IDEOLOGY
argument. Important political disagreements involve
I distantly removed from economic concerns as normally
IDEOLOGY
1. Introduction
Ideology is a thorn in the side of public choice. The disci-
pline would be more powerful and more useful if it were
not true that ideological factors were important in explain-
ing political behavior. This is true for at least two reasons.
First, public choice applies economic methods and theories
to political behavior. But economic theory has no theory of
ideology and no room for ideological factors, and so the
application of public choice is weakened by the importance
of factors that are outside of economics. Second, a key
tenet of public choice theory is that people operate in the
political realm in about the same way as in the economic
realm. But in normal economic behavior, there is little
room for ideological aspects of behavior. Although people
may try to boycott South African products or to purchase
recycled paper goods, consumption decisions of this sort
are marginal and fairly unimportant. But as shown below,
ideological factors are much more important in politics.
Nonetheless, empirical evidence indicates that ideology is
an important part of political behavior, and scholars must
take the world as we find it.
Public choice scholars have examined two main issues
associated with the influence of ideological factors on polit-
ical behavior. First is the existence question: is there such an
influence? Second is the role of the ideology of the elected
representative versus the role of constituent ideology. This
has been addressed as a form of “shirking”: do elected
representatives avoid accountability by voting their own
ideological preferences instead of the interests of their con-
stituents? We now have answers to these questions. First,
ideology does matter. Second, politicians do not shirk;
rather, elected representatives dependably pursue their con-
stituents’ interests. However, in the process of answering
these questions, the analysis of ideology has been markedly
advanced and its empirical relevance has been extensively
explored. This analysis shows that ideology may be a pri-
mary building block of political conduct. I discuss each of
these issues: existence, shirking, and current and new
problems related to ideology.
If the question of the significance of ideology were
raised today, it is unlikely that there would be much of an
291
issues, such as abortion and homosexual rights, which are
understood. Often the opinion of an official’s “liberalism”
or “conservatism” is unrelated to his or her views on eco-
nomic issues, and decided entirely by the position on mat-
ters such as abortion. Moreover, positions on social and
economic issues seem to be blended together: it is difficult
or even impossible to find an elected official who is in
favor of more economic freedom and also in favor of abor-
tion rights, and conversely. As we shall see, both of these
observations — the significance of non-economic ideolog-
ical issues, and the linkage of issues so that the ideological
space is one-dimensional — are consistent with what we
have now come to appreciate is the nature of ideology in
politics.
2. Existence
George Stigler (1971) unintentionally generated an interest
in ideology. Stigler took an particularly strong stand
against the importance of ideology, and argued that in
effect all economically relevant political behaviors could
be explained by economic self-interest (see also Posner,
1974; Peltzman, 1976). This idea was appealing to econo-
mists for two reasons. First, the hypothesis was sharp and
was empirically testable. Second, the nature of the hypoth-
esis was one that created sympathy among economists;
economists are pleased to suppose that economic self-
interest is the main motivator of human behavior. This may
be because such a belief creates a stronger position for eco-
nomics as a discipline, and hence is in the economic self-
interest of economists (the downside of this argument is
that, as Stigler 1976 indicates, economists will have little
influence on policy decisions). On the other hand, it may be
because economists themselves behave this way, and
expect others to do so as well (see Carter and Irons, 1991).
That is, economists liked this hypothesis for reasons related
both to their own self-interest and to their ideology.
However, as much as we might like the hypothesis and
wish it to be true, it is fair to say that the results are in, and
the hypothesis has been falsified. There has been an exten-
sive body of empirical investigation of congressional vot-
ing and it has shown that ideological factors have extensive
power in explaining congressional voting; see also Mueller,
1989, for a discussion agreeing with this conclusion. The
major initial contributors to the literature were James Kau
and Paul Rubin (1978, 1979, 1981; Kau et al., 1982).
