https://doi.org/10.1177/0192512119858370
International Political Science Review
1­–17
© The Author(s) 2019
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0192512119858370
journals.sagepub.com/home/ips
Consensus and dissensus
in comparative politics: Do
comparativists agree on the goals,
methods, and results of the field?
Andrew Roberts 
Northwestern University, USA
Abstract
Are comparative political scientists divided over the goals, methods, and results of their field? This article
attempts to answer this question, drawing on an original survey of US-based political scientists. The
main conclusion is that there is relative consensus on the goals of research—comparativists favor broad
generalizations and causal inference—but there is also acceptance of a variety of methodological approaches,
both qualitative and quantitative, in pursuing this goal. Comparativists, however, show less agreement on
substantive findings in the areas of democracy and democratic politics, economy and society, and political
institutions. Interestingly, generational differences are relatively infrequent, but gender differences on issues
such as rational choice and causal inference are more prominent, possibly contributing to gendered citation
bias. The findings suggest that comparative politics may not have accumulated a large amount of agreed-upon
knowledge, but that there is substantial agreement on the path forward.
Keywords
Comparative politics, expert survey, methodology, political scientists
Introduction
Comparative politics is widely regarded as a field divided. In its starkest terms, one side sees comparative
politics as a scientific discipline whose goal is building parsimonious models of politics
that are subject to rigorous, quantitative testing. The other conceives the field as part of area studies
and favors a holistic understanding of the histories and cultures of individual polities. These poles
overlap with other more specific debates: the existence of a single logic of causal inference or
multiple logics (Goertz and Mahoney, 2012; King et al., 1992), the usefulness of rational choice
modeling in understanding politics (Friedman, 1995; Green and Shapiro, 1994), and the relative
Corresponding author:
Andrew Roberts, Department of Political Science, Northwestern University, Evanston, IL 60208, USA.
Email: aroberts@northwestern.edu
858370IPS0010.1177/0192512119858370International Political Science ReviewRoberts
research-article2019
Article
2	 International Political Science Review 00(0)
importance of methodological sophistication versus deep knowledge of places (Bates, 1997;
Geddes, 2003; Johnson, 1997).
Yet, the widespread belief in these divisions has rarely been subjected to empirical confirmation.
Are there really two such poles and how far apart are they? Though these debates received
heavy airplay in the 1990s, the battle seems far less intense in the new millennium and especially
among younger scholars. But again this impression is an anecdotal one, and it is not clear if current
collegiality is due to one side winning the battle, a synthesis of the two poles, or simply increasing
civility.
A similar problem applies to the substantive results of comparative politics. Some have suggested
that political science has produced little well-confirmed and useful knowledge of politics
(Mead, 2009). Is this true? Does the field have good answers to questions like how democracy
emerges, the causes of conflict, or the institutions that produce growth, to name just a few central
issues in comparative politics? Substantive agreement is important because many comparativists
are committed to improving public policy and advocacy of policy changes should be based on
theories that receive broad support.
The aim of this article is therefore twofold. First, it attempts to assess whether comparativists
are divided on the goals and methods of the field, particularly between those who embrace a scientific
worldview and quantitative testing and those who advocate area studies and qualitative methods.
Second, it looks at whether comparative politics has produced consensus knowledge about
how politics actually functions. This means major propositions such as the natural resource curse,
consociationalism as a solution to ethnic conflict, or the perils of presidentialism. Little existing
work has taken on either of these questions.
The article attempts to answer both of these questions using an expert survey of professors of
comparative politics at US universities. Such surveys have become increasingly common as a way
of assessing academic consensus (Maliniak et al., 2011).
The headline finding of this survey is that there is a good degree of consensus about the goals
and approach of comparative politics, at least among US-based comparativists, but that much of
the consensus comes from an acceptance that there is more than one way to skin a cat. Thus, while
large majorities agree with the scientific turn in comparative politics and see its goal as producing
generalizations about politics, area studies and culture are likewise seen as important. The same
can be said about quantitative and qualitative methods. Backlash only emerges when a particular
approach is raised as the one, true way. Comparativists are united both in their scientism and in
their methodological eclecticism.
Turning to substantive results about how politics actually works, consensus moderates considerably.
This part of the article focuses on 37 research results drawn from the areas of democracy and
democratic politics, the economy and society, and political institutions. With some exceptions—in
particular, the relationship between development and democracy, the importance of collective
action problems, and veto players—there is much less agreement on prominent theories. This may
be due to the fact that the theories mentioned in the survey are too general and abstract. It may also
be because politics works in different ways in different places. In fact, there appeared to be more
consensus on the functioning of politics in advanced industrial societies which have been the object
of more research than the developing world and are arguably less diverse.
Finally, the article considers group differences in these opinions. While one might expect divergence
between older and younger comparativists due to differences in training, in fact, the survey
revealed larger differences between male and female comparativists, a finding which has important
implications for gender biases in publication and citation. In sum, comparative politics may not be
a field divided, but it is a field with some divisions.
Roberts	 3
A divided field?
What do we know about the degree of consensus in comparative politics? There is widespread
belief among practitioners that political science is a fragmented field with little holding it together.
Almond (1988) gave this phenomenon the name ‘separate tables’. He argued that political science
was “not a happy profession” (Almond, 1988: 829) and was separated along the two dimensions of
ideology (left/right) and methodology (hard/soft). At the same time, however, he claimed that ‘the
overwhelming majority of political scientists are somewhere in the center – “liberal” and moderate
in ideology, and eclectic and open to conviction in methodology’ (Almond, 1988: 830).
What about comparative politics specifically? The subfield has been divided by several controversies
in the past decades. Most prominent are the debates over area studies, rational choice, and
the applicability of statistical research practices to qualitative research. While these debates, as
well as many others within particular substantive areas, have yielded considerable fire, have they
produced a fragmented subfield or has comparative politics maintained in Almond’s phrasing a
‘cafeteria of the center’where most can dine? There has been relatively little empirical research on
this question.
