Chapter 1

The Discipline of Politics

It is important to be aware that the historical development of the

discipline, and in particular its openness to other perspectives, has
greatly influenced the emergence and use made of research methods
in politics and international relations. The first part of this chapter
shows how the development of the discipline, particularly in Britain,
continental Europe and the USA, reflected differing concerns of
academics and policy-makers and resulted in the institutionalization
of particular styles of research methodology.
However, largely under the influence of scholars in the United
States, three paradigms have shaped post-war political science
around the globe. The second part of the chapter charts how behav-
iouralism, new institutionalism and rational choice have increasingly
dominated approaches to research. Although there has been
constructive dialogue between these paradigms, at base rational
choice and institutionalist paradigms are constructed on mutually
incompatible premises. This has particular implications for choice of
research method and for understandings of political 'science'. The
final section of this chapter looks at what is at stake in the 'unity of
method' debate and considers the relationship between ontology,
epistemology and methodology.

The diverse traditions of political science

In many respects, politics is the junction subject of the social

sciences, born out of history and philosophy, but drawing on the
insights of economics and sociology and, to a lesser extent, the study
of law, psychology and geography. This openness to other perspec-
tives can enrich the discipline, but also leaves it open to the accusa-
tion that it lacks a distinctive theoretical and methodological core.
'We cannot talk about political science as a discipline if those who
call themselves political scientists and pretend to teach it are unable
to agree on its basic substance and methodology' (European

8
 The Discipline of Politics 9

Thematic Network for Political Science, or EPSNet, 2003, p.2). This

lack of a distinct core leaves the study of politics prey to the views of
those who argue that politics is a field of enquiry rather than a
distinctive discipline. It has been claimed, for example, that what
politics attempts to do as a discipline can be better achieved by polit-
ical sociology or by the deployment of the methodologies of econom-
ics to explain political phenomena.
The sense of the need to mark out clear boundaries for the disci-
pline, to provide it with a dominant and widely accepted para-
digm, and to adopt appropriate research methodologies and
techniques, has been strongest in the USA. In terms of resources
and intellectual power, American political science tends to domi- /
nate world political science. In 1980, about 80 per cent of the
world's political scientists were American (Dreijmanis, 1983,
p.214) and there is no reason to believe that the figure differs
greatly at the beginning of the twenty-first century. The annual
convention of the American Political Science Association (APSA)
attracts upwards of 6,000 delegates. They are predominantly
Americans: the rest are scholars who feel that they need to keep
up to date with the latest developments in American political
science.
Even in areas such as European integration, where one might
think European scholars would have something of a comparative
advantage, the American contribution to the body of theoretical and
empirical work has been considerable. The earlier contributions to
theorizing European integration 'emanated from the US, with
Europeans hanging on to the coat-tails of the argument' (Wallace,
2000, p.100). Similarly, Smith argues (2000, p.399) that in the case
of international relations, 'IR remains an American social science
both in terms of the policy agenda that US IR exports to the world
in the name of relevant theory and in terms of the dominant (and
often implicit) epistemological and methodological assumptions
contained in that theory.'
American political science has been characterized by repeated
attempts to establish a rigorously based academic discipline that
could at least bear comparison with, say, psychology, first through
the behavioural movement and more recently through rational
choice. British political science has been characterized by a greater
intellectual scepticism in what Jack Hayward has called a 'self-depre-
cating discipline' (Hayward, 2000, p.1).
However, political science in continental Europe has its own
distinctive traditions that are largely independent of Anglo-American
influences. Public law approaches have been an important influence,
10 Research Methods in Politics

reflecting the fact that there is usually a link in France between the
study of politics and the possibility of a career in the elite ranks of
the civil service. In France, the role of political science in the educa-
tion of elites, notably at the Ecole Nationale d'Administration gave
it a special status in national life. Sciences Po (as the Ecole Libre des
sciences politiques is usually known) in Paris was established in
1871-2 as part of a modernization effort after France's defeat in the
Franco-Prussian war. 'Initially, "political science" as such played a
very modest part in its teaching, the use of the plural (political
sciences) being indicative of the fact that law and history, as well as
other social sciences such as economics dominated the syllabus'
(Hayward, 2000, p.25). Much political science in France is taught
in law faculties. The key political science institutions are the nine
Institut d'Etudes Politiques set up after the Second World War.
Sciences Po is technically one of them, but they are essentially a
network of provincial institutions. They have a broader mission
than conventional political science departments, teaching their
students law and some economics to prepare them for careers in
business and the civil service. Effectively, the country has only two
conventional political science departments. Similarly, in Italy, the
weakness of departments when compared with the faculties that
organize teaching undermines the institutional base for political
science.
German political science was shaped in its early post-war years by
a task of democratic re-education that had American influences, but
since then has been able to draw on a more traditional German
social-philosophical approach. Rejuvenated by the student protest
movement of the late 1960s, 'This tradition may explain the
perceived gominance of abstract and theoretical traditions over the
empirical study of politics' (Saalfeld, 2002, p.139). The variability of
national experiences within Europe is illustrated by the fact that
Dutch political science went through a particularly fierce debate
about behaviouralism, and 'Soon the discipline was in total disarray'
(Lieshout and Reinalda, 2001, p.60). One consequence was that
academics studying public administration and public policy split
away from political science to form their own departments and asso-
ciation. Belgian political science is split on linguistic lines, while
Denmark did not establish its first political science department (at
Aarhus) until 1959. The early development of Danish political
science was strongly influenced by American behavioural traditions
(Rasmussen, 1985, p.320).
Swedish political science was influenced first by a German tradi-
tion in the first four decades of the twentieth century, then by the
 The Discipline of Politics 11

United States and more recently by an Anglo-American tradition.