Joseph Kalt and Mark Zupan (1984) also made a signifi-
cant contribution. These scholars have found that
ideology, calculated as a score on a voting scale such as one
292
compiled by the Americans for Democratic Action (ADA)
or measured by the presidential vote in the congressional
district, has considerable explanatory power in predicting
voting by individual congressmen. These results may be
considered as confirming Schumpeter’s (1950) claim that
ideology is important in economic affairs (although not
his claim that it would lead to the fall of capitalism, a pre-
diction which has recently been falsified).
The basic technique has been to use voting by con-
gressmen on issues as the dependent variable and a list of
factors aimed at measuring constituent economic interests
and also ideological variables as the independent variables
in probit or logit regressions. The constituent characteris-
tics generally include income, age, urbanization, race,
education, unemployment, industry of employment,
unionization, measures of government spending in the
district, and occasionally measures of particular types of
economic activity in the district. Following Kau and Rubin
(1979), it is common to regress these measures on a meas-
ure of ideology such as ratings assigned by the ADA or
other ideological pressure groups and then use the unex-
plained regression residual as the measure of “pure” ideol-
ogy. Sometimes a simultaneous model is used where
campaign contributions are also controlled for (e.g.,
Chappell, 1982; Kau et al., 1982; Kau and Rubin, 1982,
1993; also see Stratmann, 1992a for a careful analysis of
this issue). Logrolling is also taken into account in some
specifications (e.g., Kau and Rubin, 1979; Stratmann,
1992b.) In such analyses, the measure of ideology is invari-
ably statistically and economically significant.
What has perhaps been just as influential in persuading
many scholars that ideology is an important variable are the
results of the comparatively futile attempts to challenge the
hypothesis. Indeed, the original work by Kau and Rubin
was projected to show that ideology did not matter, and
failed in this endeavor. Peltzman (1984, 1985) later
engaged in a determined effort to show that ideology was
unimportant. In the 1984 paper, he controlled more care-
fully than others had for constituent characteristics, specif-
ically by measuring the characteristics of those who
actually voted for US Senators rather than characteristics
of all voters in the electoral district (here, the state.) In this
way he was able to reduce the impact of the ideological
variable, but not, in general, to eliminate it. That is,
Peltzman (1984) found that ideology mattered, although
perhaps not as much as others had thought. Indeed, in a
subsequent paper (Peltzman, 1985), he found much the
same result as others: while economic factors are relevant
in explaining congressional voting, and, in particular,
trends in such voting, non-economic factors also are highly
significant. Peltzman calls these non-economic factors
IDEOLOGY
“history,” but he indicates that “one could allude to
regional differences in ideology as easily as to ‘historical
inertia’ … ” (Peltzman, 1985, p. 666). In his 1985 paper,
Peltzman himself relegates the results of his 1984 paper to
a footnote. Thus, the existence of ideology as an important
determinant of congressional voting has survived an inten-
sive attack by a skillful econometrician (I do not offer a
complete analysis of the economic literature relating ideol-
ogy to passage of legislation. There are numerous papers
providing such analysis. Lott and Davis (1992, footnote 1)
cite 18 such papers. Kalt and Zupan (1990, footnote 2) cite
20, only partially overlapping those listed by Lott and
Davis).
Further verification sustaining the importance of ideol-
ogy is supplied by studies examining campaign contribu-
tions. Changes in campaign finance laws led to a large
increase in the ability of a variety of groups to create
Political Action Committees (PACs) and contribute to
political campaigns by. Traditionally, labor unions had con-
tributed extensively to campaigns. However, changes in the
law enabled business and ideological groups to form PACs
and use contributions to attempt to attain their goals (it is
interesting to note that the changes in the law which gener-
ated the rise of the PACs were endorsed by labor unions,
apparently because they failed to predict the effects of
these changes on the ability of businesses to donate to cam-
paigns). Both business and ideological PACs now con-
tribute substantially to political campaigns. The structure
of these contributions is analyzed more fully in Poole
and Romer (1985), who show that there is an important
ideological component to contributions in general.