Anumber of studies have considered what published articles can tell us about the methodological
predilections of the field. Sigelman and Gadbois (1983) looked at 565 articles in comparative politics
journals from 1968 to 1981 and found that the field was Eurocentric and dominated by singlecountry
studies. Hull (1999) followed up with an analysis of 727 articles from 1982 to 1996 and
found that the field had become more comparative – ns of more than one became more common
though single-country studies were still a plurality. Western Europe maintained a strong but slightly
diminished position. Hull meanwhile found that no ‘delineated research program’ dominated the
field and in fact only one in seven and a half articles utilized the most relevant paradigms.
In the most recent survey of articles, Munck and Snyder (2007) found that most work in the
main comparative politics journals used inductive and qualitative methods.1 They saw little evidence
of a turn to economic or deductive reasoning. Most scholars worked within a particular
geographic region, but area studies were not set apart from non-area studies work. Munck and
Snyder (2007: 6), however, echoed the common view that ‘comparativists hold widely divergent
views about the nature of the problems facing their field and the solutions required to realize more
fully the potential of comparative politics to produce knowledge about politics around the globe’.
However, the diversity of approaches in published articles does not necessarily imply that the field
is divided, as I argue below.
In contrast to the multiple studies of topics and methods, there is very little study of the second
research question, the level of agreement on substantive topics. One could cite the many literature
reviews in the Annual Review of Political Science, the Oxford Handbook series, as well as most
individual articles, which typically summarize what the field knows about a particular topic. But
these assessments of consensus are mainly subjective, do not follow a standard rubric, and are
limited to a single or handful of propositions. In short, there has been little systematic study of the
degree of substantive consensus within comparative politics.
Methods
A number of methods exist for measuring the degree of consensus in a field. The studies mentioned
above code journal articles by topic and approach. The drawback of this approach is that it captures
what scholars do rather than what they know and what they do represents only a small portion of
their knowledge. Further, none of these studies looked at the actual findings of these works and the
degree to which these findings were widely accepted.
4	 International Political Science Review 00(0)
A more productive approach and one that has been applied in other fields is an expert survey. This
allows researchers to capture the range of scholarly views on key debates in the discipline. This
approach may suffer from scholars’lack of familiarity with many areas of the field and difficulties in
constructing a representative sample, but it more fully utilizes the knowledge of scholars.
This method has been widely used in economics where scholars have been periodically surveyed
about their views on matters of economic policy with the typical finding being one of relative
consensus (Frey et al., 1984; Fuller and Geide-Stevenson, 2003).2 Surveys of political scientists
are less common. The main exceptions are the surveys on Teaching, Research, and Policy in
International Relations (TRIP) (Maliniak et al., 2011). The TRIP survey, however, asks few questions
on methods and particular findings in international relations, the main exceptions being items
on paradigms and US foreign policy.
To date there has been no survey of comparativists about the methods and findings in the field.3
This article attempts to fill this gap. The aim is to determine the degree to which comparativists
agree or disagree on the goals, methods, and results of the field. To that end, I developed 61 statements
about comparative politics covering five general areas: goals and approach, methodology,
democracy and democratic politics, economy and society, and political institutions. The statements
were intended to capture major perspectives which might garner widespread agreement.
They were drawn from textbooks and graduate reading lists in comparative politics as well as
discussions with colleagues. The statements were framed as an advocate of that point of view
might state their position.
Respondents were asked to rank their agreement with these statements on a five-point scale
ranging from strongly agree to strongly disagree along with a separate no opinion option.4 The full
list of statements and summary statistics are included in the online appendix. The statements were
divided into the five groups mentioned above, each consisting of approximately a dozen statements.
The order of the statements was randomized within each section of the survey.
The initial intention was to send the survey to members of the comparative politics section of
APSA, but this mailing list was unavailable to researchers at the time. Instead, I compiled a list of
comparative scholars at the top 100 political science departments and the top 100 liberal arts colleges
as rated by US News.5 While some department websites listed comparative politics faculty
separately, others did not. This sometimes made it difficult to distinguish international relations
faculty from comparative politics faculty. I erred on the side of inclusion – for example, I included
any scholar with a regional specialization as well as formal theorists working on topics relevant
outside the US – but I asked scholars at the conclusion of the survey about their primary field of
research.
I emailed an invitation to the survey to 1385 comparativists at 200 universities and colleges in
October 2011 and sent two reminder emails at two-week intervals.6 I received 474 responses for a
response rate of 34%. The response rate is similar to previous surveys of the economics profession.
This rate may be understated given that multiple scholars wrote to say that they did not consider
themselves comparativists. Just short of 90% listed their primary specialization as comparative
politics with another 9% naming international relations.
Although accurate demographics for the universe of comparativists teaching at US universities are
unavailable, the respondents seem to be a reasonable match for the field with the exception of age.
There was a bias towards relatively younger scholars: 40% received their PhD in the 2000s, 26% in
the 1990s, 15% in the 1980s, and 13% before that time. Two-thirds of the respondents were male and
one-third female. Academic ranks were divided almost evenly between full, associate, and assistant
professors. Most of the respondents (84%) came from research universities rather than liberal arts
colleges.7 A variety of methodological approaches were represented (respondents could name more
than one): 68% chose qualitative, 56% quantitative, 66% mixed methods, 18% formal modeling,
15% experimental, and 20% interpretive methods. Regional specializations matched previous studies
Roberts	 5
of journal articles (again respondents could name more than one): the leaders were Western Europe
(37%), Latin America (29%), North America (19%), Eastern Europe (17%), Russia/Post-Soviet
(15%), East Asia (15%), and Sub-Saharan Africa (14%).
Results
In presenting the results, I will focus on the degree of consensus or dissensus on the statements. As
a rough heuristic, I distinguished three different categories. A strong consensus means that more
than two-thirds of respondents were on one side of the divide – choosing either the two ‘agree’ or
the two ‘disagree’ options on the five-point scale. The middle category was not included on either
side. By contrast, a moderate consensus means that between half and two-thirds of respondents
favored one side. The third category was dissensus which means that neither agree nor disagree
received a majority. One might also consider a fourth category of polarized dissensus where the
dissensus extends to the extreme categories of strong agreement or strong disagreement, but no
question elicited this sort of polarization.
