Misinformed? The Implications of Measurement for Assessing Citizen Competence

Abstract

This dissertation addresses measurement issues underlying our understanding of the capacity of citizens to hold elected representatives accountable. A growing body of work documents the many misperceptions---or inaccurate beliefs---people hold about politics, and the prevailing narrative in the literature suggests that the public is highly misinformed. In Chapter 2, I argue that past work exaggerates the degree to which the public is misinformed by relying on methods that cannot meaningfully distinguish between inaccurate beliefs held with certainty and incorrect guesses. I adjust for differential item functioning (i.e., differences in how respondents use response scales) in an alternative measure of certainty and show that being misinformed is surprisingly rare. I find that misperceptions reported on surveys are four times more likely to represent uncertain guesses than firmly held beliefs. Chapter 3 examines how guessing under uncertainty gives the false impression of being misinformed. I show that when people are uncertain, they engage in a Bayesian process by which they systematically overestimate the size of smaller quantities and underestimate the size of larger ones. I show that this Bayesian model compares favorably to existing explanations of misperceptions about the size of minority racial groups using a large collection of data from estimates of the size of 42 demographic groups from 35,000 survey respondents in 22 countries. The findings presented in Chapter 3 raise questions about the common practice of measuring political misperceptions by asking survey respondents to estimate the size of politically relevant quantities. Chapter 4 examines how people interact with theses quantities, specifically with respect to attempts to change attitudes by correcting misperceptions. I focus on individual differences in numeracy (i.e., quantitative literacy) and show that people often lack the necessary skills required to estimate the size of political quantities and incorporate correct information into their attitudes. Together, these findings challenge the prevailing narrative that much of the public is highly misinformed about politics and suggests that much of what is known about the capacity of citizens to meaningfully engage in politics is a product of the types of questions used to measure political misperceptions.