Knowledge in Political Science

What do we know about politics?

I worry a lot about whether my chosen field of political science has actually produced real knowledge. While natural scientists can point to many well-confirmed theories, it is harder to think of similar achievements in political science.

To potentially bolster my self-esteem, I went searching for things political scientists know. I prepared a list of prominent theories in my subfield of comparative politics and did the very political-sciency thing of surveying my fellow political scientists on whether they agreed or disagreed with these theories. My thinking was that if we truly know something, then there should be a relative consensus among political scientists that this theory was true.

You can read the full results here (or see the graphs at the end of this post), but the short version is that there were not many theories on which political scientists agreed. From about 40 propositions, the ones where there was a strong consensus (ie, more than two-thirds of respondents) agreeing with a theory, looks like this:

  • The strength of civil society is a major determinant of how well democracy functions

  • Transitions to democracy are more likely to survive in wealthier countries.

  • Advanced industrial economies exhibit institutional complementarities that can be described in terms of distinct varieties of capitalism

  • The collective action problem as described by Mancur Olson is a central impediment to the formation of political groups.

  • The more veto players in a political system, the less likely is change from the policy status quo.

  • Electoral rules and social diversity are the main causes of the number of relevant parties in a political system.

  • Bureaucratic agencies tend to be captured by the industries they are supposed to regulate.

There was one other proposition where there was a consensus that a theory was wrong, namely:

  • Recent global politics can be characterized by a 'clash of civilizations'.

By contrast, consider the list of topics where there was a fair degree of dissensus;

  • Whether the economy determines elections

  • Whether democracy can be created in any country

  • Whether democracy promotes development

  • Whether order should precede democracy

  • Whether public policy follows public opinion

  • Whether there is a natural resource curse

  • Whether foreign interventions can create democracy

  • Whether institutions like private property are the best explanation for development

  • Whether ethnic diversity hurts growth  

  • Whether a strong executive helps economic reform  

  • Whether state intervention is key to development  

  • Whether consociationalism is a solution to ethnic conflict

  • Whether multiparty system are less stable  

  • Whether war makes state and state makes war  

  • Whether presidential systems are more likely to break down  

  • Whether consensus institutions are better than majoritarianism

  • Whether corporatism delivers better economic results  

  • Whether the iron law of oligarchy holds in parties

  • Whether federalism encourages free markets

And this is not to mention a number of theories that got over half but less than two-thirds support, what I called a moderate consensus.

In short, this was not an inspiring exercise. There have been some other attempts to codify our knowledge and they similarly do not yield long lists (see Noel’s Ten Things We Know, Roberts’s Ten Things We Know, about Comparative Politics, and Colomer’s Thirty Propositions in Political Science).

Does this mean things are hopeless? I can think of a number of ways to salvage political science, or comparative politics in this case.

In the first place, it was difficult to frame statements about the field in ways that were (i) specific enough to elicit agreement or disagreement but (ii) general enough that most comparativists could assess them. It is possible (likely?) that the statements here were pitched at such a high level of generality that it was easy for respondents to find exceptions. (This might also explain another curious result - the fact that those who specialized in the subfield of the theory did not have a different evaluation from those who specialized in something else.)

More specifically, many (most?) theories in comparative politics are limited in scope or domain. By contrast, most of the statements that I provided did not specify a domain (occasionally democracies or advanced industrial economies). Further, the devil may be in the details. A full specification of Duverger’s law or the median voter theorem, or a collective action problem requires a host of assumptions that were not spelled out. 

Similar polls of economists tend to focus on their agreement with different public policies - say, the negative income tax or tariffs - but political scientists tend to be less concerned with evaluating the welfare effects of policy.

For many political scientists, the best that we can hope for are focused, mid-range generalizations that are sensitive to context. These might be harder to aggregate into general statements about politics, but they would still constitute real knowledge, even though one might worry about how useful they were. While medium-range results may not generalize to policy recommendations sub specie aeternitatis, they could still generate recommendations, say, on how to proceed in a place like Syria.

Another tack would be to focus on descriptive and conceptual knowledge rather than on causal propositions. I did not focus on this sort of knowledge in my survey, but arguably much of our knowledge in political science is descriptive whether in terms of the varieties and distribution of institutional forms or the content and dynamics of public opinion and voting.

A final way to save our reputation is to focus on a number of questions that I posed about the aims and methods of comparative politics. Here there was much more agreement. Comparativists agreed that we should be searching for causal generalizations about politics and about some of the methods that would be useful for getting there. This agreement is arguably fairly recent. When I was in grad school in the 1990s, there were still fierce battles over methods and rigorous qualitative methods barely existed. And since my survey in 2010 there has been much more movement towards strong causal inference. 

Perhaps we are simply at the beginning of the path towards knowledge. Other fields like economics with its credibility revolution and psychology with its new focus on replication seem to have embarked on similar paths at around the same time. The future may be bright.

Agreement of Comparativists with Propositions about Politics

The figures below present the percentage of the surveyed political scientists who agreed with the following propositions about politics. More details are in the paper.

Figure 1: Democracy and Democratic Politics

Figure 2: Economy and Society

Figure 3: Political Institutions