“While it has long been recognized that framing plays a huge role in politics, it is only in recent years that normative democratic theorists have started incorporating this reality. This book will set the standard for this movement and become necessary reading not only for behavioral and philosophical scholars but for all political scientists.”–James Druckman, Northwestern University

“Democratic theorists–and citizens of democracies in general–ignore this book at their own peril. The case for democracy depends upon its tendency to make good decisions. Yet, in this sober, rich, and authoritative study, Kelly shows us how one set of cognitive biases–framing effects–threatens to impede sound democratic decision making.”–Jason Brennan, author of The Ethics of Voting

“Kelly’s synthesis of empirical research on framing effects and normative political theory is impressive, providing the most persuasive integration of the two fields that I have seen. This is a strongly written, clear, and compelling book.”–Paul Brewer, University of Delaware

Jamie Terence Kelly is assistant professor of philosophy at Vassar College.

Framing Democracy

A BEHAVIORAL APPROACH TO DEMOCRATIC THEORYBy Jamie Terence Kelly

PRINCETON UNIVERSITY PRESS

Copyright © 2012 Princeton University Press
All right reserved.
ISBN: 978-0-691-15519-7

Contents

Acknowledgments……………………………………………….ixIntroduction………………………………………………….1Chapter One Framing Effects…………………………………….7Chapter Two Theories of Democracy……………………………….44Chapter Three Behavioral Democratic Theory……………………….59Chapter Four Behavioral Democratic Theory Applied…………………74Chapter Five Institutional Implications………………………….97Conclusion……………………………………………………122References……………………………………………………125Index………………………………………………………..149

Chapter One

Framing Effects

1.1 The Behavioral Approach

Recently, the behavioral approach to law (Sunstein and Thaler 2008), economics (Ariely 2009), and other social sciences (Shleifer 2000; Shefrin 2002) has been gaining popularity. This approach is characterized by an attempt to reform existing disciplines (e.g., law, economics, and finance) through the development of a new model of human decision making (H. Simon 1955; Gintis 2004). Traditionally, these disciplines have employed a model of choice borrowed from classical economics. This model construes individuals as maximally rational and seeks to understand human behavior in terms of a set of optimal rules for the solution of decision-making problems. This approach has been criticized on a number of levels (Jolls, Sunstein, et al. 1998), but most important for my purposes, it has long been shown to present an inaccurate description of actual human decisions (Kahneman, Slovic, et al. 1982; Kahneman 2003).

The behavioral approach to law and economics (as well as other disciplines) arose out of an attempt to develop an account of human decision making that more accurately reflects our actual decision-making behavior. In order to do so, this approach incorporates insights from empirical psychology into theorizing about markets and the law. The behavioral subdisciplines of economics and law reject the traditional rational actor model of human decision making and attempt to generate a new, behavioral model of choice. The picture of human decision making that has emerged from these subdisciplines construes human beings as boundedly rational, where the bounds of our rationality are drawn by cognitive heuristics that, under certain specifiable conditions, result in biases in our decision making. So far, the behavioral approach has proved to be highly productive, generating important insights into how certain facts about the decision making of humans affects our behavior both in markets and in legal settings.

1.1.1 Heuristics and Biases

The behavioral approach to both law and economics has drawn heavily on an empirical literature started by psychologists Daniel Kahneman and Amos Tversky. Since the 1970s, Kahneman and Tversky have studied the effect of risk and uncertainty on human decision making. This psychological research, commonly known as the “heuristics and biases” literature, has become highly influential in economics, law, and political science.

The heuristics and biases literature represents a rejection of the rational actor model of human decision making in that it purports to show that human decisions do not operate on the basis of the rules outlined by rational choice theory. Instead, Kahneman and Tversky have argued that we rely on a small set of relatively efficient, low-information, cognitive shortcuts to solve decision problems. These heuristics do not conform to the requirements of rational choice theory, and thus this literature has helped to explain why, in so many different contexts, actual human decisions fail to be fully rational. In this way, the heuristics and biases literature presents an alternative model of human decision making: the heuristics specified by this literature can be used as a behavioral model for the study of human decisions. The predictions of this behavioral model differ significantly in many (though not all) situations from those of the rational actor model.

In recent years there has been a huge proliferation of research into behavioral models of human decision making. A large number of individual heuristics and biases have been studied, and there is much controversy regarding the proper way to characterize many of these phenomena. 12 As a result, much of the empirical literature on behavioral models of choice is still under development. Because the literature has become so vast, and so much of it remains controversial, it is not yet possible to speak conclusively about a single, unified, and complete behavioral model of choice. Instead, such a model now exists only in bits and pieces, with varying degrees of controversy and empirical disagreement attached to each. As a result of the current instability of the wider behavioral research program, in what follows, I will focus on just one well-documented aspect of the heuristics and biases literature: the phenomenon of framing effects.