Rather than an American hegemony, it is an Anglo-American axis
that seems to be dominant. In 1971 only 8 .4 per cent of the refer-
ences in the leading Swedish political science journal were from the
UK, but by 2000 some 25. 7 per cent of the references were from
that country. In the meantime, the number of references to
American publications had declined from 40 per cent to 23.8 per
cent (Angstrom, Hedenstrom and Strom, 2003, p.10). The Swedish
case raises the question of how far political science can retain a
distinctive national identity in a relatively small country with a
langauge that is not widely used. According to Swedish political
scientists themselves, Swedish political science 'stands less tall than
it did one scholarly generation ago' (Angstrom, Hedenstrom and
Strom, 2003, p.6). 'Swedish political science ... imports more than
it exports', but such small scholarly communities may nevertheless
'survive and even thrive' (Angstrom, Hedenstrom and Strom, 2003,
p.15).
As in other areas of cultural and intellectual life, Canada has made
a particular effort to ensure that its approach to the study of politics
can be differentiated from that pursued by its powerful neighbour
across the border. Of course, some Canadian political scientists are
American by origin and there is necessarily much interchange of
ideas. However, Canadian political science has developed a strong
international reputation in areas such as federalism, the study of
public policy and environmental politics. The respect with which
Canadian political science is held internationally is indicated by the
fact that the International Political Science Association currently has
its headquarters in the country. The distinctive traditions of Quebec,
expressed through the existence of a separate association for politi-
cal scientists working in the province, also gives Canadian political
science a special flavour.
Canadian political scientists have often worked with their coun-
terparts in Australia on topics such as federalism where they share a
common interest. Historically, the Australian National University /
(ANU) in Canberra, which has a strong research school, had links
with universities such as Oxford. ANU and other Australian univer-
sities still recruit leading political scientists from Britain. However,
the geographical isolation of Australia encourages Australian politi-
cal scientists to travel extensively. They are prominent at conferences
in Europe and the United States and draw from both approaches to
the study of politics. Not surprisingly, Australian political science is
also strong in the study of the Asia-Pacific region and such topics as
the politics of international trade.
12 Research Methods in Politics

American political science emerged at an earlier date than its

British counterpart, as is evidenced by the fact that APSA was I
preparing to celebrate its centenary at the millennium, while its
British counterpart (deliberately named the Political Studies
Association) was celebrating its golden anniversary. How can we
account for the earlier development of the discipline in the USA, but
in particular its more rapid professionalization (e.g., emphasis on the
possession of a PhD as a necessity rather than an embarrassment)
and its proneness to seek out 'scientific' paradigms?
As American universities evolved from their original function of
training Christian ministers, political scientists shared in the wide-
spread belief 'that universities must rely on science rather than reve-
lation to inform their teachings on natural or social phenomena'
(Ricci, 1984, p.66). It is therefore not surprising that early political
scientists turned to the natural sciences for paradigms:

Of the handful of political scientists who wrote on the nature of

the discipline between 1880 and 1903, most held a scientistic
point of view. That is to say, they believed that the methodology
they associated with the natural sciences was appropriate for
investigating problems of fundamental concern to political
science, and that proper application of this methodology would
lead to the development of 'laws' with explanatory and predictive
power. (Somit and Tanenhaus, 1967, p.76)

Universities in the United States underwent a massive expansion

between 1870 and 1920, a period in which the undergraduate popu-
lation increased from 54,300 in 1870 to 597,900 in 1920. This
created a demand for new subjects. This of itself does not explain
why political science should start to develop as a distinct discipline,
an event usually traced to the formation of a School of Political
Science at Columbia College (as it then was) in New York in 1880.
There were, however, developments in the wider society that
favoured the systematic study of politics. The influx of immigrants
into the United States, preponderantly from countries with monar-
chical or authoritarian systems of government, led to fears that the
republican and democratic traditions of the country could be under-
mined. The new immigrants needed to undergo formal political
socialisation and this led to the widespread development of 'civics'
teaching in American high schools, an issue only tackled in Britain at
the end of the twentieth century with a debate about political liter-
acy. The Progressive movement of the late nineteenth century and
early twentieth century was based on the concerns of a growing
 The Discipline of Politics 13

urban middle class about what they saw as corrupt systems of city
government. Systematic ideas for reform were needed and political
scientists were called upon to provide them. It should also be remem-
bered that law is not generally available as a first degree in the USA.
Majoring in political science is seen as an acceptable prerequisite to
attending law school and has helped to maintain buoyant demand
for undergraduate courses.
The development of the study of politics was a much more hesi-
tant process in Britain. The London School of Economics and
Political Science was founded in 189 5 with an agenda of training an
administrative elite and, more specifically, colonial civil servants. It
was part of the 'Webbs' Fabian Socialist grand design for a merito-
cratic-technocratic break with the gentleman-amateur Oxford tradi-
tion' (Hayward, 2000, p.12). Oxford had established its Gladstone
chair in politics before the First World War. The degree in
Philosophy, Politics and Economics (PPE), which so many future
political leaders were to take, was initially established in 1921. 'In
1932 changes took place which opened the way to a great increase
in the amount of teaching required in Politics. The modern form of
P.P.E. was introduced.' (Chester, 1975, p.155). Barry saw the combi-
nation of teachers of politics usually holding their first degrees in
history, and having to teach a very wide range of courses in their
colleges, as contributing to an amateurish approach to the subject.
(Barry, 2000, p.433). This comment perhaps overlooks the contri-
bution of Nuffield College in the 1930s onwards as a centre for
social science research.
British political studies in the 1950s was male-dominated and
characterized by a prevalent belief in the merits of British institu-
tions. Indeed, some influential political scientists in the USA argued
that it would benefit from moving closer to the model of a disci-
plined party system. As the 1950s came to an end, the complacency
of political scientists was beginning to be punctured by the failure
of Westminster-style constitutions in newly independent Third
World countries. In wider political life there was an increasing real-
ization that Britain was lagging behind the growth rates being
achieved by its continental competitors. There was a general ques-
tioning of British institutions and a demand for their moderniza-
tion. Political scientists such as Bernard Crick organized the Study
of Parliament Group (which rather unusually brought together
practitioners in the shape of Commons officials and academics) and
played a leading role in calls for the reform of Parliament. Other
political scientists assisted the royal commissions and committees
set up in an attempt to reform the civil service, local government
14 Research Methods in Politics

