons model
of an arms race between two nations, first drafted in the 193Os, was posthumously brought to the
forefront of political science by Anatol Rapoport (1957) and by the publication of Richardsons
(1960) Arrrzs and Insecurity. Richardson hypothesized that two countries armament levels, x and
y, change in a way which is represented by a system of differential equations:

dx /dt = ay - rn.y + g,

dyldt = bx - ny + h.

The terms dx/dt and dy/dt are time derivatives which represent changes in armament levels. The
constants a, b, m, n are assumed positive, while g and h, which represent grievances may be
positive or negative. Richardsons analysis of stability and the time paths of the armament levels
drew attention to the power of this kind of analysis.
Rapoport (1960, p. xi) called this approach social physics, because the method is similar to
that of mathematical physics: abstract, but plausible assumptions represent a process underlying
social behavior and the derived consequences are the end product. Rapoport also credits biologist
Nicolas Rashevsky (1948, 1951) in the development of this approach to social relations. Dynamic
analysis of arms acquisitions are other international processes has generated a large literature which
considers a variety of different dynamic specifications and methods of parameter estimation [e.g.
Gillespie et al. 1977; Ward 1984; Schrodt et al. 1977; for a variety of approaches, see Luterbacher
and Ward (1985)]. Though earlier research in international relations drew on biological models,
recent literature is more closely linked to operations research and engineering. For example, the
MIT world simulation projects, which modelled world growth patterns, drew a substantial amount
of attention to dynamic modelling (Forrester 1971; Meadows et al. 1972).
The use of dynamic models has slowly expanded beyond international relations to a range of
fields. A host of dynamic studies of mass behavior appeared in the 1960s. William McPhee (1963)

tThe levels of analysis problem is driving important research on parameter estimation in models of the relationship between
the economy (a variable with aggregate and individual level components) and individual voting behavior (Kramer 1983).
402 PAUL E. JOHNSON

made a substantial contribution with the publication of a general volume of dynamic models and
some essays on dynamics of voting behavior (in McPhee and Glaser 1962). By that time, the links
between this kind of model and biological models were recognized (Kemeny and Snell 1962;
Coleman 1964) [see also May (1974), for an example of the kind of biological research which has
substantially affected political science]. Some classic studies of aggregate patterns in voting
behavior used the dynamic logic, though the flavor of dynamic models was used without most of
the formalism (Key 1949; Converse 1969; Stokes and Iverson 1966). Considerably more advanced
approaches to mass behavior and public opinion phenomena have recently appeared (see MacKuen
1981; Przeworski and Sprague 1986; Brown 1987). I believe that the models in these articles will
be familiar to most system scientists.?
There are two straightforward examples of dynamic models in this volume by leaders in the field.
Huckfeldts article posits a dynamic relationship between law enforcement activity and public
compliance with unpopular laws. One of the interesting propositions in this paper is that moderate
law enforcement can bring about the highest levels of long run compliance. Compliance with
unpopular laws depends on the legitimacy of the political system, which is reduced by aggressive
punishment of noncompliance. Tsebelis and Sprague propose a model, based on the predator-prey
logic, of the linkages between coercion and support for revolution among large populations. One
of the important contributions of this piece is clarification of a long-stanging dispute concerning
the relative merits of longitudinal and cross-sectional analysis in the study of revolutionary
phenomena.

5. Probability Models
A second systems level approach uses probability and stochastic process concepts. These models
are less widely used than the dynamic models discussed in the previous section, though they are
increasingly popular. Important surveys of applications to social systems of probability models (as
well as dynamic models) are Coleman (1964) and Rapoport (1983).
In this issue, Cioffi-Revilla presents a set of models which describe war-related phenomena in
the international political system. The author shows that there are properties of war which can only
be understood from a mathematical perspective. This paper received the best paper award in the
Symposium on Systems Engineering and Peace Research at the recent European Meeting on
Cybernetic and SyStems Research. Midlarskys derivation of the Pareto distribution as a description
of inequality in land distributions is important for a variety of reasons. Previous studies assumed,
but did not derive, an exponential distribution. Midlarsky shows that, in addition to possessing
theoretical motivation, the Pareto (or log-exponential) distribution fits observed patterns of land
holdings.