The large volume of contributions generated by ideo-
logical PACs is itself evidence against the strong economic
interest hypothesis. These contributions show that many
individuals and groups are willing to spend money to attain
non-economic goals. This reveals that these contributors
have tastes for public goods. This observation is of course
consistent with economic theory (where the elements of the
utility function are unspecified) but it is not consistent with
the strong claim that political action is aimed only at
increasing money income. If some are willing to contribute
money to change public policy for non-economic reasons,
then it is not surprising that people are also willing to use
their votes to achieve these same goals. It would not be sur-
prising if citizens were willing to vote for candidates who
promised to support legislation which would gratify their
ideological, as opposed to economic, preferences.
This is particularly true when we remember that the act
of voting is itself not understandable in terms of economic
rationality, since the probability of any one vote changing
the outcome of an election is infinitesimal and voting does
 IDEOLOGY
have positive costs (Downs, 1957). However, Nelson
(1994) has recently formulated a model, discussed more
fully below, in which political activity, including apparently
ideological activity, serves a private goal (see also Morton,
1991).
Even if we believe that the ideological PACs actually
attain economic goals, economic theory still has difficulty
in explaining their survival. This is because such PACs
must overcome considerable free rider problems, as dis-
cussed by Olson (1965). Even for a potential contributor
who believes in the goals of an organization, the optimal
strategy would be to abstain from contributing and to
free ride on the donations of others. Thus, when we observe
individuals freely contributing to ideological PACs, we
have already observed a phenomenon that contradicts
the restricted view of self interested rationality. Perhaps, in
addition to tastes for policies, individuals also have tastes
for feeling that they themselves have influenced those poli-
cies (e.g., Andreoni, 1989). However, whatever theories
may be evolved to explain these organizations, their exis-
tence and size does present a puzzle for the strong versions
of the economic theory of politics.
3. Shirking
Even if it is conceded that ideology “matters” in congres-
sional voting, an important question remains. That is the
issue of whose ideology counts. Congressmen voting ideo-
logically might simply be reflecting tastes of constituents.
Alternatively, they might be indulging their own prefer-
ences (Wittman, 1977, 1983). This is essentially a principal–
agent question: are legislators good agents for their
constituents’ (principals’) ideological views, or are they
shirking and representing their own views?
This question has been addressed thoroughly in the lit-
erature. There has been some confusion in many analyses,
however. The initial work (Kau and Rubin, 1979) was con-
cerned with the issue of whether non-economic factors
(called ideology) influenced legislation; this work did not
attempt to differentiate between the ideology of the repre-
sentative and that of his constituency. For example, Kau
and Rubin (1979, p. 366) stated explicitly that, “the repre-
sentative (or his constituency) may be ideologically in
favor of the bill … .”
But others have conflated the question of ideological
impacts on voting with the question of ideological shirking.
Peltzman (1984), for example, views his analysis as testing
between the impacts of constituent interest and ideology. It
is possible for a representative to represent constituent ide-
ological interest, so Peltzman’s categories are not mutually
exclusive.
293
Conceptually, issues of ideology and of shirking should
be separated. It is possible to have ideologically based vot-
ing without shirking. It is also possible to have shirking
without ideologically based voting. For example, a repre-
sentative might vote in response to contributions received
from various special interests, and thus shirk with respect
to his constituents’ desires, but in a way unrelated to ideol-
ogy. Indeed, Kau and Rubin (1993) find exactly this form
of shirking. This last possibility has not been carefully
studied in the literature, which has focused on ideological
shirking (see Bender and Lott 1996 for an analysis).
Kalt and Zupan (1984, 1990) argue strongly that the
observed voting behavior of representatives comports well
with representatives’ own ideologies, and therefore repre-
sents shirking. Nelson and Silberberg (1987) test the
responsiveness of their measure of shirking to changes in
its relative price. However, as Bender and Lott (1996) indi-
cate, their measure (relative strength of ADA ratings in
explaining voting on general versus special-interest bills) is
flawed. A better measure is the behavior of legislators in
periods when reelection is not an issue; using this measure,
Lott (1987) finds that legislators shirk by voting less often,
but do not change their ideological positions when they do
vote. Dougan and Munger (1989), following Downs
(1957), argue that what appears to be ideology is actually
an investment in brand-name capital — a signal of reliabil-
ity and commitment providing voters with assurance that
representatives will not behave opportunistically. They
argue that past votes create a valuable reputational asset
that makes promises of future voting behavior credible.