(and so on), with devolution being added as an agenda item in the

1970s. During this period a considerable expansion of university
education took place, with politics departments being substantially
increased in size in existing institutions and being an important part
of the new 'plateglass' universities such as Essex and Warwick. The
period from 1961 to 1974 is characterized by Hayward (1991,
p.313) as 'technocratic reformism'. It was perhaps the high point of
'relevance' for British political science, although institutions
remained largely unreformed.
The period since 1975 is characterized by Hayward (1991,
p.315) as one of 'sceptical professionalism'. For critics from the
left, the subject remained 'suffused, not to say suffocated, by a
bland orthodoxy, reform-inclined at the edges but essentially emol-
lient, apologetic and structurally unproblematic' (Miliband, 1976).
After a period of expansion, there was even some contraction in
the discipline in Britain in the 1980s. However, more modest
expansion was renewed in the 1990s, while intellectually the disci-
pline developed in new areas such as international political econ-
omy. The study of north-south relations attracted many able
younger researchers. In both Britain and the USA, there was a
marked tendency towards greater specialization in subfields of the
discipline. Barry draws an important distinction between special-
ization and fragmentation. He sees specialization as 'the elabora-
tion and extension of a common body of ideas to different aspects
of the subject ... In contrast, fragmentation might be defined as
what happens when the centrifugal tendencies inherent in special-
ization are not reined in by the gravitational pull of the central
core' (Barry, 2000, p.447).
Important developments have taken place in the direction of
creating a European political science, although perhaps it would be
more appropriate to talk of Europe-wide co-operation drawing on
the distinctive traditions of the different countries. The annual
workshops and other events organized by the European
Consortium for Political Research since its formation in 1970 were
followed by the formation in 1998 of the more teaching-oriented
European Thematic Network for Political Science, later reconsti-
tuted as EPSNet. However, these events have tended to have a
strong Northern European influence, with many of the participants
drawn from Britain, Germany, the Netherlands and the
Scandinavian countries. Norwegian political science has been
particularly influential, exemplified by the work of the leading
analyst of comparative politics from an historical perspective, Stein /
Rokkan.
 The Discipline of Politics 15

The collapse of the Communist regimes in Eastern Europe, where

the teaching of politics had largely been confined to the heavily ideo-
logical study of Marxism-Leninism, presented a unique opportunity
and challenge: 'Immediately after the collapse of communism, polit-
ical science had no institutional existence as a discipline or field of
training and education. Departments had to be established, reformed
or entirely redesigned. Unlike in Western Europe, political science
was introduced with a single "big bang" ' (Ilonszki, 2002, pp.23-4 ).
It is perhaps too early to assess the results of this experiment,
although it does seem that the familiar problem of the lack of
common themes to integrate diversity has occurred.

Dominant paradigms

Three paradigms have shaped post-war political science, especially in

the USA but also with an impact elsewhere: behaviouralism, new
institutionalism and rational choice.
Behaviouralism was the dominant paradigm in American political
science in the 1950s and 1960s and it fell away thereafter. It emerged
out of a discontent with the 'old institutionalism' which was preoc-
cupied with the formal structures of government. This approach 'had
quite spectacularly failed to anticipate the collapse of interwar
German democracy and the emergence of fascism' (Blyth and
Varghese, 1999, p.346). Discontent with what had gone before was
not of itself enough to launch a new paradigm. However, the devel-
opment of probability sampling techniques and a systematic under-
standing of mass surveys had opened up new areas of research for
political science, particularly in the study of what shaped electoral
choice (Lazarsfeld, Berelson and Gaudet, 1944 ). What quickly
became apparent was that voters were less well informed and ratio-
nal than classical democratic theory assumed. Funding foundations
were eager to encourage this new 'scientific' approach to political
research.
The advocates of the 'behavioural persuasion' saw 'themselves as
spokesmen for a very broad and deep conviction that the political
science discipline should ( 1) abandon certain traditional kinds of
research, (2) execute a more modern sort of inquiry instead, and (3)
teach new truths based on the findings of these new inquiries' (Ricci,
1984, p.140). What was the essence of this new paradigm? By
combining two classic accounts (Easton, 1965; Somit and
Tanenhaus, 1967) it is possible to identify eight main claims made
for beha viouralism.
 16 Research Methods in Politics

1. There are discoverable regularities in politics which can lead to

theories with predictive value.
2. Such theories must be testable in principle. This 'proposition
was repeatedly restated by behaviouralists ... it was fundamen-
tal to the point where Easton argued that the vital difference
between most pre-World War II political scientists and behav-
iouralists lay precisely in the latter's insistence on testability'
(Ricci, 1984, p.137).
13. Political science should be concerned with observable behaviour
which could be rigorously recorded.
4. Findings should be based on quantifiable data.
Research should be made systematic by being theory oriented
and directed.
Political science should become more self-conscious and critical
/ about its methodology.
/7. Political science should aim at applied research that could
provide solutions to immediate social problems. The truth or
falsity of values could not be established scientifically and there-
fore political science should abandon the 'great issues' except
where they were amenable to empirical investigation. 'Needless
to say, the contention that political science has no proper
concern with moral or ethical questions as such has been one of
the most bitterly argued aspects of behaviouralism' (Somit and
Tanenhaus, 1967, p.179).
Political science should become more interdisciplinary and draw
more on the other social sciences.