Goals and approach
I begin with the goals and approach of the field which are presented in Figure 1. Here I probed
whether comparativists agree in the abstract on what they should be doing – not the specific methodologies
but the general outlook of the field. This included the debates between science and area
Figure 1.  Goals and approach.
33.7
47
10.1 7.5
1.7
32.8
27.4
13 14.5 12.3
18.1
45.4
22.7
12.9
.9
33.9
42
11.5 8.3 4.3
18.9
39.8
17.8 18.3
5.2
46.4
34.8
12.4
5.2
1.3
15
32.4
21.5 21.3
9.8
30.1
45.8
14.6
7.8
1.7
39.5 36.4
18.2
4.6 1.3
36.4
46.1
11.9
5
.6
18.9
40.6
16.6 15.5
8.4
22.4
55.1
12.7 8.6
1.3
0
20
40
60
0
20
40
60
0
20
40
60
0
20
40
60
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Aim is broad, general propositions Aspire to be a science CP more divided than unified
Causal relations main goal Culture an important explanation Deep knowledge of area
Methodological individualism Office seeking assumption Participate in public debates
Path dependence common Rationality assumption useful Significant progress in CP
Note: The bars represent the percentage of respondents who chose options on a five-point scale ranging from strongly
agree to strongly disagree. “Don’t know” responses were excluded from the calculation of percentages.
6	 International Political Science Review 00(0)
studies and between rationality and culture.8 Surprisingly, given controversies in the field (and the
fact that some of the propositions were seemingly contradictory), there was widespread agreement
on most of the propositions. Seven of the twelve statements I considered received greater than 75%
agreement (again combining the strongly and somewhat agree categories) and only two elicited a
degree of controversy. Political scientists accept a variety of approaches and dismiss very few.
The clearest statement about the goal of the field – ‘Comparativists should aim to develop
broad, general propositions about politics’ – met with nearly universal support – 81% in favor and
only 9% disagreeing. A nearly as strong group of 76% agreed with the idea that what political scientists
are seeking is causal relations. Asking more specifically about whether comparative politics
should aspire to be a science, where that term was left undefined, agreement was a strong but less
overwhelming 60% with 27% disagreeing.
Two other propositions received even higher levels of support – those that asked whether comparativists
‘should have deep knowledge of a country or region’(81% agreement) and whether path
dependent processes are ‘common’ in politics (82% agreement). One can see a bit of two-mindedness
when juxtaposing these statements to the ones in the previous paragraph. In the past, the main
debate in comparative politics was between those who favored ‘broad, general propositions’ and a
focus on causality and those who favored area studies knowledge and more descriptive or interpretive
approaches.9 The judgment of the profession appears to be that both sides were correct, that
both are needed.
Several of the statements probed the approach of the field a bit more concretely by looking at
several assumptions about how to approach research questions. The strongest agreement was for
the proposition that the ‘primary’ goal of politicians is to stay in office where 76% agreed.
Curiously, equal percentages of respondents believed that rationality was a ‘useful’ assumption to
make (59% in favor and 24% against) and that culture was an ‘important explanation of political
differences between countries’ (59% versus 23%). The fact that both items were stated in inclusive
ways likely affected the degree of acceptance. The most divisive statement in this section was
whether the field ‘should’practice methodological individualism to achieve full explanations (47%
agreement versus 31% disagreement).
Looking at the subfield from the outside, there was considerable agreement with the proposition
that the field is more divided (63%) than unified (14%). This judgment fits the conventional wisdom
described earlier, but contradicts the many areas of agreement in the survey. More optimistically,
77% agreed that the field had made ‘significant progress’in the past decade. As an indication
of their faith in this progress, 76% of respondents thought that comparativists should participate in
public debates about policy.
To summarize this section, comparativists seem to have both embraced the more recent ‘scientific
turn’ in the field and the rationality assumption, but they still see the value of area studies and
culture. This seems to be a case of both/and rather than either/or. Indeed, factor analysis of the
results indicated that only a single factor explained most of the variance in responses. Comparative
political scientists still want to know about places and history, but they want to do so in a more
scientific way.
Methodology
In conventional descriptions of the field, the largest divisions are methodological ones between
quantitative and qualitative scholars. These divisions do show up in the survey, but they are not as
deep as one might expect. There is relatively widespread acceptance of King, Keohane, and Verba’s
attempt to develop a universal methodology, but at the same time, members of the subfield have
not gone all-in on quantitative and formal methodologies. Eclecticism still rules and propositions
Roberts	 7
that were phrased in more hegemonic terms were typically rejected; respondents saw the value in
a variety of approaches. Figure 2 presents these results.
The headline finding is that most comparativists do not believe that ‘political phenomena can
and should be quantified’ (only 28% agreed and 52% disagreed). The wording of this statement
may have been imperfect since disagreement does not imply a rejection of quantitative techniques,
but it does indicate that comparative politics has not fully embraced the quantitative revolution.
Greater support was found for several qualitative propositions that were phrased less categorically.
66% of respondents agreed that a case study was a ‘legitimate’ way to confirm a hypothesis and
60% agreed that mixed methods were ‘the best route’ to produce knowledge.
Turning to other approaches, several achieved widespread acceptance. 60% thought that experimental
methods had an ‘important role’ to play and a very high 75% agreed that thick description
was a ‘valuable’ method. Meanwhile, there was widespread rejection of the idea that empirical
research ‘should be grounded’ in formal models – only 13% favored the technique that is standard
in economics (Granato and Scioli, 2004). Here again, use of the word ‘should’ may have prompted
a backlash as was the case with quantification and methodological individualism. Similarly controversial
was the suggestion that deterministic causation (necessary and sufficient conditions) was
‘common’ in politics – 33% agreed and 47% disagreed.
