In the last few weeks, several media outlets have reported on a new study on American democracy by Martin Gilens and Benjamin I. Page . The headlines include:
Gilens and Page’s article, forthcoming in Perspectives on Politics this fall, examines which types of actors in American politics—voters, economic elites, and interest groups—have the most influence over policy outcomes. Their findings suggest that economic elites and interest groups have the most influence over policy outcomes, while the average voter has little influence. Based on these findings, Gilens and Page conclude that “America’s claims to being a democratic society are seriously threatened.” Note that Gilens and Page never actually conclude the US is not a democracy, even though the news article titles suggest that they do.
While I sympathetic to the idea that wealthier Americans have more influence over the policy making process than the middle class and the poor, this does not mean the US is not a democracy. I explain why in more detail below.
The consensus among political science research is that the United States is a democracy.
Almost all classification schemes in political science define the United States as a democracy. This is not a minor point. When scientists study a specific research question, they produce several studies in order to assess what they know about how the world works and what they do not know. If they come to a consensus, they have some confidence that they have identified one way in which the world works. In this case, most studies of democracy in political science suggest that the US is a democracy.
We now have a study that may challenge the conventional wisdom that the US is a democracy. While the results of this study are interesting, they should not outweigh the results of hundreds of other studies. If we allowed the results of one study to outweigh the conventional wisdom, it would be like cherry picking evidence to suit our needs.
Nevertheless, Gilens and Page’s study is still important in the sense that it should encourage other political scientists to continue to explore the idea of whether the US and other countries in the world are democracies. Since their results are different from other studies, maybe we can learn something from their study that we have not thought about before.
Gilens and Page define democracy based on policy outcomes, not the political process
It is important to know why Gilens and Page’s study comes to a somewhat different conclusion about American democracy than most other studies in political science. They attempt to determine whether the US is a democracy based on whether policy outcomes in the US represent the interests of the median voter. However, most studies that classify countries as democratic or non-democratic use different criteria to classify countries as democratic and non-democratic.
Most classification schemes of regime type in political science define countries as democratic or non-democratic based on the political process in the country. These classification schemes generally include criteria such as whether there are free, fair, and competitive elections in a country, whether election losers respect the results of election outcomes, whether there are limits on leaders’ power once they are elected, and whether there are regularly held elections resulting in a turnover of power. Under these criteria, the US is clearly a democracy.
Instead of focusing on the political process in the US, Gilens and Page focus on policy outcomes to determine whether the US is democratic. They argue that in order for the US to be a democracy, public policy has to follow the interests of the median voter. What Gilens and Page do not take into account is that democratic political systems are not always designed to represent the interests of the median voter. For example, consensus democracies (such as Belgium and Switzerland) are designed to ensure to several different interests are represented in government. These democracies have multiparty systems and electoral rules that allow several parties to gain access to legislative and executive power. On the other hand, majoritarian democracies (such as the United Kingdom) are designed to ensure that only the interests of the majority are represented in government. These countries tend to have a limited number of political parties and their electoral rules tend to only allow one party to influence the legislative process at any given time. As such, the interests of the majority are well represented in government.
In a sense, both consensus and majoritarian democracies both represent the interests of the median voter, but in different ways. In consensus democracies, government policy tends to reflect the policy interests of the median voter. This is a direct result of how these systems are designed. Since several different parties are represented in government, no one party can control the policy making process. Hence, parties have to work together to develop policy. As a result, policy tends to move naturally towards the preferences of the median party (or interest) in parliament, as the median party tends to be the pivotal actor in determining policy.
On the other hand, in a majoritarian democracy, instead of representing the policy interests of the median voter, government policy reflects the partisan interests of the median voter. When the median voter’s party wins office, the party tends to control the policy making process. The interests of the opposition party tends to be excluded from the process. As a result, policy tends to move towards the median interest of the majority party, as the median interest of the majority party tends to be the pivotal actor in determining policy. Hence, while policy in majoritarian democracies may not reflect the specific policy interests of the median voter, it may reflect his/her partisan interests.
Gilens and Page’s democracy criterion only takes into account that public policy in the US should follow the policy interests of the median voter, not the partisan interests of the median voter. If public policy reflected the partisan interests of the median voter in the US, then we could conclude that in the very least, elected officials are responding to their constituents’ partisan interests. Then we could reasonably conclude that the US meets the criteria of being a majoritarian democracy. If subsequent studies build upon Gilens and Page’s work, those studies should take into account how parties influence policy in the US.
Several other long-standing democracies in the world would not be classified as democratic according to Gilens and Page’s criterion.
I am simply reiterating this point from the previous section to emphasize the implications of using Gilens and Page’s criterion to classify other democracies in the world. According to their criterion, consensus democracies probably would be considered democratic, whereas majoritarian democracies probably would not be considered democratic. In order to resolve this issue, as stated above, I think future studies need to take into account that elected officials may represent the interests of the majority party, not just the median voter.
Gilens and Page make a series of methodological decisions that may limit the conclusions of their study.
According to Gilens and Page’s empirical results, economic elites and interest groups that represent business interests have the most influence over public policy outcomes. In contrast, the median voter and interest groups that represent the public have less influence over policy outcomes. However, the way they identify the median voter and interest groups in their models may limit their ability to conclude if elected officials respond to these different actors.
First, Gilens and Page wanted to know whether the median voter influenced public policy in the US. To identify the median voter, they look at the policy preferences of individuals who have incomes at the 50th income percentile. Traditionally, we define the median voter as the individual with the median policy preference, not the median income. It is unclear whether every person at the 50th income percentile has the same policy preferences. I would be interested to see the extent to which these individuals agree with one another on public policy.
Second, Gilens and Page wanted to determine whether interest groups influenced public policy. However, they only included interest groups in their study that were from a list of the 25 most powerful interest groups in the country. Hence, it is not surprising that they found that these interest groups influenced public policy. It would be interesting to see if their results changed if they include more interest groups in their study.
And last, in their bivariate analyses, Gilens and Page found that the policy preferences of the median voter and the policy preferences of the elites both influence public policy. However, when they ran a multivariate analysis, only the policy preferences of the elites significantly affected policy, not the policy preferences of the median voter. They take this to mean that only elite preferences matter. However, Gilens and Page also recognized that the policy preferences of the median voter in their model and the policy preferences of the economic elites were highly correlated (r=0.73). This high correlation might suggest that it might be difficult for the model to determine whether (1) elite preferences matter, (2) median voter preferences matter, or (3) if they both matter. Highly correlated independent variables tend to produce larger standard errors, making it more difficult to find significant results.