The reaction against behaviouralism came from three broad

directions. First, there were those who pointed out that behav-
iouralism had failed to fulfil its own goals. Even in an area such as
voting behaviour which 'is the closest thing to a scientific theory
that we have ... even a casual review of the findings of voting
research ... shows how unstable these regularities are, and how far
short of hard science our efforts to stabilize them must fall'
(Almond and Genco, 1990, p.37). Much of the time political scien-
tists were not observing actual behaviour, but trying to make sense
of reports of behaviour. Political scientists were not in the voting
booth with the citizen, but undertaking post hoc analyses of the
accounts voters gave of what they claimed to have done and their
reasons for it.
Second, behaviouralists were criticized for placing an 'undue
emphasis on process' at the expense of the 'content and substance
of political events and systems'. Moreover, 'As the behavioral
 The Discipline of Politics 17

methodologies became increasingly elaborate and complex, the

problems seemed to get ever more narrow and insignificant. The
very demand for precision of method imposed limits on the kinds of
subjects that could be dealt with' (Prinati, 1983, p.192). This kind of
criticism acquired additional force as the discipline developed a new
interest in the analysis of public policy, stimulated by work such as
that of Lowi (1964), Ranney (1968) and Pressman and Wildavsky
(1973).
Third, the ferment in American politics associated with the
upheavals caused by the civil rights movement and the Vietnam
War, and later by the growth of feminist consciousness, penetrated
the political science profession. 'In retrospect, the behavioral era in
political science appears to be a time of optimism' (Ricci, 1984,
p.171) both politically and within the profession. The Caucus for
a New Political Science set up within the APSA in 1967 attacked
the complacency, conservatism and lack of relevance of American
political science, rejecting the behaviouralist paradigm. In his 1969
address to the American Political Science Association, David
Easton tried to reconcile the various groups within the APSA by
referring to the 'post-behavioral revolution' and calling for more
applied research. Post-behaviouralism 'appeared on the political
scene as a phenomenon whose name ... indicated a shared deter-
mination to leave something definite behind rather than a common
notion of the direction in which the discipline should move
forward' (Ricci, 1984, p.189). Never again would a paradigm
enjoy such an unchallenged position of intellectual leadership in
political science.
In what ways did behaviouralism leave a mark on political
science? The impulse for a more scientific and rigorous approach
remained a force in American political science, as we shall see later.
It was generally accepted that the behaviouralist attempt to separate
value judgements and empirical research was doomed to failure.
Similarly, their view that 'regularities and generalizations are the
only proper objects of scientific political inquiry' came to be seen as
'an unnecessary delimitation of the discipline's subject matter'
(Almond and Genco, 1990, p.41). Behaviouralism could lead to the
neglect of 'the fact that much social and political change has to be
explained neither by strong regularities nor by weak regularities, but
by accidental conjunctions - by events that had a low probability of
occurring' (Almond and Genco, 1990, p.39). Indeed, new develop-
ments in chaos theory would suggest that very rare events will occur
more often than the Gaussian or standard bell-shaped curve model
would predict.
18 Research Methods in Politics

Three of the propositions made by the proponents of behav-

iouralism would probably win acceptance by most political scientists
today: that the study of politics should be theory-oriented and L
directed; that it should be self-conscious about its methodology; and~
that it should be interdisciplinary. At its worst, pre-war political
science offered low level descriptions of political structures which
treated constitutions as an accurate reflection of political reality.
Methodology was often unsystematic and rarely discussed. In those
respects, behaviouralism did contribute to the development of polit-
ical science as a discipline, although it was not necessary to have
been a behaviouralist to recognize the importance of theory and
method.
One response to the demiseof behaviouralism was the rebirth of
institutionalism in a seminal article by March and Olsen (1984).
They sought to argue that institutions do matter, that political
phenomena could not be simply reduced to the aggregate conse-
quences of individual behaviour, 'that the organization of political
life makes a difference' (March and Olsen, 1984, p. 74 7). The
choices that people make are to a significant extent shaped by the
institutions within which they operate. They sought to revive the
tradition within political science that saw political behaviour
'embedded in an institutional structure of rules, norms, expecta-
tions and traditions that severely limited the free play of individual
will and calculation' (March and Olsen, 1984, p.736). It was a way
of bringing the state back into political analysis, although in Europe
it had probably never left. New institutionalism 'insists on a
autonomous role for political institutions. The state is not only
morel
affected by society but also affects it' (March and Olsen, 1984,
p.738).
How does the new institutionalism differ from the old institution-
alism? March and Olsen are perhaps most vague and opaque on this
point, stating that the new and the old are not identical and that one
could describe 'recent thinking as blending elements of an old insti-
tutionalism into the non-institutionalist styles of recent theories of
politics' (March and Olsen, 1984, p.738). One difference is that old
institutionalism generally held a more narrow definition of what
constitutes an institution, seeing it principally in terms of formal
structures. 'For example, the old institutionalism argued that presi-
dential systems are significantly different from parliamentary
systems based upon the formal structures and rules' (Peters, 1999,
p.1). Such an explanation has some value, but by itself it is not suffi-
cient. Current work tends to define institutions more broadly 'as a
set of rules, formal or informal, that actors generally follow, whether
 The Discipline of Politics 19