6. Artljicial Intelligence
This is the newest area of research, and unfortunately, the least familiar to political modellers
with a general background (including the editor of this Special Issue). Aside from expert systems
to aid decision makers and budgeting in applied political science (called public administration), the
method has found most of its applications in international relations. The artificial intelligence
models are distinguished from other systems models because they specify processes through
algorithms, rather than equations. Almost invariably, models are implemented on digital comput-
ers. The models are particularly useful in dealing with three characteristics of human judgment
which have not been successfully incorporated into the traditional models of human decision. These
characteristics are the use of a large amount of memory, sequential information processing and
the general process of learning. In this issue, Schrodts article shows that application of a Holland
classifier can lead to successful short-term prediction of international events. A Holland classifier
learns by using event sequences it has encountered in the past to construct rules by which it can
predict future sequences.

tone indicator of the progress in the field is the high quality of available textbooks (Huckfeldt, Kohfeld and Likens 1982;
Boynton 1980; Hanneman 1988).
 Mathematical modelling in political science 403

CONCLUSION

This issue offers a cross section of the mathematical research which is currently being conducted.
The topics addressed range widely, as do the methods. Since the origin of formal modelling in
political science in the early 1950s the field has done extremely well in a brief period of time.
Careful analyses of the meaning of the word science and its applicability to the subjects of
politics have been offered elsewhere (e.g. Kramer 1983; Riker 1977; Rapoport 1983). Some of our
colleagues in the natural sciences might suppose that the study of politics is not scientific because
the subject matter is too complicated and unpredictable. The consensus among political scientists
is that this is incorrect. Human behavior is more complicated than the behavior of simple physical
processes, such as gravitational acceleration or the response of the pressure of a gas to changes
in temperature. Human beings have preferences, they act under a set of complex social, economic
and political constraints, they learn, and are influenced by events in the distant past (through
memory) and in the present (by contextual variables). At the same time, human behavior is
patterned and in large part predictable-this predictability makes possible complex social
organizations and a lifestyle which is not, for most people, nasty, brutish or short. The goal of
the political scientist is to uncover both the existence of those regularities and the processes which
generate them, in the same way that physical and biological scientists set out to discern laws
governing their observations.

Acknou~ledgements--I would like to thank my colleagues in political science who acted as anonymous referees in the process
of selecting these articles. In drafting this paper. I have received helpful advice from Scott Ainsworth, Mike Krdssa, Phil
Schrodt and Jeff Sutter.

REFERENCES
Arrow, Kenneth J. 1951, 1963. Socictl Choice and Individual Values. First and Second Editions. New Haven: Yale University
Press.
Axelrod, Robert. 1984. The Evolution of Cooperation. New York: Basic Books.
Bendor, Jonathan, Serve Taylor and Roland van Gaalen. 1987. Stacking the Deck: Bureaucratic Missions and Policy
Design. American Political Science Review 81: 873-896.
Black, Duncan. 1984. On the Rationale of Group Decision Making. Journal of Political Economy 56: 23-34.
Black, Duncan. 1958. The Theory of Commirtees and Elections. Cambridge: Cambridge University Press.
Boynton, G. Robert. 1980. Mathematical Thinking about Politics. New York: Longman.
Brams, Stephen. 1985. Superpower Comes. New Haven: Yale University Press.
Brams, Stephen and Marc Kilgour. 1987. Threat Escalation and Crisis Stability: A Game-theoretic Analysis, American
Political Science RevieMa 8 1: 833 -850.
Brown, Courtney. 1987. Voter Mobilization and Party Competition in a Volatile Electorate. American Sociologicul RerieM.
52: 59-73.
Bueno de Mesquita, Bruce. 1981. The War Trap. New Haven: Yale University Press.
Coleman, James S. 1964. Introduction to Mathematical Sociology. Glencoe: Free Press,
Converse, Phillip. 1969. Of Time and Partisan Stability. Comparative Political Studie.r 2: 139917 I.
Cortes, Fernando, Adam Przeworski and John Sprague. 1974. Systems Analysis for Social Scientisrs. New York: Wiley.
Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper & Row.
Ferejohn, John and Morris P. Fiorina. 1975. The Paradox of Not Voting: A Decision Theoretic Analysis, American Political
Science Rerien, 69: 525-536.
Fiorina, Morris P. 1975. Formal Models in Political Science. American Journal qf Political Science 19: 133 159.
Forrester, Jay W. 1971. World Dynamics. Cambridge: Wright-Allen.
Friedman, James W. 1986. Game Theory With Applications to Economics. New York: Oxford University Press.
Gillespie, John V. et al. 1977. An Optimal Control Model of Arms Races. American Political Science Review 71: 226244.
Goodman, Leo. 1957. Some Alternatives to Ecological Correlation. American Journal qf Sociology 64: 610-625.
Hanneman, Robert A. 1988. Computer-assisted Theory Building: Modeling Dlxamic Social Systems. Beverly Hills: Sage.
Huckfeldt, Robert, C. W. Kohfeld and Thomas W. Likens. 1982. Dynamic Modelling: An Introduction. Beverly Hills: Sage.
Kramer, Gerald H. 1973. On a Class of Equilibrium Conditions for Majority Rule. Econometrica 41: 2X5-287.
Kramer, Gerald H. 1983. The Ecological Fallacy Revisited: Aggregate versus Individual-level findings on Economics and
Elections an Sociotropic Voting. American Political Science Review 77: 92-l 11.
Kemeny. John G. and J. Laurie Snell. 1962. Mathemafical Models in the Social Sciences. New York: Blaisdell.
Key, V. 0. 1949. Sourhern Politics. New York: Knopf.
Lute, R. Duncan and Howard Raiffa. 1957. Game.y and Decisions: Introduction and Critical Survey. New York: Wiley.
Luterbacher, Urs and Michael D. Ward, Eds. 1985. Dynamic Models in International Conflict. Boulder: Lynne Riener.
MacKuen, Michael B. 1981. More than Ne~ss: Media Power and Public Affhirs. Beverly Hills: Sage.
May, Robert M. 1974. S/ability and Complexit~~ in Model Ecosystems. Princeton: Princeton University Press.
McKelvev, Richard D. 1976. Intransitivities in Multidimensional Voting Models and Some Imrlications for Agenda
Control. Journal of Economic, Theory 12: 472 482.
McKelvey, Richard D. and Norman Schofield. 1986. Structural Instability of the Core. Journal of Mathematic~al Economics
15: 179-198.
404 PAUL E. JOHNSON