Unfortunately, it is difficult to gauge relative support for different methodologies from these
statements. This is due to a limitation of the survey design. Each technique was described in different
terms which limits inter-item comparability. The reason was my attempt to phrase the
Figure 2. Methodology.
17.4
28.5
19.8 19.4
14.9
8
20.3 19.2
26.9 25.6 25.1 26
22 19.2
7.7
25.3
41
13.5 12.2 8
16.1
33.2 32.9
13.8
4 5.1
27.5
19.9
29.5
17.9
2.7
10
20.7 24.9
41.6
20
39.6
19.1
15.1
6.2
21.3
45.6
19.1
11.3
2.7
27.2
33.2
23.2
11.8
4.7
41.7
29.2
10.7 12.5
5.8
37.3 37.5
13.2 9.9
2
0
20
40
60
0
20
40
60
0
20
40
60
0
20
40
60
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
CP overvalues methodological skill Can and should quantify Can't create and test theory on same data
Case study can confirm hypothesis Causal inference based on counterfactual Deterministic causation common
Empirics grounded in formal models Important role for experiments Increase observations for better inference
Mixed methods are best Need variation on dependent variable Thick description valuable method
Note: The bars represent the percentage of respondents who chose options on a five-point scale ranging from strongly
agree to strongly disagree. ‘Don’t know’ responses were excluded from the calculation of percentages.
8	 International Political Science Review 00(0)
propositions in ways that an advocate of each approach might. Thus, an advocate of experiments
would likely claim that experiments have an ‘important role’ to play rather than that they are
unequivocally the best method. This technique likely introduced a bias in favor of consensus. In
future iterations, a better approach would be to ask a single question of the form: ‘Rate the usefulness
of the following techniques for producing knowledge in comparative politics’. The best
we can say here is that there is acceptance of a variety of approaches and a rejection of any one
true way of doing things. The field remains eclectic.
More useful results were found for four propositions that attempted to capture the main methodological
advice of King, Keohane, and Verba (1992, hereafter KKV).10 The conception of causality
as involving counterfactual reasoning received solid but not overwhelming support—49%
agreed and 18% disagreed with a fairly large percentage, 33%, standing in the middle (perhaps due
to lack of familiarity with this conception). Large majorities did agree with two other propositions:
that variation on the dependent variable is necessary for showing causality (71% versus 18%) and
that scholars should try to increase their number of observations (67% versus 14%). There was
reasonable agreement with the tenet that scholars should not create and test a theory on the same
set of data (51% in favor and 27% opposed). Overall, the KKV approach received considerable
support.11 Even if US-based comparativists remain eclectic in methods, certain principles are seen
as applying to them all.12
Democracy and democratic politics
I now turn to substantive propositions about politics. Here the question switches from how to produce
knowledge to whether we have produced knowledge. I begin with theories related to democracy.
The conceptualization, causes, and effects of democracy might be characterized as the heart of
comparative politics, a sort of holy grail. As Figure 3 shows, comparative politics is still a considerable
distance from the grail. While a handful of propositions about democracy received considerable
support, many more elicited division. There is much we still need to learn about democracy, both its
causes and effects. Indeed, even the best way to define democracy was controversial with 36%
endorsing the minimalist definition pioneered by Schumpeter and 34% disagreeing with it.
By far the strongest agreement came on offshoots of modernization theory. There was 59%
agreement with the classic version of the theory that the likelihood of a democratic transition
increases with a country’s wealth and only 22% disagreement. Even stronger was the overwhelming
84% agreement (and only 8% disagreement) with Przeworski and collaborators’ revised version
that democracy is more likely to survive in wealthier countries. Indeed, 36% of respondents
strongly agreed with this proposition. Comparativists seem to agree that income matters for democracy,
despite several recent claims to the contrary.
Only one other proposition about democracy received strong consensus support. The idea popularized
by Putnam that the strength of civil society is a ‘major determinant’of how well democracy
functions was supported by 72% of respondents and opposed by only 10%. There was also fairly
substantial but not overwhelming opposition to the idea that an Islamic religious tradition inhibits
democracy – 52% thought that it was not true and only 21% agreed, though a substantial group
(27%) chose to neither agree nor disagree.
The remaining propositions on democratization elicited much more controversy. There was
considerable ambivalence about the idea that democracy can be crafted in any country (almost
equal numbers in favor and opposed to this proposition) and that external intervention can help to
promote democracy (32% agreed and 40% disagreed). Two well-known theories also received only
lukewarm support. Huntington’s claim that order must come before democracy elicited only 39%
agreement and 30% disagreement. Similarly, the popular idea of a natural resource curse was supported
by 37% of respondents and opposed by 30%.
Roberts	 9
Turning to the functioning of a democratic order, there was reasonable endorsement of the spatial
model of voting –‘voters choose the party closest to them in policy space’ (51% versus 21%)
– and that election outcomes are ‘primarily determined’ by economic conditions (47% versus
29%). This is somewhat surprising as one might view them as alternative theories of voting.
Comparativists were more ambivalent about the influence of public opinion on policy in democracies
with approximately equal numbers on each side (38% agreeing that policy follows public
opinion versus 35% disagreeing). There was some approval of the idea that democracy promotes
economic development (41% in favor), but an even larger group would neither agree nor disagree
(43%), a claim that fits with the ambiguous findings in the literature.
If asked what they know about democracy, comparativists might say that income and civil society
matter and that they have some understanding of voting. As I suggest below, one interpretation
of the lack of consensus, particularly on democratic politics, is that the propositions typically did
not specify a domain. If politics works in different ways in different places, then respondents might
be disagreeing simply because the theory applies in some places but not others.
Economy and society
If democracy is the holy grail of comparative politics, the understanding of economic prosperity and
conflict are not far behind. Large literatures in the field consider how to best attain a prosperous and
Figure 3.  Democracy and democratic politics.