1.1.2 Framing Effects

An influential part of the heuristics and biases project involves the study of how experimental subjects respond to decisions involving risk. Kahneman and Tversky have attempted to provide a general account of the heuristics that determine in which situations individuals will display risk-averse and risk-seeking behavior. As part of this project, they conducted experiments to show how the wording of a decision problem served to influence the responses they received from experimental subjects. The result was the first systematic empirical examination of framing effects. The most widely cited example from their research continues to be their “Asian disease problem.”

The Asian disease

Imagine that the United States is preparing for the outbreak of an unusual Asian disease, which is expected to kill 600 people. Two alternative programs to combat the disease have been proposed. Assume that the exact scientific estimates of the consequences of the programs are as follows:

If Program A is adopted, 200 people will be saved

If Program B is adopted, there is a one-third probability that 600 people will be saved and a two-thirds probability that no people will be saved

In this version of the problem, a substantial majority of respondents favor Program A, indicating risk aversion. Other respondents, selected at random, receive a question in which the same cover story is followed by a different description of the options:

If Program A’ is adopted, 400 people will die

If Program B’ is adopted, there is a one-third probability that nobody will die and a two-thirds probability that 600 people will die

A substantial majority of respondents now favor Program B’, the risk-seeking option. Although there is no substantive difference between the versions, they evoke different associations and evaluations. (Kahneman 2003, 1458)

Here, Program A and Program A’ are equivalent (200 people saved; 400 people dead) and Program B and Program B’ are equivalent (one-third probability of 600 people saved; two-thirds probability of 600 dead), while A and A’ (the risk-averse options) are different from B and B’ (the risk- seeking options). The manner in which these outcomes are framed, however, causes respondents to express contrary preferences (i.e., the certain outcome is more popular when the decision is framed in terms of saving people, whereas the risky outcome is more popular when the decision is framed in terms of avoiding death). This serves as evidence for the conclusion that the description of a problem scenario can significantly affect the judgments and preferences of subjects. This is the case despite the fact that the different descriptions are, and can easily be recognized to be, equivalent.

1.1.3 Invariance

Kahneman and Tversky have presented the results of the Asian disease problem and other experiments as a challenge to rational choice theory’s principle of invariance:

Invariance requires that the preference order between prospects should not depend on the manner in which they are described. In particular, two versions of a choice problem that are recognized to be equivalent when shown together should elicit the same preference even when shown separately. We now show that the requirement of invariance, however elementary and innocuous it may seem, cannot generally be satisfied. (Kahneman and Tversky 2000b, 4)

The principle of invariance (referred to by Kenneth Arrow as the principle of extensionality) requires that rational decisions must be invariant across equivalent formulations of the same problem. This principle forms a core part of the rational actor model of human decision making traditionally relied on in economics and related disciplines. It requires that individuals ignore arbitrary changes to the presentation of a choice scenario and focus only on outcomes. If individuals make different decisions when presented with different formulations of the same decision problem, then they fail to satisfy what Arrow called a “fundamental element of rationality, so elementary that we hardly notice it” (Arrow 1982, 6). Kahneman and Tversky, for their part, claim that the “moral of these results is disturbing: Invariance is normatively essential, intuitively compelling, and psychologically unfeasible” (Kahneman and Tversky 2000b, 6).

Abandoning the principle of invariance leads to a model of choice that predicts that individuals will respond in different ways to a given decision problem, depending on how it is framed. It should be emphasized that abandoning the principle of invariance does not entail that human decisions will be chaotic, random, or ultimately unpredictable. Instead, the central insight of Kahneman and Tversky’s work was that framing a risky choice in terms of losses or gains would have a predictable effect on individual decisions (i.e., certain kinds of frames lead individuals to make risk- averse decisions, while other frames lead them to be risk-seeking). Human decision making, despite its quirks, appears to result from the interaction of a number of relatively simple, discernible rules. Although there is still disagreement about the specific nature of these behavioral rules, empirical research has for some time now shown that the framing of decisions reliably affects human decision making. As a result, the empirical study of framing effects has become a mainstay of behavioral research in economics, law, and related social sciences.

1.2 Research on Framing

Given the massive amount of existing research, I will not attempt to describe all the empirical complexities of the social scientific literature on framing effects. Instead, I will provide a quick sketch of some of the most important studies and attempt to explain how they are relevant to the normative study of politics. In particular, I will be concerned to show how rejecting the principle of invariance results in a more plausible model of human decision making. In later chapters I will demonstrate how acknowledging our susceptibility to framing effects leads to specific conclusions about the design of public institutions in a democracy. For now, however, I am concerned with adequately describing the phenomena in question.

1.2.1 Equivalency Framing Effects

Initially, the manner in which Kahneman and Tversky defined the phenomenon of framing effects was conservative. In their seminal studies, they restricted framing effects to cases in which alternative presentations (the frames) of the decision problem were equivalent and recognized by the respondents as such. Thus, it was only because the experimental subjects in the Asian disease study would assent to the fact that Program A was equivalent to Program A’ and that Program B was equivalent to Program B’ that Kahneman and Tversky were willing to attribute the divergence in choices to the effects of framing (Kahneman 2002, 457). Let us define this kind of phenomenon as an equivalency framing effect: An equivalency framing effect occurs when different but formally equivalent presentations of a decision problem elicit different choices.