Both Kalt and Zupan and Dougan and Munger present
empirical evidence that they argue is consistent with that
hypothesis. Glazer and Grofman (1989) argue that ideol-
ogy is a method used by politicians to communicate with
constituents, who pay limited attention to political debates.
In a recent and important paper, Lott and Davis (1992)
have criticized the methodology both of Kalt and Zupan
and of Dougan and Munger. More importantly, Lott and
Davis (1992) and Bender and Lott (1996) have shown that
voters punish shirking, to the extent that it exists, and that
they are empirically quite sensitive to wandering in ideo-
logical space. One result is that senators who deviate from
the interests of their constituents by as little as 1.27 per-
centage points are ultimately defeated. Thus, they conclude
that whether or not shirking exists is unimportant since if
there is shirking it is strongly punished by political mar-
kets. Kau and Rubin (1993) also find that if there is ideo-
logical shirking, it is strongly and quickly punished. The
key argument is that political markets do a good job of sort-
ing legislators. That is, a representative must be in ideolog-
ical agreement with his constituency to be elected to office.
294 IDEOLOGY
If ideological voting is based on constituent ideology,
this also has created a puzzle for public choice scholars. We
do not have a theory of constituent ideology. Indeed, we
have no theory of political behavior in general by individ-
uals. Revealed theory cannot explain why rational individ-
uals vote.
Recently, however, Nelson (1994) has offered a theory
that depends on ideology to motivate voting. Nelson begins
with the standard Downsian observation that, because the
chance of any given voter influencing the outcome of an
election is trivial, there is no private motive for voting
related to the expectation that any one vote will be decisive.
Nelson goes on to argue, however, that there is a private
motive for voting and for expressing a political (ideologi-
cal) position that is distinct from “instrumental” consider-
ations. In particular, he suggests that “political positions
are … chosen not because these positions are the desired
outcome for voters, but rather because one wants to associ-
ate with certain people and they have certain positions.
People imitate others in choosing political positions”
(Nelson, 1994, p. 92).
Nelson presents empirical evidence, relating to political
behavior of ethnic groups, which is consistent with his
hypothesis. For example, he shows that membership in var-
ious ethnic groups is significant in determining political
affiliation (after adjusting for economic variables) and that
the income of the ethnic group’s members 60 years ago is
significant in explaining its political orientation today.
Thus, Nelson has presented a theoretical and empirical
basis for constituent ideology based on private motives and
normal utility maximization. This model should make the
use of variables related to constituent ideology less contro-
versial among economists.
4. Current Research and Outstanding Questions
Any future work on ideology must begin with a very
important analysis that is probably the most significant
public choice analysis of the US Congress. This is Poole
and Rosenthal (1997), a book-length treatment of roll call
voting by the House and Senate for all roll calls from the
first Congress in 1789–1985. It is based on numerous
articles by these authors and others; I will refer to the book
rather than to the articles. Poole and Rosenthal find that
what they call ideology is the basic organizing principle
behind all such voting. By their definition, “voting is along
ideological lines when positions are predictable along a
wide set of issues” (Poole and Rosenthal, 1997, p. 4). They
show that individual congressmen can be arrayed along a
unidimensional continuum for most roll call votes. Almost
everyone to one side of a “critical point” on this continuum
will vote one way and almost everyone to the other side
will vote the other way, with errors being clustered near the
critical point. The winner in any vote is then determined by
the location of this critical point.
Ideology as so defined is more important than con-
stituent economic interest in explaining voting by legisla-
tors. Moreover, it is even more important in explaining the
policies that are actually selected than in explaining voting
per se. Policies adopted tend to be systematically biased
away from the center of the distribution of legislators and
towards the ideological center of the majority party.
Ideology is more important in influencing the outcome of
the legislative process than in influencing voting by legis-
lators, even though voting is what has most often been
studied.
A major issue raised by Poole and Rosenthal is the
nature of the voting continuum. They characterize it in sev-
eral ways. It is roughly defined in terms of conflict over
economic redistribution. It also generally reflects party loy-
alty. A key point is that the continuum reflects logrolling.