19.3
53.1
17.2
9.4
.9
7.8
32.2
19.3
27.5
13.2
5.5
35.1
42.7
13.5
3.2
4.6
42.9
23.9 22.2
6.4 4.5
27.7 27.9 25.8
14.1
2.2
19
26.7 32
20
8.3
27.8 29.5
23.1
11.3 6.8
30.2 32.8
22
8.2 9.5
29.3 30.7
21.4
9
1.8
36
27.1 28.9
6.2 4.2
46.4
28
17.7
3.7
13
46.5
18.8 17.1
4.6
36.4
47.5
8 6.6 1.6
0
20
40
60
0
20
40
60
0
20
40
60
0
20
40
60
0
20
40
60
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree
Civil society key for democracy Democracy in any country Democracy promotes development
Economy determines elections Interventions can create democracy Islam hinders democracy
Minimalist definition of democracy Natural resource curse Order should precede democracy
Policy follows public opinion Spatial voting Transition increases with wealth
Transitions survive in rich countries
Note: The bars represent the percentage of respondents who chose options on a five-point scale ranging from strongly
agree to strongly disagree. ‘Don’t know’ responses were excluded from the calculation of percentages.
10	 International Political Science Review 00(0)
equitable economy and the causes of and solutions to civil conflict. Though it would be impossible
to include statements of all the major theories pertaining to these subjects, a handful of prominent
theories were included in this section. Unfortunately, few garnered a wide consensus except a handful
pertaining to advanced industrial economies and some propositions about the origins of ethnicity
and ethnic conflict. Figure 4 presents these results.
There was relatively little consensus on the classic question of why some countries are rich and
others poor. Respondents showed slightly more agreement with the institutional/property rights
theory currently associated with Acemoglu and Robinson and North than with the developmental
state approach pioneered by Gerschenkron (with modern exponents including Johnson, Evans, and
many others). Results showed that 49% agreed and 27% disagreed with the proposition that differing
levels of development are ‘best explained by institutional factors, particularly the protection of
property rights’. A somewhat smaller 41% agreed and 27% disagreed with the proposition that an
‘autonomous and interventionist state is a key element in successful economic development’. A
final statement about development considered whether concentrated executive power is necessary
for economic reform. Here more disagreed (43%) than agreed (27%). In sum, there is little consensus
on the causes of development, at least at this level of generality.13
There was much stronger support for the three propositions about the political economy of
advanced industrial states. Respondents strongly agreed that these economies exhibit complementarities
which can be characterized as distinct varieties of capitalism (73% support versus 10% disagreement),
that strong leftist parties and labor unions are the ‘key cause’ of the welfare state (63%
Figure 4.  Economy and society.
1.2
8.3 10.9
31
48.7
23.8
48.7
12.9 12
2.6 2.3
26.2
35.9
26
9.7
13.1
44.6
21.1
16.9
4.4 1.5
14.9
36.2 32.9
14.4
9.7
38.9
24.4 22
5.1
14.4
48.4
18.5 16.8
1.9
27.2
36.2
10.9
18
7.8 10.5
48
19.4 18.4
3.7
7.8
33.5 31.3
19.8
7.6
2.8
23.8
30.3 31.5
11.5
16.4
56.3
17.7
8.6
1
0
20
40
60
0
20
40
60
0
20
40
60
0
20
40
60
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Clash of civilizations Collective action problem important Consociationalism solution to ethnic conflict
Ethnic conflict instrumental Ethnic diversity hurts growth Institutions best explanation for development
Left power key for welfare state Nationalism modern phenomenon Postmaterialist values important
State key to development Strong executive helps economic reform Varieties of capitalism
Note: The bars represent the percentage of respondents who chose options on a five-point scale ranging from strongly
agree to strongly disagree. ‘Don’t know’ responses were excluded from the calculation of percentages.
Roberts	 11
support versus 19% disagreement), and that postmaterialist values play a ‘significant and increasing
role’in the politics of industrialized countries (58% support versus 22% disagreement).14 These high
levels of consensus may reflect the large amount of attention that comparativists have devoted to
these countries. Past literature reviews have shown that studies of Western Europe dominate the
subfield. That sustained attention to a group of countries can yield such a consensus is both a sign
of hope and a call for more scholars to turn to the study of less developed countries.
Moving from political economy to conflict and its roots, there is considerable consensus on the
nature and effects of nationalism and ethnicity, but less on solutions to conflict. The argument
associated with Gellner and Anderson that nationalism is a modern phenomenon received widespread
support (63% versus 26%). So did the idea that ethnic conflict is explained more by instrumental
calculation than cultural factors (58% versus 21%). Meanwhile, Huntington’s claim that a
clash of civilizations explains recent global politics received the least support of any proposition in
the study (9% agreeing versus 80% disagreeing). This consensus on ethnic conflict should perhaps
be more fully conveyed to journalists who trumpet ancient hatreds and civilizational divides in
reporting on current conflicts. On the origins of the state, meanwhile, more agreed than disagreed
with the Tillyean ‘war makes states and states make war’, but not by much (43% versus 36%).
The most unfortunate result here is the lack of agreement on a solution to intrastate ethnic conflict.
Respondents were just about evenly divided on the merits of the most commonly proposed
remedy, Lijphart’s consociationalism (29% agreeing and 36% disagreeing). Though it was not
asked, it is unlikely that a statement of Horowitz’s integrative approach, the main alternative,
would elicit greater support given that it is less well-known. A final question in this section asked
whether Olson’s collective action problem is a ‘central impediment to the formation of political
groups’. A large majority of 72.5% agreed here, suggesting that some aspects of the rational choice
approach are part of the standard toolkit.
To summarize, there is some agreement on the political economy of advanced industrialized
societies, but much less on the roots of development. As far as conflict and groups are concerned,
there is more agreement on causes than on consequences or solutions.
Institutions
Some comparativists have suggested that the study of formal political institutions is the area of
comparative politics that most approximates a normal science. Scholars have developed a number
of well-known hypotheses, grounded them in formal theory, and subjected them to multiple tests.