for normative, cognitive or material reasons' (Hall and Soskice,

2001, p.9).
The experience of behaviouralism has probably also made the new
institutionalism more methodologically systematic. Old-style struc-
turalism tended to slip into the assumption that political structures
determined behaviour. The behaviouralists particularly objected to
this assumption and new institutionalists would see structure (more
broadly defined than by the old institutionalists) as a shaping and
constraining force rather than a dominant one. Where there is a line
of continuity with contemporary historical institutionalism is the
notion that institutions are historically embedded, influenced by
their formation and critical junctures in their development. Old insti-
tutionalism was highly normative and displayed a bias in favour of
the established democracies as offering models of good government
(Peters, 1999, p.13). New institutionalists do not display a prefer-
ence for one set of institutions over another; instead, they are more
interested in exploring how and in what ways institutions affect the
behaviour of their members.
New institutionalism has developed as a broad, almost protean
approach rather than as a school that sets guidelines for what
constitutes acceptable research. Peters ( 1999, pp.19-20) identifies
seven distinct variants of new institutionalism. One of the most
influential variants has been historical institutionalism. This has
two central hallmarks: an interest in 'the distribution of power in
the polity' (Hall and Taylor, 1998, p.960), reflecting what we have
emphasized as one of the central concerns of the study of politics,
and an emphasis on path dependency resulting from key historical
choices made by states. Critical moments create branching points
from which historical development moves on to a new path and
once that new path has been taken it is difficult to change track.
Thus for historical institutionalists, 'the basic point of analytic
departure is the choices that are made early in the history of any
policy, or indeed of any governmental system. These initial policy
choices, and the institutionalized commitments that grew out of
them, are argued to determine subsequent decisions' (Peters, 1999,
p.19). See also Box 1.1.
Developments in economics offered the possibility of a new
approach through what came to be known as rational or public
choice, drawing on the methodology of economics rather than the
sociological and psychological approaches favoured by behavioural-
ists (Ward, 2002, p.65). William Riker and his collaborators at
Rochester played a key role in introducing this approach to
American political science, and now 'It has arguably become the
20 Research Methods in Politics

Box-1.1

1. A broad undersranding of the relationsfl~p betweeh institutions and~-

individual behaviour, -
2. An emphasis on the asymmetries 0£ pow~ that arise'frorn the way in -
which instututions wmk. - -
3. !\view of institutionaldeveloprnent that.gmphasizes path dependence
arid unintended consequences.
4. Aii.,emphasis on the integratiop-of institutional analysis with that of-~:
ideas. -

Source:

dominant approach to political science at least in the United States' /

(Ward, 2002, p.65). However, its apparent emerging hegemony in
American political science provoked a counter-movement in the USA
known as the perestroika movement, although it also had other
concerns.
The rational choice approach was based on methodological indi-
vidualism and it was this core assumption that made it unpalatable
to many historical institutionalists. Like behaviouralism, it viewed
individuals as the key political actors and tended to reduce 'collec-
tive behaviour to individual behaviour' (Peters, 1999, p.16). A
second core assumption, borrowed from economics, is that individ-
uals are utility maximizers pursuing their own personal goals. Thus,
political actors, whether they are voters, politicians or bureaucrats,
'are assumed to be material interest maximisers, seeking benefits in
the form of votes, offices, power, and so on, at least cost' (Almond,
1990, p.123). Critics argued that like behaviouralism, there was an
emphasis on method rather than substance. Nevertheless, one of its
important insights was to draw attention to the 'free-rider' problem.
The rational pursuit of individual interests can make the collective,
co-operative action that politics requires difficult to achieve (Hay,
2002, pp.9-10).
One of the advantages the advocates of rational choice enjoyed
was that they were a more coherent group defined by a very clear
methodology compared with the more fragmented historical insti-
tutionalists who were less convinced of the innate superiority of any
one approach. To some commentators it seemed as if rational choice
was in the ascendant. Some of the criticisms made of the activities
of supporters of rational choice were rather exaggerated or
 The Discipline of Politics 21

alarmist. Rational choice has much to offer political science: for

example, one of the earliest models within this tradition, Downs's
median voter theorem, has turned out to be one of the most robust.
First published in 1957 (Downs, 1957), it can readily be applied to
contemporary elections. Rather than allowing the discipline to be
divided into two opposed camps, was some reconciliation between
the new institutionalism and rational choice possible? Some schol-
ars thought so.
Utility-maximizing individuals might find 'that their goals can be
achieved more effectively through institutions, and find their behav-
ior is shaped by the institutions' (Peters, 1999, p.44). Hall and
Taylor (1996, p.956) pointed to the way in which 'rational choice
analysts have begun to incorporate "culture" or "beliefs" into their
work to explain why actors move toward one outcome when a
conventional analysis specifies many possible equilibrium outcomes'.
Historical institutionalists could also benefit from an opening to
rational choice. 'Many of the arguments recently produced by this
school could readily be translated into rational choice terms' (Hall
and Taylor, 1996, p.957). Coleman and Tangermann (1999) offer an
example of a political scientist who had deployed historical institu-
tionalist approaches working with an economist to present an analy-
sis of agricultural trade negotiations in terms of a game theory
analysis of two autonomous, linked games. A narrative account of
the negotiations is linked with the theoretical perspective offered by
game theory.
Nevertheless, fundamental difficulties may frustrate any attempt
to generate a constructive dialogue between the paradigms of histor-
ical institutionalism and rational choice. Hay and Wincott (1998,
p.951) 'argue that the prospects for such a dialogue are more limited
than Hall and Taylor suggest [since] rational choice and sociological
institutionalism are based on mutually incompatible premises or
"social ontologies" '. A central concern of the new institutionalist
project is to restore collective action to the heart of political analysis
and this is at odds with the individualistic premises of rational
choice. 'For historical institutionalists, institutions are ontologically
prior to the individuals who constitute them' (Blyth and Varghese,
1999, p.355).
A further difficulty is that for rational choice theorists, the pref-
erences of actors are exogenous and the ways in which they are
derived is of relatively little interest. For historical institutionalists,
the way in which the rational actor is viewed is difficult to accept.
For them 'Actors cannot simply be assumed to have a fixed (and
immutable) preference set, to be blessed with extensive (often
22 Research Methods in Politics