That is, votes are structured so that coalitions are main-
tained across most issues. This explains in part why the
economic interests of constituents are not significant in any
one vote. The vote trading reflected in the continuum in
part accounts for these interests. One theoretical treatment
consistent with these arguments is Hinich and Munger
(1996). However, as discussed below, the nature of this
continuum is one of the most important research questions
in the contemporary study of ideology.
Poole and Rosenthal also show that for most of American
history, a single ideological dimension is all that is required
to array votes. During two periods (the 1830s–1840s and the
1940s–1960s) a second dimension dealing with race was
also useful. From the New Deal until the 1970s, they find
that there was actually a three-party system in the United
States (Republicans, Northern and Southern Democrats).
More recently, the country has returned to a single ideologi-
cal dimension and a two-party system.
This is not a logical necessity. For example, Poole and
Rosenthal point out that it would be possible for two
dimensions to be required, and the actual number needed is
an empirical matter. They give the example of social and
economic ideology, with economic liberals favoring gov-
ernment intervention in economic matters and social liber-
als opposing intervention in social or behavioral issues. It
would then be possible for two dimensions to be needed to
explain voting. But the fact that one dimension is sufficient
is because social conservatism and economic liberalism are
highly correlated, and the reverse. There are relatively few
libertarians, and also relatively few individuals who favor
both social and economic interventionism. This may be
 IDEOLOGY
because individuals attracted to politics tend to be those
who are interested in controlling others (Rubin, 2002).
The procedure used to estimate the continuum is called
“NOMINATE.” It is an iterative procedure, aimed at max-
imizing the probabilities assigned to the observed votes.1
They also develop a dynamic procedure, D-NOMINATE,
based on assuming that each legislator moves at most along
a linear trend over his career. This assumption and resulting
trend line enables Poole and Rosenthal to estimate a com-
mon issue space for all US history. (Groseclose et al., 1996
estimate a similar function.) Because of the volume of data
involved (11,473 legislators, 70,234 votes and 10,428,617
total decisions), the estimation requires the use of a super-
computer. A two-dimensional model (i.e., an issue space
allowing for two ideological dimensions) and a linear trend
for each legislator provide as good a fit (about 85% of indi-
vidual votes predicted correctly) as higher-order models
(with either more dimensions or a more complex polyno-
mial time path for legislators). Indeed, a one-dimensional
model assuming that each legislator maintains a constant
position predicts about 80% of the votes correctly.
Poole and Rosenthal show that the NOMINATE variable
is highly correlated with more traditional measures of
ideology, such as the ADA rating. This is a nice result
since many researchers, including many of those discussed
above, have used these scores in measuring ideology.
NOMINATE is a preferred rating scale since it is more
comprehensive and since traditional interest group ratings
are subject to “folding” problems. That is, a legislator just
a little more liberal than the ADA would get the same rat-
ing as a representative who was more conservative by the
same amount. However, many of the major interest groups
(e.g., the ADA and the American Conservative Union) are
at or very near the (opposite) ends of the political space, so
this problem is not acute.
The major theoretical issue regarding ideology is the
low dimension of the ideological space, as found by
Poole and Rosenthal (this is consistent with others who have
examined ideology, but Poole and Rosenthal document the
result much more carefully and completely). Indeed, it
appears that the US political system can only handle a one-
dimensional space. In those two periods when the space
increased to two dimensions, catastrophe followed. The
first was the Civil War. The second was the chaos of the
1960s. Thus, it appears that our institutions may have diffi-
culty with a policy space of more than one dimension. This
of course would not be surprising; many of the theoretical
results following Arrow show that the only guarantee of
stability is a unidimensional, unimodal issue space. But
what has not been fully explored is the mechanism that
constrains the US to remain in such a space.
295
The work of Poole and Rosenthal shows that such a
mechanism must exist and that it is exceedingly important.