This can be seen in the very strong degree of consensus on several institutional propositions as
Figure 5 shows. However, this consensus mostly applies to fairly focused propositions and does
not extend to larger institutional constellations or to developing countries.
The very high degree of consensus can be seen for veto points theory, the causes of the number
of parties, and the median voter theorem. In the first two cases, over 80% agreed that the more veto
points, the less likely is change from the status quo and that electoral rules and social diversity are
the ‘main causes’ of the number of relevant parties in a political system. Support for the median
voter theorem was not quite so high (63% versus 23%), but was still substantial. The only other
institutional proposition that received nearly this level of support was the idea that bureaucratic
agencies tend to be captured by the industries they are supposed to regulate (69% versus 4%), a
surprising triumph for public choice theory.
Most other institutional propositions received only ambivalent support. Linz’s perils of presidentialism
met with more disagreement than agreement (28% versus 41%). The same applied
to Michel’s iron law of oligarchy as applied to political parties (37% versus 28%), the instability
of multiparty systems (30% versus 44%), and the superiority of consensus institutions to
12	 International Political Science Review 00(0)
majoritarian ones (41% versus 25%). Few were willing to even hazard an opinion on the proposition
that federalism encourages free markets – 52% chose neither agree nor disagree and 20%
chose no opinion.
In short, institutional findings were divided between propositions that received some of the
highest levels of support in the survey and those with strong splits.Again, one sees more agreement
on propositions that are more applicable to advanced industrial societies (electoral laws, veto
points) and less to those equally or more applicable to developing nations (presidentialism, broader
constitutional choices).
Subgroup differences
The previous analyses have been of comparativists as a whole. Another question is whether there
are differences across subgroups. In particular, I focused on generational and gender differences.
To assess these differences, I conducted difference of means tests between different groups (full
results are available in the online appendix).
Generational differences
Conventional wisdom would predict that the greatest differences are generational. The practice of
comparative politics has changed significantly in the past half century. Training and practice have
become more quantitative and focused on methodology. Younger comparativists might also be
Figure 5. Institutions.
15.8
56.6
22.8
4.3 .5
9.3
32.1 33.4
20.6
4.6 4.5
17.9
36.6
29.9
11
37.9
48
8.3 4.8
1 1.2
13.1
52.5
23.7
9.5
3.5
33.2 35
23.5
4.9
11.5
51.3
13.9
19.6
3.7 3.5
26.2 26.4
34
9.8
1.2
5.4 7.3
34.9
51.2
4.3
24.1
30.2 32
9.4
31.3
57.1
7.9
2.7 1
5.7
37
21.7
26.2
9.4
0
20
40
60
0
20
40
60
0
20
40
60
0
20
40
60
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Agree Disagree Agree Disagree Agree Disagree
Bureaucracies tend to be captured Consensus better than majoritarian Corporatism delivers better economy
Electoral rules cause number of parties Federalism encourages free markets Iron law of oligarchy
Median voter theorem Multiparty systems less stable Partisanship mainly American
Perils of presidentialism Veto players reduce change War makes state and state makes war
Note: The bars represent the percentage of respondents who chose options on a five-point scale ranging from strongly
agree to strongly disagree. ‘Don’t know’ responses were excluded from the calculation of percentages.
Roberts	 13
attracted to more recent theoretical advances and less in thrall to trends that were more popular in
the past (for example, political cultural analysis or modernization theory). To assess the extent of
this change, I compared respondents who received their PhDs after 2000 and those who received
their PhDs in the 1980s or before (I left out those who received their PhDs in the 1990s as being
trained in a transitional era).
In a few cases, these predictions were fulfilled, but overall older comparativists were not so different
from younger ones. Statistically and substantively significant differences between the two
groups showed up on only about one-third of statements (21/61).
Even more surprisingly, these differences were not particularly concentrated on issues of methodology
or approach (only 6 of 25 statements in these sections showed large differences). The only
foundational issue on which the older generation differed from the younger was whether the field
should aspire to be a science where older comparativists were less in agreement. They were also
more likely to see the field as divided (perhaps from having lived through past battles). More in
line with expectations, older comparativists were more likely to see culture and thick description
as important and less likely to embrace experimental methods.
On substantive issues, older comparativists were more likely to accept propositions that were in
vogue in the past, for example, modernization theory, the benefits of external intervention, the
negative effects of Islam on democracy, and the usefulness of consociationalism. Interestingly,
then, the older generation, or at least those responding to the survey, has kept pace with the changing
methodological approach of comparative politics but has remained more attached to some
older substantive findings.
Gender differences
Surprisingly, gender differences were considerably larger and touched more fault lines in the field.
Nearly one-half of the goals and approach and methodology statements produced significant differences
and in some cases they were quite large (over half a point on the five-point scale). These
very large differences included the aspiration to be a science and the necessity of quantification and
EITM where female comparativists were less in agreement. Women were also less accepting than
men of the goal of developing broad, general propositions, the usefulness of the rationality assumption,
the counterfactual approach to causality, the utility of increasing the N, and the need for variation
on the dependent variable. Fully 12 of the 25 statements on these themes showed significant
differences between men and women. This supports the claim in Spierings’(2016) symposium that
the feminist academic movement tended to privilege interpretive and qualitative research over the
dominant positivist and quantitative approaches.
These differences extended to theories of democracy where women were less in agreement
with modernization theory (in both forms), with minimalist definitions of democracy, with
Huntington’s order before democracy thesis, with the beneficial effects of external intervention,
and with the negative impact of Islam. Here 8 of 13 statements showed differences. These differences
were less pronounced for the economy and conflict – only 3 of 12 statements – and
institutions – only 5 of 12. In the latter areas, the major areas of disagreement were ethnic conflict
(women were more likely than men to reject the clash of civilizations and to see ethnic
conflict as instrumental) and the usefulness of consensus democracy and corporatism (women
were more likely than men to agree).