perfect) information and foresight and to be self-interested and self-

serving utility maximisers.' (Hay and Wincott, 1998, p.954). Any
institutionalist version of rational choice would have to be
concerned with 'how individuals and institutions interact to create
preferences' (Peters, 1999, p.44).
At this point, values, culture and history would return in a way
which most rational choice theorists would find unnecessary or
superfluous. They would accept that they make simplifying assump-
tions that are sometimes distorting, but would argue that these are
necessary to construct elegant models, particularly at an early stage
of theoretical development. Even if they distort reality, their models
and methodology are more robust than the alternatives. Indeed, it
may be questioned whether the historical institutionalists have a
distinctive methodology.
One fundamental problem remains the question of how far ratio-
nal choice theorists are open to alternative approaches. Almond
suggests that they neglect other social science literatures 'that display
the variety of values, preferences and goals in time and space'
(Almond, 1990, p.135). In particular he argues:

The failure to relate the economic model of rationality in any way

to the sociological, psychological and anthropological literatures
and particularly to the work of Max Weber, whose great theoret-
ical accomplishment was an analysis of modern civilization and
culture in terms of rationality and rationalization, is the most
striking consequence of the almost complete 'economism' of the
rational choice literature. (Almond, 1990, p.134)

What is evident from our discussion so far is that political science

has been 'more eclectic than most disciplines in borrowing the
approaches of others' (Peters, 1999, p.20). Political science has to be
understood within the broader context of social science and it is to
some of the more significant methodological disputes within social
science that we now turn.

Political studies or political science:

some methodological considerations

Our review of the dominant paradigms of behaviourism, institu-

tionalism and rational choice has necessarily omitted close analysis
of other, less popular, heterodox contenders- such as Marxism,
 The Discipline of Politics 23

feminism, post-modernism and environmentalism. This list is by

no means exhaustive and points not only to the eclectic character
of modern political science but also to the divergent methodolog-
ical assumptions which exist within the discipline. These assump-
tions concerning the character of knowledge (epistemology) and
the nature of social reality (social ontology) may seem far removed
from a discussion of research methods yet, as Archer (1995, p.5)
rightly emphasizes, 'the practical analyst of society needs to know
not only what social reality is, but also how to begin to explain it,
before addressing the particular problem under investigation'. In
other words, before we can locate a researchable topic, and
choose the methods most appropriate to our study, it is important
to be aware of basic debates within the philosophy of social
science which will allow us to form a coherent link between epis-
temology/social ontology and research methods (Stoker, 1995,
pp.13-16; Grix, 2002). This is essential if we are to be consistent
and resist the temptation of 'perspectivism', which 'denies that
there are serious underlying reasons for theoretical variety and
slides via instrumentalism into a marriage of inconsistent
premises' (Archer, 1995, p.5). A useful way into these complex
debates is to consider what is commonly seen as the primal prob-
lem of the philosophy of the social sciences: to what extent can
society be studied in the same way as nature (Bhaskar, 1979,
p.107)?
The starting point of modern versions of this debate is the recog-
nition that the common sense 'man-in-the-street' worldview of
empiricism or positivism is wholly inadequate as a philosophical
theory. Positivism, as Benton (1977, p.11) points out, is a variant of
the philosophical theory of knowledge - empiricism - which is
usually attributed to John Locke. At its heart is the claim that all
knowledge is obtained from human sense experience. For classical
positivists, knowledge is gained frorri sensory obseryation (and is
therefore rooted in inductive reasoning); 'facts' exist independently
of the observer and his or her values; and the goal of science is thus
to build an objective empirical foundation for knowledge which will
produce testable and verifiable statements to explain, predict and
attribute causality to events and processes in the world. Finally, it is
claimed that all science, natural and social, is united in accordance
with these positivist principles and that if a proposition is neither
true nor false in relation to ascertainable fact nor by virtue of reason
or logic, it is, by definition, non-science (and by implication,
nonsense).
In terms of social ontology, positivism would seem to be perfectly
24 Research Methods in Politics