More recently (especially in economics), the study of framing effects has been expanded to cases in which researchers can prove the values to be equivalent, even if the experimental subjects do not themselves recognize this. For example, even if subjects did not readily see that Programs B and B’ were equivalent, we could demonstrate their equivalence by uncontroversial mathematical means. In order to retain this conservative definition of framing, the study of framing effects had to focus on situations involving quantifiable values. Here, it is easy to demonstrate equivalence when dealing with things like lottery payoffs, probability calculations, and outcome statistics for medical treatments.

For an example, take the case of an influential study conducted in 1982 by a group of researchers investigating the possibility that framing effects might impact the decision making of patients in health-care scenarios. In this study, the researchers tried to determine whether framing outcome statistics for cancer treatment in terms of survival rates or mortality rates would influence individual choices. Posing a hypothetical choice scenario, they gave subjects the option of pursuing either surgery or radiation treatment for lung cancer. They then monitored the respondents’ choices in order to determine whether the manner in which the outcome statistics were presented would have any effect on the rate at which each option was chosen:

We presented a large number of outpatients, physicians, and graduate students with information describing the possible outcomes of two alternative therapies for lung cancer. The respondents appeared to comprehend and use these data. An interview with the patients after the experiment indicated that they understood the data and were able to recall important items of information. However, the choices of both naive subjects (patients) and sophisticated subjects (physicians) were influenced by several variations in the nature of the data and the form in which they were presented. (McNeil et al. 1982, 392)

Perhaps our most notable finding is the effect on people’s choices of presenting the data in terms of survival or death. Surgery appeared to be much more attractive when the outcomes where framed in terms of the probability of survival rather than in terms of the probability of death. We attribute this result to the fact that the risk of perioperative death looms larger when it is presented in terms of mortality than when it is presented in terms of survival. Unlike the preceding effects, which can be justified or at least rationalized, this effect of using different terminology to describe outcome represents a cognitive illusion. The effect observed in this study is large (25 percent vs. 42 percent) and consistent: It holds for both cumulative-probability and life-expectancy data, for both identified and unidentified treatments, and for all three populations of subjects. Much to our surprise, the effect was not generally smaller for the physicians (who had considerable experience in evaluating medical data) or for graduate students (who had received statistical training) than for the patients (who had neither). (McNeil et al. 1982, 392–393)

In cases such as these it is a simple matter to demonstrate that a 90 percent short-term survival rate is equivalent to a 10 percent immediate mortality rate. As a result, the fact that individuals respond differently to the choice between radiation treatment and surgery (depending on whether the decision is framed in terms of the probability of living or the probability of dying) can be attributed to the efficacy of framing.

1.2.2 Emphasis Framing Effects

There are many instances, however, where there exist no reliable means of demonstrating equivalence but where framing still seems to occur. Let us define this kind of phenomenon as an emphasis framing effect: An emphasis framing effect occurs when emphasizing different elements of a decision problem elicits different choices. These cases are more problematic than equivalency framing effects because, in any given instance, it is possible to claim that the different frames in fact represent different choices. Given that there are no means of demonstrating their formal equivalence, many experimental subjects will not assent to the frames being equivalent, providing instead some explanation of how the frames differ. This results in a problem. If, in any purported case of framing, we cannot demonstrate that the frames are equivalent, then how can we be sure that emphasis framing effects ever occur? Here is another way of presenting this problem: perhaps emphasizing different elements of a decision problem in fact always produces different decisions.

A number of things can be said here. First, there is an inductive argument to be made for the existence of emphasis framing effects. Second, there is an intuitive case for framing in these situations. Third, we must realize that it is not imperative that we arrive at a consensus that any individual purported instance of framing really represents a framing effect, but only that we show that framing effects do occur in many instances.

We should note that in domains where the demonstrability condition is met (such as those studied in psychology, economics, and related social sciences), framing effects appear regularly. Equivalency framing effects have been documented in situations as diverse as personal finance (Kahneman, Knetsch, et al. 2000a), the labor supply of taxi drivers (Camerer, Babcock, et al. 2000), the choice of medical treatments (McNeil et al. 1982), jury deliberations (McCaffery et al. 2002), and consumer choices regarding ground beef (Levin and Gaeth 1988). These examples all involve at least three things: uncertainty, risk, and quantifiable values. Emphasis framing effects, should they exist, would involve only the first two: uncertainty and risk. There is, however, no reason that to think that our susceptibility to frames extends only to domains in which there exist quantifiable values. Therefore, given the pervasiveness of frames in those areas in which we can verify their existence through formal proofs, I think we are entitled to expect the efficacy of framing in other areas as well.

(Continues…)

Excerpted from Framing Democracyby Jamie Terence Kelly Copyright © 2012 by Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.