The size and frequency of these gender differences were somewhat surprising and certainly
deserve further research especially considering recent findings of large biases against women in
citation patterns (Maliniak et al., 2013). Simmons (2013) has suggested that the lower rate of citations
for women may be due to the different approaches that male and female scholars are likely to
take and the fact that men (and by extension their approaches) are still a majority in the field.
14	 International Political Science Review 00(0)
Conclusion
Do comparativists agree on the goals, methods, and major results of the field? To answer this question,
I placed each item into one of three categories: strong consensus where more than two-thirds
of respondents with an opinion were on one side or the other; moderate consensus where one-half
to two-thirds were on one side or the other; and dissensus where neither the agree or disagree side
reached 50%. Full results are available in an online appendix.
The results are ambiguous. More than half of the propositions show strong and moderate consensus
(18 and 17 propositions respectively) and something less than half (26 propositions) show
dissensus. More telling is where the agreements and disagreements lie. Contrary to the conventional
wisdom that comparativists are divided by goals and methods, these areas show the highest
level of consensus. Disagreement is more common on substantive results.
Specifically, large majorities of comparativists agree that the aim of the field is to develop
broad, general propositions framed in causal terms and that the field has made significant progress
in recent years. They also largely agree that the field should have scientific aspirations. At the same
time, commitment to a scientific comparative politics coexists with a strong belief that deep knowledge
of an area is important for carrying out these goals.
When it comes to putting these goals into practice, this degree of consensus moderates, but
does not disappear. What we see instead is eclecticism – most comparativists see the value in
multiple approaches – from rational choice to thick description, from office-seeking to culture.
Only where a method was presented as the ‘one true way’was there pushback. Nevertheless, these
different approaches would need to conform to certain methodological precepts like variation on
the dependent variable and increasing the N. This is where the scientific orientation affects the
diverse approaches.
Turning to actual empirical results, the degree of consensus drops considerably. It is hard to say
that we have achieved clear and accepted knowledge in many areas of comparative politics. Of
course, this comes with several caveats. In the first place, the survey included only a limited selection
of possible findings and most were phrased in relatively general and abstract terms in order to
be understood by respondents. Connected to this, many comparativists may be unfamiliar with the
cutting edge of research in these areas. Few respondents chose the ‘don’t know’ option and so
many may have been relying on outdated knowledge. Factor analysis of the results for each substantive
area revealed only a single dominant factor for each, which possibly suggests a divide
between those agreeing and disagreeing that comparative politics has created knowledge.
To put a positive spin on the results, one could say that there is very strong agreement on a
number of foundational issues: the connection between economic development and democracy, the
importance of collective action problems, the existence of varieties of capitalism, veto players
theory, bureaucratic capture, the effect of electoral laws, and the non-existence of a clash of civilizations.
Comparativists do know something or at least they agree on some things.
However, there is a larger class of issues on which there is considerable dissensus. The relation
between politics and development is one of these. The effects of political institutions (aside from
veto players and electoral laws) is another. The field seems particularly torn on issues with important
policy consequences like the best solutions for conflict and economic development as well as
the best (or worst) constitutional forms and the origins of states.
As suggested above, one interpretation of this dissensus is that politics works in different ways
in different places. The subfield of comparative politics has traditionally been divided across
regional lines as well as between developed and developing nations. It may be that theories do not
cross these lines very well (and the survey did not usually mention the proper domain). Knowledge
in comparative politics may be specific to regions, countries, and even smaller units. Interestingly,
Roberts	 15
there appeared to be more consensus on the politics of the advanced industrial societies. Possibly
this has to do with more resources being devoted to their study or to a Tolstoyan regularity that
happy families are alike, but unhappy families are unhappy in distinctive ways.
A number of interesting findings emerged from the analysis of group differences. Surprisingly,
generational differences were not particularly pronounced while gender differences were. This
belies standard views of the field.
So where does comparative politics stand? Despite claims of division, it does seem to have a
reasonable foundation on which to build research. There is considerable consensus among
US-based comparativists on what they are trying to do, the usefulness of particular assumptions
and techniques, as well as a handful of methodological precepts. This bodes well for collegiality
and collaboration in the field. To a good degree comparativists are on the same page.
However, this leads to two surprises. One is from within the survey. A majority of comparativists
still believe that the field is divided. This might be explained as a legacy of past battles or perhaps the
narcissism of minor differences. A second is the lack of consensus on key findings. If the field’s
approach is right, why has it not yielded more findings that command general agreement? Perhaps
because this approach is of relatively recent vintage. Though we have little evidence of what comparativists
believed in the past, the consensus on causal inference is unlikely to have prevailed as recently
as the 1990s. It may be a matter of time before better findings emerge. Indeed, economics is currently
undergoing whatAngrist and Pischke call a ‘credibility revolution’that discredits much previous work.
The survey questions may also have been at fault. They were pitched at a level of generality that does
not correspond with the sort of qualified findings that typically emerge from research.
Where should we go from here? As mentioned above, the survey questions often suffered from
problems like abstractness and non-comparability. The focus on US-based comparativists limits
the generalizability of the results. These problems could be remedied in future iterations.
Replications might also focus on specialists in narrowly defined subfields. If political science is to
defend its bona fides and have an influence on policy, then it needs to pay greater attention to the
knowledge it has produced.
Acknowledgements
The author is grateful to Jordan Gans-Morse, Jim Mahoney, Jason Seawright, Josh Tucker, and IPSR’s
reviewers for their valuable comments.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
ORCID iD
Andrew Roberts   https://orcid.org/0000-0002-0416-5975
Supplementary material
Supplementary material is available for this article online on the journal website.
Notes
 1.	 Looking at the main general-interest journals (APSR, AJPS, and JOP), Mahoney (2007) found more
evidence of deductive and quantitative work.
 2.	 Most surveys of economists focus on their agreement with the benefits of specific policy choices—for
example, the minimum wage, trade restrictions, or budget deficits. However comparative politics has developed
fewer theories about the value of different public policies, making this approach harder to replicate.
16	 International Political Science Review 00(0)
 3.	 Other surveys that include comparativists are Tetlock (2005) and the Bright Line Watch (brightlinewatch.org),
but they do not separate the opinions of comparativists.
