With Who Governs? James Druckman and Lawrence Jacobs will change the debate about the relationship between public opinion and what elites do. The debate itself is not new. While studies in political science have long investigated how much voters’ preferences shape politicians’ actions,^{Footnote 1} in recent decades research has started to consider the possibility that politicians might be shaping public opinion too.^{Footnote 2} In Who Governs? Druckman and Jacobs show that presidents have created an apparatus to help them follow public opinion and then potentially insulate themselves from it. They have shown that politicians have the tools necessary to shape opinion and not simply respond to it.

Druckman and Jacobs are able to document the extent to which presidents have created a public opinion-shaping apparatus, having compiled an amazing dataset using the papers of three U.S. presidents. From these new documents, we learn that presidents are extensively polling on public opinion. Presidents are also using those polls to learn about how voters respond to different frames. Druckman and Jacobs show that not only are presidents gathering poll data, but this information shapes what issues presidents talk about. While their work focuses on the presidency, it is likely that individual MCs, the major political parties, and other political actors have created similar tools for themselves.

Druckman and Jacobs, by documenting the apparatus that presidents have created, have also laid out an important research agenda for future scholars. While Who Governs? will inspire numerous studies, I want to highlight a few questions that deserve scholars’ attention.

First, I believe that a two-way street connects politicians’ positions and public opinion; each affects the other.^{Footnote 3} What is unclear is how much and under what circumstances. Druckman and Jacobs find that public opinion, at least in some cases, is shaped by presidents’ rhetoric, but we need more work to understand how big the effect is. Is this effect bigger than the effect of public opinion on politicians’ positions? I do not think that we currently have the appropriate research design and measurement approach to document how much each factor influences the other; however, I am more convinced than ever that we should try to measure the relative impact that public opinion and politicians’ positions have on each other.

Second, when can public opinion be shaped? Druckman and Jacobs lay out a convincing argument that presidents can only shape public opinion on some issues. “[W]hen the public holds strong opinions about an issue, it will be difficult to change basic preferences or the importance attached to the policy” (p. 15). Politicians may gather public opinion so that they can learn where the public is persuadable. Research can and should test when politicians are convincing in shaping opinion.

Third, should we be worried that politicians have this apparatus? Druckman and Jacobs raise flags of concern about this development; however, they also acknowledge that there are times when we might want politicians to try to shape public opinion (p. 98). Answering this question requires both normative and empirical work.

One place to start might simply be to ask ourselves how we would like to politicians to act in relation to public opinion. One reasonable answer is that we might want politicians to represent the public on issues that voters feel strongly about, but exercise discretion on other issues. If we wanted politicians to always do what the public wanted on every issue, then we could limit all questions to public referendum. However, there are advantages to delegating to representatives. If the public does not feel strongly about an issue, then we could be better off by trusting a representative who would do the research on our behalf and then make the best decision possible. Of course for a representative to know when they should exercise discretion would require them to collect data on public opinion.

If politicians approached representation in the manner outlined above, we would observe something very close to what Druckman and Jacobs actually find: politicians who poll extensively about public opinion and then deviate from the public’s preferences on issues that the public does not feel strongly about. While I personally want politicians who are responsive to public opinion, I also want politicians who exercise opinion leadership when it is in their citizens’ interests. The difficulty is that one individual’s opinion leadership is another’s opinion manipulation.

At this point I am still more optimistic than Druckman and Jacobs. I think that most politicians are genuinely striving to make decisions that they think are best for their constituents. However, that is an open empirical question and one that we should be pursuing. Druckman and Jacobs lay groundwork for us and so we should take note and build on their work.

While I am more optimistic about representation than Druckman and Jacobs, I think that they actually undersell the potential negative consequences of the development they document for American democracy. Because of the nature of their data, they focus on communication between the president and the public. In practice however, there are competing voices all vying for attention and serving as potential counterweights in the struggle for public opinion. This leaves open the possibility that no single voice will be able to move opinion much. Indeed, one of the solutions raised at the end of the book is that we can minimize undue influence by promoting competition. Such competition will only work if in fact all voices are represented; however, if all of the sides shared the same bias, competition would do nothing to mitigate such a bias. Druckman’s and Jacobs’ results are important because they raise the stakes. Politicians now have tools that allow them to better learn about and shape public opinion. While I am more optimistic that many of them use these tools for good, the fact that they have more powerful tools should cause us to be vigilant in learning and monitoring how those tools are used.