PS
Online www.apsanet.org
271
Editor’s Note
I
t is rather ironic that just as rational
economic models of politics have
emerged as the most influential paradigm
in political science, experimental econo-
mists have increasingly begun to question
the descriptive validity of expected utility
theory. The rise of the behavioral econom-
ics research program owes a great deal to
psychologists Daniel Kahneman and
Amos Tversky, whose research on
systematic deviations from rationality
quickly transcended their own discipline
and significantly influenced other fields,
including management science, finance,
investment, and consumer economics.
Their work has recently begun to influence
political psychology, international
relations, and other areas of political
science. This brief summary of some of
Kahneman’s major contributions is
intended to provide a context for his
upcoming address at the 2002 APSA
meetings.
It is impossible to do justice to the full
range of Kahneman’s research in the
limited space available here, and I restrict
my focus to his research programs on
judgment and on decision making.
1
Each is
defined by an extraordinarily influential
article coauthored with Tversky: a 1974
Science article on “Judgment under
Uncertainty: Heuristics and Biases”
(which was followed by a 1982 anthology
with the same title with Slovic and
Tversky, and a 1979 Econometrica article
on prospect theory). One measure of the
impact of these articles is that they have
each been cited over one hundred times per
year on average during the last two
decades (Laibson and Zeckhauser 1998),
and the article on prospect theory is
reported to be the most widely cited article
ever published in Econometrica.
2
Judgment under Uncertainty:
Heuristics and Biases
Judgment involves assessments of the
probability or likely frequency of an event.
In the tradition of bounded rationality
(Simon 1955), the Kahneman/Tversky
research program on judgment posits that
people rely on a limited number of
cognitive shortcuts or judgmental heuris-
tics that simplify the complex task of
assessing probabilities in an uncertain
world. Three of the most important
heuristics are representativeness, availabil-
ity, and anchoring and adjustment. While
these heuristics are often useful and
economical, they neglect important
information relevant to probabilities and
consequently may lead to systematic and
potentially severe errors.
The representativeness heuristic refers
to assessments of the likelihood that one
object or event belongs to a particular
category based on the similarity of that
object or event to typical members of that
category. In one experiment (Tversky and
Kahneman 1974), subjects were given
brief personality descriptions of several
individuals and asked to assess, for each
description, the likelihood that it referred
to an engineer or a lawyer. In one
experimental condition subjects were told
that the descriptions were sampled
randomly from a group of 70 engineers
and 30 lawyers, and in a second experi-
mental condition the proportion of the two
professions was reversed. A rational
assessment of probabilities, following
Bayes’s law, would involve some
combination of prior probabilities (based
on the relative percentage of engineers and
lawyers) and current information (based
on the description), leading to different
probability estimates in the two experi-
mental conditions because of differences
in prior probabilities. Subjects’ estimated
probabilities were nearly identical,
however, demonstrating the tendency to
emphasize representativeness or similarity
and to neglect prior probabilities or base
rates. Similar patterns have been identified
in countless other experiments.
The influence of representativeness also
contributes to the violation of the conjunc-
tion rule in probability theory. Given two
events A and B, the probability of the
conjunction or intersection of A and B
cannot exceed the individual probability of
either A or B. In one experiment,
Kahneman and Tversky (1982) gave
subjects the following description: “Linda
is 31 years old, single, outspoken, and
very bright. She majored in philosophy.
As a student, she was deeply concerned
with issues of discrimination and social
justice, and also participated in anti-
nuclear demonstrations.” Subjects were
asked to assess the likelihood that various
statements about Linda were true, among
them being that “Linda is a bank teller”
and “Linda is a bank teller and is active in
the feminist movement.” Over 85% of
subjects believed it was more likely that
Linda was both a bank teller and a feminist
than just a bank teller, contrary to the laws
of probability.
One interesting implication of represen-
tativeness is that assessments of the
likelihood of a particular outcome might be
less than assessments of the likelihood of
the conjunction of that outcome and a
particular causal mechanism that might
have led to it. A study in the early 1980s
found that subjects believed that “a
Russian invasion of Poland, and a
complete suspension of diplomatic
Daniel Kahneman: Judgment, Decision,
and Rationality
Jack S. Levy,
Rutgers University
Daniel Kahneman is the Eugene
Higgins Professor of Psychology and
professor of psychology and public
affairs in the Woodrow Wilson
School at Princeton University. He
will deliver the Decade of Behavior
Lecture at the Annual Meeting in
Boston on Friday, August 30 at
8:30 pm.