  4.	 In general, few respondents chose the no opinion option—typically less than 5%.
  5.	 Although these ratings are obviously flawed, they should capture a broad swathe of the profession.
  6.	 Only those listed as regular faculty members on the website of the political science department were
included. I did not include faculty at public policy, business, or law schools or other institutes unless they
were listed as members of the political science department. I excluded my home institution because of
frequent testing on my colleagues and worries about anonymity.
  7.	 The main reason is that research universities tend to have much larger faculties.
  8.	 This should not be taken as a complete set of approaches given that it omits class analysis, identity politics,
and critical approaches, among others.
  9.	 Path dependent processes might be placed on the latter side as well. They make it difficult to develop
broad generalizations since contingent historical factors tend to set off these processes.
10.	 The survey did not explicitly address methods associated with the recent ‘credibility revolution’ like
randomized control trials and regression discontinuity.
11.	 KKV has come under considerable criticism, particularly in the volume Redesigning Social Inquiry
(Brady and Collier, 2004). My initial aim was to include several propositions from RSI, but the advice
of the first edition was difficult to summarize.
12.	 Factor analysis of these results again revealed only a single factor solution.
13.	 Perhaps more nuanced statements of these options would have yielded better results—for example,
Evans’s theory of embedded autonomy. But here, as elsewhere, the more complexities that are introduced,
the greater the likelihood that respondents will not be familiar with the finding and the harder it is
to gauge the aspect of the theory they are agreeing or disagreeing with. Again, future surveys might ask
about the relative effects of a range of factors.
14.	 The proposition that corporatism delivers better economic performance than pluralist or liberal forms
of interest representation was included in the institutions section of the survey, but may belong here. It
received only ambivalent support, possibly on its own merits, but also potentially because corporatism
may not work in developing economies. The proposition did not specify the domain of the argument.
References
Almond, Gabriel (1988) Separate Tables: Schools and sects in political science. PS: Political Science and
Politics 21(4): 828–842.
Bates, Robert H (1997) Area Studies and the Discipline: A useful controversy. PS: Political Science and
Politics 30(2): 166–169.
Brady, Henry E and David Collier (eds) (2004) Rethinking Social Inquiry: Diverse Tools, Shared Standards.
Lanham, MD: Rowman & Littlefield, Berkeley Public Policy.
Frey, Bruno S, Werner W Pommerehne, Friedrich Schneider and Guy Gilbert (1984) Consensus and
Dissension Among Economists: An empirical inquiry. American Economic Review 74(5): 986–994.
Friedman, Jeffrey (ed) (1995) The Rational Choice Controversy: Economic Models of Politics Reconsidered.
New Haven: Yale University Press.
Fuller, Dan and Doris Geide-Stevenson (2003) Consensus Among Economists: Revisited. Journal of
Economic Education 34(4): 369–387.
Geddes, Barbara (2003) Paradigms and Sand Castles: Theory Building and Research Design in Comparative
Politics. Ann Arbor: University of Michigan Press.
Goertz, Gary and James Mahoney (2012) A Tale of Two Cultures: Quantitative and Qualitative Research in
the Social Sciences. Princeton, NJ: Princeton University Press.
Granato, Jim and Frank Scioli (2004) Puzzles, Proverbs, and Omega Matrices: The scientific and social
significance of the empirical implications of theoretical models (EITM). Perspectives on Politics 2(2):
301–312.
Green, Donald and Ian Shapiro (1994) Pathologies of Rational Choice: A Critique of Applications in Political
Science. New Haven, CT: Yale University Press.
Roberts	 17
Hull, Adrian Prentice (1999) Comparative Political Science: An inventory and assessment since the 1980s.
PS: Political Science and Politics 32(1): 117–124.
Johnson, Chalmers (1997) Perception versus Observation, or the Contributions of Rational Choice Theory
and Area Studies to Political Science. PS: Political Science and Politics 30(2): 170–174.
King, Gary, Robert O Keohane and Sidney Verba (1992) Designing Social Inquiry: Scientific Inference in
Qualitative Research. Princeton, NJ: Princeton University Press.
Mahoney, James (2007) Debating the State of Comparative Politics: Views from qualitative research.
Comparative Political Studies 40(1): 32–38.
Maliniak, Daniel, Ryan Powers and Barbara F Walters (2013) The Gender Citations Gap in International
Relations. International Organization 67(4): 889–922.
Maliniak, Daniel, Amy Oakes, Susan Peterson and Michael J Tierney (2011) International Relations in the US
Academy. International Studies Quarterly 55(2): 437-464.
Mead, Lawrence M (2009) Scholasticism in Political Science. Perspectives on Politics 8(2): 453–464.
Munck, Gerardo L and Richard Snyder (2007) Debating the Direction of Comparative Politics: An analysis
of leading journals. Comparative Political Studies 40(1): 5–31.
Sigelman, Lee and GH Gadbois (1983) Contemporary Comparative Politics: An inventory and assessment.
Comparative Political Studies 16(3): 275–305.
Simmons, Beth (2013) The Gender Citations Gap: A glass half full perspective. The Washington Post, October
4. Available at: http://www.washingtonpost.com/blogs/monkey-cage/wp/2013/10/04/the-gender-citations
-gap-a-glass-half-full-perspective/.
Spierings, Niels (2016) Introduction: Gender and comparative methods. Politics and Gender 12(6): 1–5.
Tetlock, Philip (2005) Expert Political Judgment: How Good Is It? How Can We Know? Princeton, NJ:
Princeton University Press.
Author biography
Andrew Roberts is an associate professor of Political Science at Northwestern University. His research interests
are in the quality of democracy, political institutions, political economy, and post-communist Europe. He
is the author of The Quality of Democracy in Eastern Europe: Public Preferences and Policy Reforms
(Cambridge University Press, 2009) and articles in such journals as the British Journal of Political Science,
Comparative Political Studies, Legislative Studies Quarterly, and Party Politics.