In a functioning representative democracy, citizens pressure politicians to act in accordance with public opinion. However, politicians need to be able to read public opinion correctly to serve as good delegates of the public will. In their new book, Politicians’ Reading of Public Opinion and Its Biases, Stefaan Walgrave, Karolin Soontjens, and Julie Sevenans argue that politicians are not good at reading public opinion, and that this undermines democratic representation. The book makes an important contribution to the analysis of mass-elite interaction in modern democracy and specifically the bottom-up link of democratic representation. Existing research focuses on how citizens express their views and how politicians behave, but there is very little research about how politicians perceive citizens’ signals. Politicians’ Reading of Public Opinion and Its Biases addresses this neglected question with an elegant argument and an insightful analysis of Flanders (the Dutch-speaking region of Belgium) using rich and systematically aligned evidence from public opinion surveys as well as a survey and in-depth interviews with Flemish politicians.

The book suggests that politicians care a lot about public opinion, and that their actions follow public opinion. However, despite all their efforts to understand what the people want and their honest desire to act in accordance with the public will, politicians are responsible for a “drama of representation” because their judgments of what the people want are inaccurate. The book reports a survey asking participants to express their opinions about eight important policies, and goes on to show that Flemish politicians, in the aggregate, misjudge public support for these policies by 13 percentage points. Importantly, politicians consistently locate both the general electorate and their own voters further to the right than they really are. When adding up individual errors in estimating public opinion, politicians feature an average inaccuracy of 18 percentage points, and they incorrectly guess the direction of majority opinion (whether people are for or against the policy on average) 30% of the time.

The book reports considerable variation between politicians in their ability to estimate correctly, and then goes through great lengths to show very convincingly that this variation cannot easily be attributed to differences in politicians’ characteristics. Specifically, the accuracy with which politicians make good or bad estimates does not depend on their political roles, their estimation ability, their expertise, their seniority, their level of engagement with public opinion, or their investment in their local constituency. By contrast, the direction of inaccurate estimations in the aggregate can be explained. First, the book draws on its surveys of citizens and politicians to show that politicians engage in wishful thinking by overestimating the level of public support for policies they like. And second, the book uses evidence from in-depth interviews with politicians to support its argument that the aggregate right-wing bias of public opinion estimates is a function of a right-wing bias of the information environment in Belgium.

Politicians’ Reading of Public Opinion and Its Biases starts with a brief introduction that describes the research question, the research design, the evidence, and the structure of the book. After that, Chapter 1 spells out the function of politicians’ reading of public opinion in a general model of democratic representation. The following Part I of the book comprises four chapters that rely on the surveys and interviews with politicians to show how politicians read public opinion and how public opinion affects their behavior. Part II of the book analyzes the accuracy of politicians’ perceptions based on data about politicians and surveys of public opinion in four additional chapters. Finally, the brief conclusion summarizes the argument, evaluates the generalizability of the Flemish case, and discusses the negative consequences of politicians’ inability to correctly estimate public opinion for representative democracy.

The best books get us to think about what else we could do to study the questions they raise, and as a great book, Politicians’ Reading of Public Opinion and Its Biases inspires lots of ideas like that. So, once the authors start working on a follow-up study (or the second edition), they (or others building on their many contributions) might want to consider the following suggestions. First, the book could make better use of the evidence generated by the in-depth interviews and connect it more extensively to different theoretical considerations about politicians’ motives and behavior. For instance, the authors use the interview with FJ, a senior politician and former minister (pp. 29-31), to prove their general point about the ambivalence of politicians vis-à-vis public opinion. They show how FJ asserts initially that public opinion is irrelevant, only to claim a few minutes later with the same level of conviction that it is highly significant. This is consistent with the authors’ expectation of finding ambivalence, and the authors make that point very convincingly, but what else can we learn? Why did the politician change his view during the interview? Was there something in the conversation that prompted it? Are there alternative explanations for this course of events during the interview?

Second, at the individual level, politicians’ average accuracy scores deviate between 12 and 24 percentage points from the average of public opinion for different policies, and the overall average for all eight policies is 18 percentage points. There are some worryingly large deviations on the upper side of the distribution around that mean, but is it really such a terrible judgment when, say, 61% of people like a policy, and politicians think it is 79%? Relatedly, is it really a bad estimate when politicians misjudge the direction of public opinion (for or against a policy) 30% of the time? The authors address the problem of finding a reasonable benchmark for the quality of estimates, and it is a neat idea to compare the accuracy of politicians to the accuracy of citizens. However, this point would deserve a more extensive discussion (including an appraisal of research about expert judgments, which a sizable literature finds to be worse than the judgments of non-experts). This is important, because the key claim of the book that we are facing a “drama of representation” requires the observed inaccuracies to be substantial and meaningful.

Third, the normative argument that representative democracy suffers from the inaccuracy of politicians’ estimates of public opinion also depends on a valid measurement of public opinion. The authors carefully and convincingly address problems of survey design and question wording that are inevitable in any study of public opinion. However, there is an additional issue here that might deserve more attention, and this is the question of whether policy views really represent the kind of public opinion politicians are able to perceive. Is asking about views of specific policy proposals too demanding and an overly idealistic expectation of accuracy that exaggerates the size of the “drama of representation” highlighted in the book? The authors address the possibility that politicians might be better at other kinds of estimation tasks. Future studies could shed light on whether politicians would be better at judging the general policy mood or the abstract ideological leaning of their electorate than they are at estimating public support for specific policy proposals.