consistent with Individualism (the social whole can be ultimately

reduced to individual beliefs and dispositions) and, in principle but
to a lesser extent, with certain versions of Collectivism (the whole is
not reducible to the sum of its parts and emergent properties may
arise from the collective). The emphasis placed by this approach on
observation as the sole means of verification, combined in particu-
lar with a crude individualistic social ontology, gave it great appeal
to behaviourists in political science, and it continues to be popular
(though often unacknowledged) within the rational choice camp
(Sanders, 1995; Ward, 1995). A clear line can therefore be drawn
between this theory of knowledge and the practical application of
specific research methods (usually statistical, all conforming to the
strictures of verification, causality and prediction). However, posi-
tivism encounters problems when linked to a realist social ontology
inasmuch as it cannot entertain a commitment to 'ontological
depth' (hidden generative mechanisms at work in nature and soci-
ety) which, as we will see below, is the hallmark of many 'critical
realist' writers in the Marxist, feminist and environmentalist tradi-
tions.
This is not the place to rehearse the well known critique of posi-
tivism which emerged in philosophical circles in the post-war period
(Keat, 1981; Chalmers, 1982 and 1990; Hay, 2002, pp.80-1). It will
suffice simply to indicate that three broad traditions developed out
of this criticism, each having important implications for political
science and the researchers' choice of methods. The responses may
be categorized in terms of neo-positivism, conventionalism, and crit-
ical realism.
Neo-positivism continues to endorse the naturalistic claim that
there is a broad unity of method between the natural and social
sciences. However, through the work of Karl Popper, and later in the
hands of followers such as Imre Lakatos, neo-positivist writers came
to reject induction in favour of deductive modes of reasoning and
replaced verification with the notion of falsification (for an overview,
see Hay, 2002, pp.81-8). Through his focus on the inevitability of
conceptualization (and thereby the impossibility of empiricism)
Popper developed the 'hypothetico-deductive model' of scientific
reasoning. In this model, from a new idea, tentative conclusions are
drawn by means of logical deduction. With the help of other state-
ments previously accepted, certain singular statements (predictions)
are derived from the theory. These are then compared with the
results of practical applications and experiments. If positive, the
results are acceptable for the time being; if negative, then they are
falsified and rejected:
 The Discipline of Politics 25

So long as a theory withstands detailed and severe tests, and is not

superseded by another theory in the course of scientific progress,
we may say it has 'proved its mettle' or that it is 'corroborated'
... [but] I never assume that by force of 'verified' conclusions,
theories can be established as 'true' or even as merely 'probable'.
(Popper, 1959, p.33)

Falsification, Popper's solution to the Kantian problem of demar-

cation (distinguishing between scientific and non-scientific knowl-
edge), is not - as he admits - an absolutely sharp criterion since
there are degrees of testability. But, whilst science may admit 'well
testable' and 'hardly testable' theories, non-testable theories are 'of
no interest to empirical scientists. They may be described as meta-
physical' (Popper, 1963, p.257). In short, Popper proposes a clear
unity of method, 'all theoretical and generalizing sciences make use
of the same method whether they are natural or social sciences.
The methods always consist in offering 'deductive causal explana-
tions' and in testing them by way of predictions' (Popper, 1957,
p.131).
The Popperian model of science has been hugely influential in
the development of political science in both Europe and the USA.
It can, by and large, be seen as the received model of science
prevailing within the discipline (Hay, 2002, p.81). However it is
not without its critics who claim that Popper merely displaces the
problem of induction (since falsification rests on inductive reason-
ing); that he misunderstands the real development of natural
science; and that crucially he fails to acknowledge that knowing
subjects play an active role in constituting the world they know
(Habermas, 1972; Horkheimer, 1976; Chalmers, 1980; Held,
1980). Lakatos's (1970) attempt to develop 'sophisticated method-
ological falsificationism' (the 'hard-core' of a research programme
is not falsifiable whereas statements deriving from the 'positive
heurstic' are testable) may be able to more accurately account for
the development of 'real science' and yet the exercise remains
bound by Popper's neo-positivist straitjacket and the postulate of
methodological individualism.
The second response to the crisis of positivism is that which can
be labelled the conventionalist, humanist or hermeneutic. In
essence, it argues that the objects of study of the natural and social
worlds are so utterly different that they require totally different
methods and forms of explanation (Benton, 1977, p.12). The
philosophical lineage of this tradition is traceable through Weber
and Dilthey to the transcendental idealism of Kant (Bhaskar, 1979,
26 Research Methods in Politics

p.107). The most usually cited representative of this tradition,

whose work we will briefly review, is the philosopher Peter Winch.
Winch's argument focuses on the claim 'that all behaviour which is
meaningful (therefore all specifically human behaviour) is ipso
facto rule-governed' (Winch, 1958, p.52). The social 'sciences',
which thus claim to analyse aspects of human behaviour, are in
essence studies of rule-following. It may be appropriate for the
natural scientist to follow the rules of scientific method but, Winch
claims, for the social scientist the object of study, as well as the
study of it, is a human activity and is therefore carried out accord-
ing to rules. It is these rules, rather than the abstract rules of
science, which must be understood to gain knowledge of social
activity. In short, the 'nature of this knowledge must be very differ-
ent from the nature of knowledge of physical regularities' (Winch,
1958, p.88). Concepts in social science are thus, in Alfred Schutz's
terms, 'second order constructs': that is, they are built on 'the
actual common sense first order constructs of actors in their daily
lives' (Phillipson, 1972, p.144). It follows that actions are only
intelligible 'in the context of, ways of living or modes of social life
... one cannot apply criteria of logic to modes of social life as
such' (Winch, 1958, p.100). Two important consequences flow
from Winch's analysis. First, since voluntary behaviour is behav-
iour to which there is an alternative (and since following a rule
involves interpretation and allows for the possibility of making a
mistake), prediction, in the sense implied by Popper, is ruled out in
social science. Second, Winch rejects Popper's claim that it is possi-
ble to develop a causal analysis of action. The relationship between
an act of command and an act of obedience is not the same as the
relation between thunder and electrical storms, since human
actions belong to the realm of concepts and 'the relation of belief
to action is not external and contingent, but internal and concep-
tual' (Winch, 1958, pp.124-5; Macintyre in Laslett and Runciman,
1962, p.52). Theory construction in social science is different in
kind from that found in natural science and ultimately events can
only be understood by rejecting Popperian forms of methodologi-
cal individualism in favour of a focus on 'modes of social life' or
context-specific Wittgensteinian 'language games' (Winch, 1958,
p.128).
This broad approach in which human beings are seen as distinc-
tively 'free subjects', as the agents of 'meaningful acts', and as the
'creators' of their social world (Benton, 1977, p.12) has spawned a
whole new tradition of social and political analysis based on the
ideas of indexical (context-specific) reasoning and situational logic.
 The Discipline of Politics 27