The Decade of Behavior Distin-
guished Lecture Program is designed
to showcase behavioral and social
science research addressing the
goals and themes of the Decade of
Behavior initiative
. The
lecture by Kahneman is one of five
sponsored addresses held by
scholarly societies in 2002.
A renowned scholar, Kahneman
has worked in diverse areas of
psychology, most significantly work
done with the late Amos Tversky, in
which they developed some chal-
lenges to the descriptive validity of
the rationality assumption, and
provided psychological analyses of
a number of significant facts of
judgment and choice. Their
pathbreaking work has crossed
disciplines and influenced political
psychology, international relations,
and other areas of political science.
––
Jack Levy is the Board of Gover-
nors’ Professor of Political Science
at Rutgers University and a scholar
of international relations, specializ-
ing in the causes of war, foreign
policy decision making, and
methodology.
ANNUAL MEETING FOCUS: THE DECADE OF BEHAVIOR LECTURE
272
PS June 2002
relations between the USA and the Soviet
Union sometime in 1983,” was more likely
than the suspension of diplomatic relations
alone (Plous 1993).
People use the availability heuristic
when they judge probabilities based on the
ease with which instances of the phenom-
enon in question come to mind, based on
familiarity, vividness, salience, or
emotional impact, regardless of whether
such cognitively retrievable events
constitute a representative sample of the
universe of relevant events. Television
coverage of a single act of terrorism or
airlines crash has far greater impact than
statistical evidence on peoples’ perceptions
of risk. Availability is a central mechanism
underlying many important hypotheses
about learning from the past based on
analogical reasoning. People tend to learn
from events that have a major impact,
affect the individual or society directly,
have occurred recently, and that are
observed first-hand and at a formative
period in a person’s life (Jervis 1976),
despite such events’ not necessarily
representing the class of events that one
wants to predict.
A third important judgmental heuristic
is anchoring and adjustment. Whereas
Bayes’s law prescribes the optimum way
to combine new information with prior
probabilities, evidence suggests that
people’s adjustment mechanisms are
inefficient, that initial values have a
disproportionate impact on final estimates,
that people update their beliefs more slowly
than a rational Bayesian model would
predict, and that updated probabilities
converge with true values slowly if at all.
3
Prospect Theory
Kahneman and Tversky’s research on
prospect theory has had an even greater
impact, at least in economics, than their
work on judgment. Reacting to the
hegemonic position of expected-utility as a
theory of decision under conditions of
risk, they demonstrated experimentally that
people systematically deviate from the
predictions of expected utility-theory and
some of the axioms upon which it is
based.
4
Kahneman and Tversky (1979)
constructed prospect theory to integrate
these behavioral anomalies in rational
decision theory into an alternative theory
of risk choice.
The central analytic assumption of
prospect theory is that people define value
relative to a reference point (reference
dependence) rather than in terms of net
assets, based on evidence that people are
more sensitive to changes in assets than to
net asset levels. People give more weight
to losses from that reference point than to
comparable gains (loss aversion), and they
value what they have more than compa-
rable things not in their possession (the
endowment effect), which in turn makes
actual losses hurt more than foregone
gains. Individuals’ strong aversion to
losses, particularly to losses that are
perceived as certain (as opposed to those
that are perceived as probabilistic), induces
them to take significant risks in the hope
of avoiding loss, even though the result
may be an even greater loss and even
though the expected value of the gamble
may be considerably worse than the value
of the certain loss. The result is risk-
averse behavior with respect to gains and
risk-acceptant behavior with respect to
losses.
With value defined in terms of devia-
tions from a reference point, how people
identify their reference points and hence
“frame” a choice problem is critical. A
change in reference point may lead to a
change in preference (preference reversal)
even if the values and probabilities
associated with possible outcomes remain
unchanged. People facing decisions over
medical treatments, for example, respond
differently to a 90% survival rate than they
do to a 10% mortality rate, although the
two are logically equivalent. These
framing effects are difficult to reconcile
with rational choice, which assumes that
logically identical choice problems lead to
identical outcomes.
5
Similarly, evidence
that an individual’s preference between
two goods can be affected by the introduc-
tion of a third alternative (which may
induce a new reference point) violates
formal utility theory’s basic assumptions
of “independence of irrelevant alterna-
tives” (Kahneman and Tversky 1982).
