Democracy and Struggle for Definition

Common definitions happen to be useful in research. Instead of arguing what “democracy” means, economists agree on the same definition and move on to important things, for example, the relationships between democracy and economic growth:

Acemoglu et al., "Democracy Does Cause Growth."
Acemoglu et al., “Democracy Does Cause Growth.”

So, you study relationships between specific “democracy” (usually Polity IV) and specific “growth” (real GDP per capita). No problem in studying another democracy and another growth. But until economists conquer the world and turn everyone into an economist, very few heads are working on the issue right now. And these heads have to focus on very specific definitions, like the Polity IV components of “democracy”:

Competitiveness of Executive Recruitment
Openness of Executive Recruitment
Constraint on Chief Executive
Competitiveness of Political Participation

This makes communication with the public difficult, though. For one reason, the public understands “democracy” differently. The great source for making a representative public definition of “democracy” is the World Values Survey (WVS). The survey reaches thousands of respondents in 52 countries, and since 1995, asks questions about democracy. Specifically, it asks to “tell me for each of the following things how essential you think it is as a characteristic of democracy,” and offers nine scales. Summary stats of these scales (min 1 — not an essential characteristic; max 10 — an essential characteristic):

Definition of democracy given by the WVS respondents
Definition of democracy given by the WVS respondents

The first row is the answer to the question how good democracy is for the respondent’s political system (min 1 — “Very good”; max 4 — “Very bad”). The rest of the variables are the components of democracy as defined by 74,000 people from 52 countries. You can think of the respective means (third column) as weights each variable has in the public definition of democracy. The standard deviations aren’t huge, which implies some consensus across many people.

These components seemingly have little in common with the Polity IV’s definition. To make sure, let’s compare the Polity IV index of political regimes with question V141 from the WVS, which asks “how democratically is this country being governed today.” A scale ‘from 1 to 10, where 1 means that it is “not at all democratic” and 10 means that it is “completely democratic.”’ Hence, I compare two things: the respondent’s opinion about the state of democracy (as she understands the concept) in her country and Polity IV’s expert’s opinion about the state of democracy (as defined by Polity IV) in the same country.

Democracies described by the public in the WVS and by experts in the Polity IV dataset have almost nothing in common:


Though the regression line is not horizontal and the relationship is significant at 0.01, R² is just 14%. Not what you may expect from two things with the same name.

This nerdy fact is important because results in the social sciences are relevant only for concepts as they’ve been defined in the social sciences. If your definition X is different from my definition X, then my investigations of X are useless to you. If someone thinks that “democracy” is when “the army takes over when government is incompetent” (row six in the table) then this democracy does not cause economic growth that the first figure shows. In fact, the authors of the figure describe in detail how they constructed the measure of democracy that is relevant for growth.

Of course, when it comes to elections and policy making, economists become as humble as dentists. That is, not even dentists, but also physicians know how to fix the economy—contrary to what economists think on the same matter. In addition, there’s no organized communication between people and researchers. When the public is so poorly informed about actual research, the difference in definitions—of which “democracy” is just one example—seem unimportant. It is, to an extent. But whenever serious research makes it into the media, it’s better to doublecheck the words.

Doing Harm by Way of Habit

Facebook’s experiment on emotions got more feedback than any academic research had before. Many quitted, some raged. The reaction concerned, first, experimenting, second, manipulating.

Facebook experimented with News Feed by reducing or increasing posts containing positive and negative emotions. Then it measured the users’ reaction, which happened to be small but statistically significant, mostly because of the huge sample.

So, Facebook once tried what reputable The New York Times and Washington Post do every day (not to mention TV, penny press, and advertising industry). In fact, traditional media scrutinized the study and maybe raised more emotions than the original researchers did in the experiment.

Facebook surely learned the lesson and won’t publish significant research readily. Just like any other relatively open private company, including Google and Microsoft. Going stealth is safer for employees, which also means less collaboration with outsiders from universities and less openness to the public in general.

If doing “experiments” is punishable, then it’s better to leave everything as is. But that does most harm.

Experiments test hypotheses. When you have no experiments, you have hypotheses alone, true and (more often) false. Meanwhile, decisions are still being made. In Facebook, in GE, or in the government. Policymakers have to. They have many hypotheses for that, even if they never mention the word.

If you’re a CEO of a large corporation, you have hypotheses about your employees. You may think that praise works better than salary, and praise more. Or increase the payroll, or threaten the employees. But without experiments you can’t know if you’re right. Having positive experience indicates a little, because anyone can be right by chance. Even coin flipping gives a 50% chance. Without systematic evidences, you end up doing Machiavellian stuff and hurting people who trust you.

Current practices do much harm because they came out of all the crazy theories authors had. Governments have their own, corporations do. Experiments in the present help find better solutions. And all attention to the Facebook study, Ebola, and Michael Jackson’s death you could devote to questioning what happens every day. It makes the difference.

How to Define a Word

Saul Kripke needed one hundred pages to show some difficulties with using the word “Aristotle” in our language.

Politicians had similar problems with common nouns. When Jefferson mentioned “All men are created equal,” he didn’t mention that “men” meant white men with a fortune.

“Democracy” eclipses other words by the number of alternative meanings. Churchill and Lenin imagined totally different things when they used it. And so did thousands of other authors. Good luck finding two authors who meant the same by the word.

But economists have to put democracy in equations. Words fit equations badly, and you need a number. A number for “democracy” usually comes from Polity IV, World Bank, Freedom House, and other organizations. The numbers have been even brought together in a single dataset.

And yes, these numbers still mean different things because organizations compose them from different facts. Here’re Polity IV’s democracy components:

Competitiveness of Executive Recruitment
Openness of Executive Recruitment
Constraint on Chief Executive
Competitiveness of Political Participation

Still, an indicator consists of facts: whether elections are competitive, presidents have to ask permissions, or country has elections at all. This indicator can jump into equations and help understand relations with other facts, like wages. In the end, you may be interested in the outcomes of democracy, like better living. Indicators happen to be very specific about this: if you had this-and-that variable equal one, then your income would be 20% higher.

A huge improvement over Churchill, who said basically the same thing, is that numbers can prove it. As far as democracy means a fixed list of facts, these facts can be systematically studied. On the other hand, storytelling about democracy proves nothing, it only convinces. People got convinced by various lies for centuries. That’s what history is about. And that’s also why you need a mechanism called science to check what happened to be true after all.