Poole and Rosenthal have measured the issue space for con-
gressional voting. However, the results discussed above on
the absence of evidence of shirking show that the congres-
sional issue space is congruent with the issue space of voters
as well. In other words, if the ideological space for Congress
is one-dimensional and unimodal, and if Congress faithfully
reflects preferences of constituents, then the issue space of
voters must also be one-dimensional and unimodal. This
conclusion raises the question of which comes first — does
the issue space facing Congress come from the underlying
preferences of constituents, or does the political process
somehow define the issue space for individuals?
Poole and Rosenthal provide a mechanism for using ide-
ology in a time series analysis. Without their data, this
would be difficult because each Congress faces different
issues. For example: In year 1, the legislature passes a min-
imum wage of $2.00 per hour by a 51–49% margin. Liberal
ratings services such as the Americans for Democratic
Action count a vote for the minimum as a plus, so 51% of
the legislature receives a positive score for this vote. In
year 2, the legislature is more liberal, so it passes a $2.50
minimum wage by the same margin. The liberal rating
service still counts a vote for as a plus, so again 51% of the
legislature gets a positive score. Thus, even though the
legislature is more liberal, conventional measures of
ideology will show no change. However, by relating votes
by the same legislator from year to year, Poole and
Rosenthal are able to derive an intertemporally comparable
index. This index can then be used in time series analyses of
changes in policy. For example, Kau and Rubin (2002)
show that the ideological composition of the Senate has
only a small effect on the size and growth of government,
and the House has no effect. Thus, it appears that ideology
is important in the cross section but may be less so in the
time series. This issue is clearly worth additional analysis.
Beyond this, there are fundamental questions of the
nature of ideology itself. North (1990) has raised this issue
and stressed its importance. He believes that the basic
direction of society is determined by its ideological prefer-
ences and that we do not sufficiently understand these pref-
erences. This means that, in addition to determining the
mapping between constituent and representative prefer-
ences, the underlying structure of these preferences itself is
an issue of fundamental importance.
5. Summary and Implications for Future Research
The initial interest in ideology was an effort to determine
whether non-economic factors influenced economic
296 IDEOLOGY
legislation. The conclusion is that they do. The emphasis on
shirking following this initial analysis has, in my view,
been misguided. Constituent ideological preferences do
have impacts on legislative outcomes, and the profession
should devote its efforts to determining the source and
nature of these beliefs. The recent work of Nelson (1994) is
a good attempt at beginning this effort, and future research
on ideology should focus on this issue, rather than on
devising ever more scholastic tests to measure relative
strength of constituent and legislator ideology. In addition,
the factors that make the issue space unidimensional and
unimodal are worthy of attention because these factors
serve to eliminate problems of cycling and instability.
Finally, the work of North (1990) tells us that the underly-
ing structure of ideology is itself of crucial importance in
ordering an economy, and we have no good theory to
explain this structure.
PAUL H. RUBIN
NOTES
1. The program and data are available at Poole and Rosenthal’s
web site, http://voteview.gsia.cmu.edu/.
REFERENCES
Andreoni, James (1989). “Giving with impure altruism: applica-
tions to charity and Ricardian equivalence.” The Journal of
Political Economy, 97(6): 1447–1458.
Bender, Bruce and John R. Lott, Jr. (1996). “Legislator voting and
shirking: a critical review of the evidence.” Public Choice, 87:
67–100.
Carter, John R. and Michael D. Irons (1991). “Are economists
different, and if so, why?” Journal of Economic Perspectives,
5: 171–178.
Chappell, Henry W. (1982). “Campaign contributions and
congressional voting: a simultaneous probit-tobit model.”
Review of Economics and Statistics, 64: 77–83.
Dougan, William R. and Michael C. Munger (1989). “The ration-
ality of ideology.” Journal of Law and Economics, 32: 119–142.
Downs, Anthony (1957). An Economic Theory of Democracy.
New York: Harper & Row.
Glazer, Amihai and Bernard Grofman (1989). “Why representa-
tives are ideologists although voters are not.” Public Choice,
61: 29–39.
Groseclose, Tim, Steve Levitt, and Jim Snyder (1996). “An infla-
tion index for ADA scores,” Unpublished manuscript, Ohio
State University.