6
Despite the importance of framing for
choice, we know little about how people
identify their reference points. In static
situations, people often (but not always)
frame their choice problems around the
status quo. In dynamic situations, evidence
suggests that people renormalize their
reference points much more quickly after
making gains (the instant endowment
effect) than after suffering losses.
Experimental studies also show that
individual choice behavior demonstrates a
nonlinear response to probabilities, in
contrast to the linear combination of
utilities and probabilities posited by
expected-utility theory. People overweight
outcomes that are certain relative to
outcomes that are merely probable (the
certainty effect). They also overweight
small probabilities and underweight
moderate and high probabilities. This
means that, except for small probabilities,
people tend to give more weight to the
utility of a possible outcome than to its
probability of occurrence.
7
Kahneman and Tversky (1979)
developed prospect theory to integrate
these observed patterns into an alternative
theory of risky choice. They distinguish
two phases in the choice process. In the
editing phase the actor identifies the
reference point, the available options, the
possible outcomes, and the value and
probability of each of these outcomes. In
the evaluation phase she combines the
values of possible outcomes (as reflected
in an S-shaped value function, which is
characterized by concavity above the
reference point, convexity below it, and a
steeper slope on the loss side) with their
weighted probabilities (as reflected in the
probability weighting function) and then
maximizes over the product. Attitudes
toward risk are determined by the
combination of the S-shaped value
function and the probability weighting
function and not by the value function
alone. Although this combination usually
generates risk aversion for gains and risk
acceptance for losses, it can also produce
(depending on the precise shape of the two
functions) risk acceptance for gains and
risk aversion for losses when probabilities
are small, as illustrated by gambling and
insurance behavior, respectively.
These basic principles lead to a rich and
varied set of propositions about political
behavior.
8
(1) When actors define their
reference points around the status quo,
there is a “status quo bias” that tends to
reinforce the status quo. If actors frame
their choices around a reference point that
is preferred to the status quo (an expecta-
tion or aspiration level, for example), there
is a tendency to move away from the status
quo, which is destabilizing. (2) State
leaders take more risks to maintain their
international positions, reputations, and
domestic political support against potential
losses than they do to enhance their
positions. The same is true for political
leaders and bureaucratic actors in domestic
politics. In addition, domestic publics
punish their leaders more for losses than
for the failure to make gains. (3) Because
people renormalize their reference points
after gains but not after losses, losers are
risk acceptant in their efforts to recover
losses (sunk costs matter) while winners
are risk acceptant in their efforts to defend
the new status quo against subsequent
losses, so that both sides engage in more
risk-seeking behavior than expected-value
theory predicts. (4) People (and the law)
treat errors of commission or action as
more blameworthy than errors of omission
or inaction. Social norms against hurting
another are more compelling than norms to
help another. Similarly, both contract and
tort law distinguish between losses
incurred and gains denied, and judges are
much more reluctant to compensate people
for unrealized profits than for losses.
There are related propositions about
strategic interaction and bargaining. (5)
PS
Online www.apsanet.org
273
Deterring an adversary from making gains
is easier than deterring her from recover-
ing those losses or compelling her to
accept losses. (6) It is easier for political
actors to cooperate in the distribution of
gains than in the distribution of losses,
because they will take more risks and
bargain harder to minimize their share of
the costs than to maximize their share of
the gains. (7) Knowing that “losers” in
political primaries who outperform low
expectations are often better off than
“winners” whose margins of victory are
less than expected, political actors try to
influence observers’ reference points by
lowering expectations of their own
performances and raising expectations of
their adversaries’ performances.
This is not the place for a critique of the
potential utility of prospect theory for the
study of politics.
9
Let me note, however,
that one key set of questions concerns the
generalizability of highly controlled
laboratory studies of individual behavior
in simple choice problems to political
contexts in which the stakes are far higher;
the actors are collective decision-making
bodies rather than individuals; choices are
interactive or strategic rather than deci-
sion-theoretic; time frames are sequential
and dynamic; and key variables of interest
are difficult to measure or even conceptu-
alize along an interval-level scale that is
necessary to test the theory empirically.
10
There are also questions about the utility
of a referent-dependent theory that lacks
both a theory of the reference point and an
established methodology for identifying
reference points empirically.
