What makes a good comparison?
Comparative politics is about comparing political systems to each other, but how do you decide what countries to look at? This video introduces principles for and limitations of comparing two or more countries to each other that you want to bear in mind when analyzing international politics.
In 1843, philosopher John Stuart Mill wrote A System of Logic, in which he described principles of comparison that would allow you to draw conclusions about cause and effect – we call these “Mill’s Methods.” Today, I am going to talk about one of them, what is referred to a “Most Similar Systems” design.
When we study the politics of a single country, we can get a good sense of what caused what in that case (what causes corruption in Kenya or democratization in Poland, e.g.), but we can’t be sure if what we learn about that country applies to other countries. Are we talking about a general rule or are the circumstances of one country so unique that what happened there isn’t relevant beyond its borders? For example, we have learned that war contributed to state formation in Europe. But can we apply that in other countries?
One way we address this question is by comparing countries to each other. By looking at more than one case, we can generalize beyond a single country.
First, we’ll talk about comparing two countries. The principle behind a Most Similar Systems design is that you choose to look at two countries that are as alike as possible. They should be different in just one way – and that way or variable should be the one you think is the cause of the outcome you are interested in. The idea is that if the outcome is different in those two countries, where everything else is the same, then the factor that is different could be the cause.
To give two examples, I wrote my senior thesis in college on “why was there ethnic conflict in Bosnia, but not Macedonia?” Bosnia-Herzegovina and the country formerly known as the Former Yugoslav Republic of Macedonia, and now known as North Macedonia, are both former parts of communist Yugoslavia so they share similar histories and economies. They both were very ethnically diverse, which was the most popular explanation for the war in the 1990s. But there were different outcomes – civil war and genocide in Bosnia but relative peace in Macedonia. I used the similar histories but different outcomes to highlight what I thought was the cause of war (institutions of course!).
Similarly, Daron Acemoglu and James Robinson contrast North and South Korea. These two countries share a common culture and much of their history, but gained different political institutions during the Korean War. They argue that these different institutions are the cause of the different economic outcomes in each country.
To make comparisons like these, you need to think about the following questions:
What do you think are the cause and effect in your case?
What else could cause that effect?
This is where academic research can help you – they explain all the other potential reasons – alternate explanations – for an outcome.
Can I find another case that is similar on the alternate explanations to my country, but differs in the way I think is the cause?
Some things you want to think about when choosing cases include:
Choosing a country in the same region or a different region.
Countries in the same area often share institutions and history (like North and South Korea or former Yugoslav republics, or more generally, former Communist states or former French colonies)
But sometimes you want to contrast regions to each other, or demonstrate that an effect is not dependent on what part of the world you are talking about. In those cases, you might want to look at countries from more than one part of the world.
What is important is that you are clear about what you are holding constant and what you are varying across the cases.
Even comparing two countries can leave you open to the problem of selection bias, though. This is the problem where, as Barbara Geddes in her book Paradigms and Sand Castles calls it, “the cases that you choose determine the answers that you get.” These two graphs demonstrate the problem. If you look at the countries in East and Southeast Asia in the 1970s, you might draw the conclusions that high levels of economic growth are associated with high degrees of labor repression – that you need to limit workers ability to form unions in order to boost economic productivity. If you expand your analysis to all developing countries, however, the you don’t see the same relationship – there is no correlation between growth rates and labor repression.
Case studies and paired comparisons are still valuable because they provide you a lot of detail on cause and effect, but the desire to avoid drawing wrong conclusions from a small group of cases is the reason why we also do statistical analysis, which allows us to compare a larger number of countries to each other.