Hinich, Melvin J. and Michael C. Munger (1996). Ideology and
the Theory of Public Choice. Ann Arbor: University of
Michigan Press.
Kalt, Joseph P. and Mark A. Zupan (1984). “Capture and ideology
in the economic theory of politics.” American Economic
Review, 74: 279–300.
Kalt, Joseph P. and Mark A. Zupan (1990). “The apparent
ideological behavior of legislators: Testing for principal-agent
slack in political institutions.” Journal of Law and Economics,
33: 103–132.
Kau, James B., Donald Keenan, and Paul H. Rubin (1982).
“A general equilibrium model of congressional voting.”
Quarterly Journal of Economics, 97: 271–293.
Kau, James B. and Paul H. Rubin (1978). “Voting on minimum
wages: a time series analysis.” Journal of Political Economy,
86: 337–342.
Kau, James B. and Paul H. Rubin (1979). “Self-interest, ideology
and logrolling in congressional voting.” Journal of Law and
Economics, 22: 365–384.
Kau, James B. and Paul H. Rubin (1982). Congressmen,
Constituents, and Contributors. Boston: Martinus Nijhoff.
Kau, James B. and Paul H. Rubin (1993). “Ideology, voting, and
shirking.” Public Choice, 76: 151–172.
Kau, James B. and Paul H. Rubin (2002). “The growth of gov-
ernment: sources and limits.” Public Choice, 113: 389–402.
Lott, John R. Jr. (1987). “Political cheating.” Public Choice, 52:
169–187.
Lott, John R. Jr. and Michael Davis, L. (1992). “A critical review
and extension of the political shirking literature.” Public
Choice, 74: 461–484.
Morton, Rebecca (1991). “Groups in rational turnout models.”
American Journal of Political Science, 35: 758–776.
Mueller, Dennis C. (1989). Public Choice II. New York:
Cambridge University Press.
Nelson, Douglas and Eugene Silberberg (1987). “Ideology and
legislator shirking.” Economic Inquiry, 25: 15–25.
Nelson, Philip (1994). “Voting and imitative behavior.” Economic
Inquiry, 32: 92–102.
North, Douglass C. (1990). Institutions, Institutional Change and
Economic Performance. Cambridge: Cambridge University
Press.
Olson, Mancur. (1965). The Logic of Collective Action.
Cambridge, MA: Harvard University Press.
Peltzman, Sam (1976). “Toward a more general theory of regula-
tion.” Journal of Law and Economics, 19: 211–240.
Peltzman, Sam (1984). “Constituent interest and congressional
voting.” Journal of Law and Economics, 27: 181–210.
Peltzman, Sam (1985). “An economic interpretation of the history
of congressional voting in the twentieth century.” American
Economic Review, 75: 656–675.
Poole, Keith T. and Howard Rosenthal (1997). Congress:
A Political-Economic History of Roll Call Voting. New York:
Oxford University Press.
Poole, Keith L. and Thomas Romer (1985). “Patterns of political
action committee contributions to the 1980 campaigns for the
United States House of Representatives.” Public Choice
(Carnegie Papers on Political Economy), 47: 63–113.
Posner, Richard A. (1974). “Theories of economic regulation.”
Bell Journal of Economics and Management Science, 5:
335–358.
Rubin, Paul H. (2001). “Ideology,” in William E. Shughart II and
Laura Razzolini (eds) The Elgar Companion to Public Choice.
Northampton, MA: Edward E. Elgar, pp. 328–336.
 THE IMPORTANCE OF THE MIDDLE IN SPATIAL POLITICS
Rubin, Paul H. (2002). Darwinian Politics: The Evolutionary
Origin of Freedom, Newark: Rutgers University Press.
Schumpeter, Joseph A. (1950). Capitalism, Socialism, and
Democracy. New York: Harper & Row.
Stigler, George J. (1971). “The theory of economic regulation.”
Bell Journal of Economics and Management Science, 2: 3–21.
Stigler, George J. (1976). “Do economists matter?” Southern
Economic Journal, 42: 347–354.
Stratmann, Thomas (1992a). “Are contributors rational?