Conclusion
Kahneman has made seminal contribu-
tions to the interdisciplinary study of
decision and judgment. His work is central
to the bounded rationality research
program and to ongoing debates about the
descriptive accuracy of the expected-utility
microfoundations of rational choice
models of politics. Determining whether
his theory of individual choice can be
extended to the strategic behavior of
collective actors, and tested empirically
against observed behavior in real-world
settings, remains a formidable task for
future research.
References
Farnham, Barbara. 1994. Taking Risks/
Avoiding Losses. Ann Arbor: University
of Michigan Press.
Jervis, Robert. 1976. Perception and
Misperception in International Politics.
Princeton, NJ: Princeton University Press.
Kahneman, Daniel. 1994. “New Challenges
to the Rationality Assumption.” Journal
of Institutional and Theoretical Econom-
ics 150 (1): 18–36.
——, and Amos Tversky. 1979. “Prospect
Theory: An Analysis of Decision Under
Risk.” Econometrica 47:263–91.
——, and Amos Tversky, eds. 2000. Choices,
Values, and Frames. New York: Cam-
bridge University Press.
——, Paul Slovic, and Amos Tversky, eds.
1982. Judgment Under Uncertainty:
Heuristics and Biases. New York:
Cambridge University Press.
——, Ed Diener, and Norbert Schwarz. 1999.
Well-Being: The Foundations of Hedonic
Psychology. New York: Russell Sage.
Laibson, David, and Richard Zeckhauser.
1998. “Amos Tversky and the Ascent of
Behavioral Economics.” Journal of Risk
and Uncertainty 16 (1): 7–47.
Levy, Jack S. 1997. “Prospect Theory,
Rational Choice, and International
Relations.” International Studies Quarterly
41 (1): 87–112.
——. 2000. “The Implications of Framing
and Loss Aversion for International
Conflict.” In Handbook of War Studies II,
ed. Manus I. Midlarsky. Ann Arbor:
University of Michigan Press.
McDermott, Rose. 1998. Risk-Taking in
International Politics: Prospect Theory in
American Foreign Policy. Ann Arbor:
University of Michigan Press.
Plous, Scott. 1993. The Psychology of
Judgment and Decision-making.
Philadelphia: Temple University Press.
Simon, Herbert A. 1955. “A Behavioral
Model of Rational Choice.” Quarterly
Journal of Economics 69:99–118.
Tversky, Amos, and Daniel Kahneman. 1974.
“Judgment under Uncertainty: Heuristics
and Biases.” Science 185:1124–31.
——. 1986. “Rational Choice and the
Framing of Decisions.” Journal of
Business 59: S251–78.
Notes
1. These are Kahneman’s most influential
lines of research. His more recent work on
the psychology of happiness and well-being
(Kahneman, Diener, and Schwarz 1999)
constitutes another major research program.
2. Lowenstein, Roger, “Intrinsic Value:
Outside Who Challenged Dismal Science,”
Wall Street Journal, 6 June 1996.
3. The influence of the anchor is illustrated
in one experiment in which high school
students, given five seconds to estimate the
product of 8x7x6x5x4x3x2x1, produced an
average figure that was four time higher than
for the product of 1x2x3x4x5x6x7x8
(Tversky and Kahneman 1974, 15).
4. The evidence is quite robust. Findings
of departures from rationality have been
confirmed with subjects of considerable
expertise in probability and statistics,
including medical professionals, and it has
been supported by empirical studies of
investment and insurance behavior
(Kahneman and Tversky 2000). The same
patterns have been confirmed by experimen-
tal economists who were determined to (and
expecting to) demonstrate that the findings
were the artifacts of flawed experimental
designs (Camerer 1995).
5. Tversky and Kahneman (1986, S252–
57) refer to this as the “invariance” assump-
tion. Arrow (1982, 6) labels it “extensional-
ity” and describes it as a “fundamental
element of rationality.”
6. Kahneman’s (1994) recent research on
people’s difficulty of correctly anticipating,
at the time they make decisions, the utility
they will experience from a given outcome
also has potentially profound implications for
rationalist theories of behavior.
7. Applications of prospect theory to
political science have given far more
attention to loss aversion and framing than to
probability weighting.
8. This discussion builds on Levy (2000).
See also Farnham (1994) and McDermott
(1998).
9. See Levy (1997) and O’Neill (2001).
10. Kahneman has always been extremely
cautious about possible generalizations from
highly controlled laboratory settings to real-
world behavior.
Dostları ilə paylaş: |