For example, the phenomenological and ethnomethodological

traditions which focus on the taken for granted, the mundane level
of the intersubjective world of everyday life, (Garfinkel, 1967;
Filmer et al., 1972), have pioneered the development of discourse
analysis and the study of natural language in political settings
(Heritage, 1984; Howarth, 1995). The inherent 'relativism' of
Winch's position has actually been embraced as an antidote to posi-
tivism by a number of radical sociologists committed to techniques
of participant observation and new feminist methods designed to
'hear women's voices' (Reinharz, 1992). Finally, it has helped
bolster wider conventionalist critiques of science such as that artic-
ulated by Michel Foucault (1980), Paul Feyerabend (1975) and, in
particular, Thomas Kuhn (1962); Kuhn's view that science is not a
set of universal standards guided by logic and reason but rather is
part of a specific form of culture and its conventions has itself
gained widespread acceptance, particularly in feminist circles
(Harding, 1986).
Critical realism, the third major response to the crisis of classi-
cal positivism, breaks with both Popper and Winch and argues for
a position best characterized as 'qualified anti-positivist natural-
ism' (Bhaskar, 1979, p.108; 1989; Hay, 2002, pp.122-6). In other
words, it is possible to give a common account of the development
of science but it does not deny that there are important differences
in methods grounded in the real differences that exist in their
subject matters. While critical realists may disagree on the precise
character of the social ontology they wish to endorse, they are
united in their rejection of pure forms of Individualism and
Collectivism, pointing instead to notions of ontological depth,
stratification and emergent group properties (Benton, 1977;
Archer, 1995, p.159). There is a unity of method inasmuch as
scientific explanations must make reference to 'underlying struc-
tures and mechanisms' involved in the causal process, but this idea
of 'generative mechanisms underlying observed phenomena' (Keat
and Urry, 1975, pp.42-3) provides, it is claimed, a basis for causal
analysis without the problems which beset the positivist regularity
thesis. Since the objects of social science manifest themselves in
'open systems' (Bhaskar's term for systems where 'invariant
empirical regularities do not obtain': p.12 7) social sciences are
denied 'decisive test situations' for their theories. Thus the criteria
for the confirmation or rejection of theories cannot be predictive
and so must be exclusively explanatory. However, the 'openness'
of social systems also indicates that our knowledge is 'necessarily
incomplete' (Bhaskar, 1979, p.130). In short, society is not, as the
28 Research Methods in Politics

positivists would claim, a mass of separable events and sequences;

and neither is it constituted by the momentary context-specific
meanings we attach to events. Rather, 'it is a complex and causally
efficacious whole - a totality, whose concept must be transformed
in theory, and which is being continually transformed in practice'
(Bhaskar, 1979, p.134 ). As an object of study, it cannot be read
straight off the empirical world, yet its conclusions are subject to
empirical check and its effectiveness in explaining events is not
merely accidental (Sayer, 1992, p.4).
This approach, although not without its cntics (Hay, 2002,
pp.122-6), has proved remarkably popular among critical social
scientists. It seems to sidestep the problems associated with posi-
tivism, whilst at the same time providing a coherent basis for
understanding the development of science throughout the ages
without resorting to the sociological simplicities of Kuhnian
conventionalism. Moreover, its rejection of positivism's a priori
restriction of knowledge to that which can be observed allows for
the reintroduction of concepts such as 'ideology', 'hegemony' and
the 'power of capital' into social science (Benton, 1977). In fact,
both Marxism and psychoanalysis can, on this reading, be read-
mitted into the scientific community. In terms of methods, it should
be clear that critical realists will not rest content with simply
recounting 'actor's views' or with bald statistical presentation;
instead, the aim must be to reject surface explanation and, wher-
ever possible through the use of primary sources (particularly
documentary material and elite interviews), reconstruct and rein-
terpret the events under investigation.
Earlier in this chapter it was proposed that the discipline of
politics may be usefully seen as a 'junction subject' born out of
history and philosophy but drawing on the insights of cognate
disciplines such as economics and sociology. This openness has
indeed been a strength, fostering not only interdisciplinary work
but also reflection on the methodological problems which the
discipline shares with the other social sciences. This chapter has
sought to show that whilst these 'meta-theoretical' problems
cannot easily be resolved, researchers in the field of politics will
benefit from engagement with such issues (Grix, 2002).
Methodological pluralism, in the sense of enabling all approaches
to flourish, is certainly to be encouraged (Bell and Newby, 1977);
yet it must also be distinguished from the slide into the theoretical
morass of eclectic pragmatism where 'anything goes' and perspec-
tives are sampled regardless of methodological consistency.
Appreciating both the historical development of a discipline and
 The Discipline of Politics 29

the links between epistemology/social ontology and research

methods is the necessary first step towards becoming a competent
researcher. Once these tasks are accomplished the next stage is to
focus on the complex issue of research design.