McKelvey, Richard D. and Norman Schofield. 1987. Generalized Symmetry Conditions at a Core Point. Econometrica 55:
9233933.
McPhee, William. 1963. Formal Theories of Mass Behavior. Glencoe: Free Press.
McPhee. William and W. Glaser, Eds. 1962. Public Opinion and Congressional Electrons. Glencoe: Free Press.
Meadows, Donella H. et al. 1972. Limits To Growth. New York: Universe Books.
Miller, Gary and Terry Moe. 1983. Bureaucrats, Legislators, and the Size of Government. American Political Science Review
17; 297-322.
Nash, John F. 1950. Equilibrium Points in n-Person Games. Proceedings of the National Academy of Science 36: 4849.
Niskanen, William. 1971. Bureaucracy and Representative Government. Chicago: Aldine.
Ordeshook, Peter C. 1986. Game Theory and Political Theory. Cambridge: Cambridge University Press.
Owen, Guillermo. 1982. Game Theory. Second Edition. New York: Academic Press.
Palfrey, Thomas and Howard Rosenthal. 1985. Voter participation and Strategic Uncertainty. American Political Science
Review 79: 62-78.
Plott, Charles. 1967. A Notion of Equilibrium and its Possibility Under Majority Rule. American Economic Review 57:
787-806.
Powell, Robert. 1987. Crisis Bargaining, Escalation, and MAD. American Political Science Reoiew 81: 717-736.
Przeworski, Adam and John Sprague. 1986. Paper Stones: A History of ElecforaI Socialism. Chicago: University of Chicago
Press.
Rapoport, Anatol. 1957. Lewis F. Richardsons Mathematical Theory of War. Journal of Conjict Resolution 1: 2499299.
Rapoport, Anatol. 1960. Fights, Games, and Debates. Ann Arbor: University of Michigan Press.
Rapoport, Anatol. 1983. Mathematical Models in the Social and Behavioral Sciences. New York: Wiley.
Rashevsky, Nicolas. 1948. Mathematical Theory of Human Relations. Chicago: University of Chicago Press.
Rashevsky, Nicolas. 1951. Mathematical Biology of Social Relations. Chicago, University of Chicago Press.
Richardson, Lewis A. 1960. Arms and Insecurity: A Mathematical Study of the Causes and Origins of War. Pittsburgh:
Boxwood Press,
Riker, William. 1977. The Future of a Science of Politics. American Behavioral Scientist 21: llm~38.
Riker. William. 1982. Liberalism Against Populism: A Confrontation Between the Theory of Democracy and the Theory of
Social Choice. San Francisco: Freeman-
Riker. William and Peter C. Ordeshook. 1968. The Calculus of Votine. American PoIitica[ Science Review 62: 2542.
Schelling. Thomas C. 1971. Dynamic Models of Segregation. Journal if Mathematical Sociology 1: 143-186.
Schelling, Thomas C. 1978. Microstructure and Macrobehavior. New York: Norton.
Schofield, Norman. 1978. Instability of Simple Dynamic Games. Review of Economic Studies 45: 5755594.
Schrodt, Philip A. et al., 1977. Parameter estimation by numerical minimization methods. international Interactions 4:
2799301.
Shepsle, Kenneth A. 1979. Institutional Arrangements and Equilibrium in Multidimensional Voting Models. American
Journal of Political Science 23: 27760.
Stokes, Donald E. and Gudmund R. Iverson. 1966. On the Existence of Forces Restoring Party Competition. In Elections
and the Political Order, Eds. Angus Campbell et al. New York: Wiley.
Von Neumann, John and Oskar Morgenstern. 1944, 1947. Theory of Games and Economic Behavior. First and Second
Editions. Princeton: Princeton University Press.
Ward, Michael D. 1984. Differential Paths to Parity: A Study of the Contemporary Arms Race. American Political Science
Review 78: 297-3 17.