Untangling strategies of political action committees.” Journal
of Political Economy, 100: 646–664.
Stratmann, Thomas (1992b). “The effects of logrolling on congres-
sional voting.” American Economic Review, 82: 1162–1176.
Wittman, Donald (1977). “Candidates with policy preferences: a
dynamic model.” Journal of Economic Theory, 14: 180–189.
Wittman, Donald (1983). “Candidate motivation: a synthesis of
alternative theories.” American Political Science Review, 77:
142–157.
THE IMPORTANCE OF THE MIDDLE
IN SPATIAL POLITICS
1. Introduction
Since Hotelling (1929), almost as an after-thought in a paper
on the location of economic activity, observed that there was
a reason why winning Democrats and Republicans tended to
favor each other, there has been a recognition of the impor-
tance of the middle in political competition. It remained for
Black (1948, 1958) and Downs (1957) to develop single
dimensional models of spatial politics and to observe that
the median constituted a dominant strategy for a candidate
or party to prevail in an election in which there are only two
contenders.
Arrow (1951) turned the discussion in a different direc-
tion from that outlined by Black (1948) by proving that
under reasonable conditions there was no guarantee that a
social preference ordering might even exist. The fact that
there might be no solution, or no dominant strategy, or no
equilibrium, was a shockingly novel idea. Elections almost
always produce a winner, but it was not clear just what the
winner might represent. There was obviously the possibil-
ity of arbitrariness in the outcome, and that very thought
was unsettling.
It would not be long, however, before the search turned to
the issue of finding conditions under which there might be an
equilibrium or a dominant strategy. This search is the topic
that we review here. Unfortunately, since the literature is
large, it is probably impossible to do full justice to all the con-
tributions. We attempt, however, to cover the major aspects.
297
2. A Spatial Model
Arrow’s Impossibility Theorem first inspired researchers to
start to search for examples of cycles in elections (Riker,
1958). There was also the issue of interpreting the outcome
of an election and, perhaps even more important, of under-
standing how institutions might act to resolve or diminish
the importance of the possibility of encountering cycles
(Riker, 1980). However, as noted above, the largest focus of
attention was the search to find conditions under which
there was a dominant political strategy.
Of course, it was immediately clear that if the number of
dimensions (issues) is one, and if voter preferences are
single-peaked, then the results of Black and Downs are appli-
cable. A median is guaranteed to exist and it is the dominant
strategy in two party elections. Since countries with serious
multiple parties usually have a system in which there is a
runoff election between the two parties or candidates who
received the largest number of votes in the first round, this
analysis is thought to be applicable so that the median still
serves as the dominant strategy for a second round. Hence,
the median is a powerful and attractive position for politi-
cians who want to win and remain in office.
However, there remained a feeling, or maybe an
intellectual conviction, that there had to be more than one
dimension of choice. When there are two or more dimen-
sions to the space in which issues are measured, then the
Impossibility Theorem is applicable. No dominant strategy
can be guaranteed. Thus, the question of finding conditions
under which dominance does exist is an intellectual
challenge.
Plott (1967) may have been the first to explore this
issue, and he established sufficient conditions for equilib-
rium in multidimensional choice spaces. The most impor-
tant and best known of these conditions is pairwise
symmetry, which states that all nonzero utility gradients at
the equilibrium must be divisible into pairs that point in
opposite directions. Enelow and Hinich (1983) develop
examples to demonstrate that Plott’s conditions are indeed
sufficient, but they are not necessary. Indeed, Plott’s condi-
tions would lead one to believe, as Shepsle (1979, p.28)
states, “… unless the assumption of unidimensional or
symmetrically distributed preference is satisfied, the solu-
tion to the election game posited by spatial theory does not
generally exist.” However, Enelow and Hinich’s argument
demonstrates that Plott’s conditions are far stricter or con-
straining than need be, so equilibrium may not be so rare.
Davis and Hinich (1966, 1967) introduced a multidi-
mensional spatial model that utilized the tools of probabil-
ity theory, especially the notion of distributions. The main
results of these papers, along with some extensions, is