Академический Документы
Профессиональный Документы
Культура Документы
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of
Conflict Resolution.
http://www.jstor.org
By applying the criteria suggested by Lakatos for comparing rival theories, I show that
the expected utility research program developed in The War Traphas yielded an integra-
tive, fairly comprehensive theory that has provided a better empirical understanding of
international conflict than any widely tested alternative. I go on to demonstrate that
Majeski and Sylvan are mistaken with regard to their criticisms of the accuracy, impor-
tance, relevance, and fairness of the theory set out in The War Trap.Also, I show that they
have misunderstood fundamental aspects of the theory and have, therefore, attributed
premises and characteristics to it that it does not possess.
AUTHOR'S NOTE: I would like to thank John Ferejohn, Seifeldin M. Hussein, and
Grace Jusi, who read and commented on earlier versions of the article.
341
indeed I note many of them in the body of the book-but I do claim that
the theoretical approach taken there, rather than being counterproduc-
tive, yields an integrative, fairly comprehensive theory that provides a
better empirical understanding of international conflict than any widely
tested alternative. This is a bold claim, but one which I will support with
clear and explicit evidence.
At the heart of the current critique is the conviction that "Bueno de
Mesquita makes assumptions that are unjustified and inconsistent."
Consistency, as I note repeatedly in TWT, is a central theoretical con-
cern. Propositions derived from an inconsistent argument have, in my
view, no theoretical standing. Whether assumptions are justified or not,
on the other hand, is an empirical question. To quote from TWT, "The
theorist's task is to make assumptions so 'realistic' that subsequent
deductions are not, in Sir Isaac Newton's words, 'dreams and vain
fictions of our own devising. ' " (p. 10) It must be noted at the outset that
Majeski and Sylvan concede the internal consistency of my argument,
and that doing so is critical to their basic challenge of the rational choice
approach. They contend,
we believe that empirical evidence must be adduced indicating that nations' actions
are consistent with those rules and indices [delineated in TWT] and inconsistent
A scientific theory T is falsified if and only if another theory T' has been proposed
with the following characteristics: (1) T' has excess empirical content over T: that is,
it predicts novel facts, that is, facts improbable in the light of, or even forbidden, by
T; (2) T'explains the previous success of T, that is, all the unrefuted content of T is
included (within the limits of observational error) in the content of T'; and (3) some
of the excess content of T' is corroborated [Lakatos, 1978: 32].
ipation" (p. 1). Using a very similar data set, Altfeld and BdM (1979)
apply a "rational-choice theory of how decision makers choose sides, or
neutrality, in ongoing wars" (p. 87). The expected utility model deve-
loped in that study is virtually identical to the multilateral component of
the model developed in TWT. Thus, we have a direct test of the relative
explanatory power of two studies, using essentially the same data, but
with different theoretical foci.
Siverson and King's discriminant classification results explained 29%
of the variance in third-party choices to participate in ongoing wars. In
an analogous test (i.e., did a third party participate or not participate in
the war, as predicted by the theory), the expected utility model explains
56% of the variance. Siverson and King's results yield a 34% reduction in
error while the expected utility approach yields a 64% reduction in error.
What is more, Siverson and King's approach only allowed predictions
of whether a third party would participate in an ongoing war. Altfeld
and BdM's study not only predicted participation, but on which side.
The expected utility approach proved very successful at discriminating
between those who would choose one side or the other-using the
alliance patterns as indicators of utilities-so that on the more difficult
question of predicting on which side nations would fight, Altfeld and
BdM achieved a 42% reduction in error.
"Choosing Sides in Wars" satisfies the Lakatos criteria, at least in
comparison with Siverson and King's study of the same problem.
"Choosing Sides" has the added virtue that it allows us to predict when a
nation will honor its alliance commitments and when it will not. That is
a question of central importance to almost all of the Realpolitik litera-
ture. For instance, in The War Ledger, Organski and Kugler argue,
"Most of the time alliances are simply not a realistic method of pre-
venting threatening changes in the distribution of world power, given the
skewness of relations between the great and the lesser nations, and also
among the half-dozen great powers themselves" (1980: 25). A competing
balance of power hypothesis (Morgenthau, 1973; Claude, 1962; Kaplan,
1957) suggests that it is exactly when two rivals are about equal in power
that allies are critical in tipping the balance in favor of one side or the
other, providing the margin of victory. The trouble is that these two rival
explanations have no way of reconciling their differences so that we can
tell when alliances will not matter. However, the expected utility
approach is capable of doing exactly that.
Organski and Kugler derive their hypothesis from reasoning that says
when two dominant nations fight each other, allies are unimportant
because the dominant states are so much stronger than their allies that
the allies are unable to make a meaningful contribution of power in the
war. Without any empirical evidence beyond their own, it is possible on
purely mathematical grounds to demonstrate that the Organski-Kugler
hypothesis is a subset of the expected utility approach to war. To see
this, I restate the basic form of the expected utility argument as it
appears in my recent paper "The War Trap Revisited" (with the note
that everything deduced in TWTis shown in that paper to be deducible
from the revised model):
E'(U i) =
[(Sj(P.psi)tl + (1 - pi)(Ufi)rl) + (1- Sj)(Usi
~[Q.
11
(Ur)+Q()+
sql i2 sq2 i3 (Ur)i
sq3
Focusing on the component that calculates the expected utility contri-
bution of third parties,
(% [Pik + - - U ri)Cni)
Pjk 1] (Ui
we see that if k is assumed to have virtually no power (with the "P" terms
referring to the probability of success of the subscripted actors and the
"U" terms referring to the utility contribution of k to the relevant
subscripted actor) then,
*
(Pik + Pj k) (Pi + P-)
then
thus
studies have been able to predict the rate of alliance formation, but not
the specific members of such partnerships. The studies by Newman and
Berkowitz use an expected utility framework based on the design set out
in TWTto identify the specific nations joining alliances with each other
and, in Berkowitz's case, even decisions to modify or terminate existing
alliance relationships.
Newman, furthermore, is able to construct a critical test that con-
trasts expectations from the balance of power theory with those from
expected utility theory. The balance of power theory leads to the expec-
tation that as power becomes more concentrated in a few hands, counter-
coalitions are likely to form, thereby increasing the proportion of
aligned nations in the international system. The expected utility frame-
work, on the other hand, suggests that as power becomes more concen-
trated, the opportunities for expected utility maximizing alliances di-
minishes, therefore leading to the hypothesis that the proportion of
nations in alliances is inversely related to the concentration of power in
the international system. Here, then, we have two diametrically opposite
hypotheses. The balance of power notion supports a direct association
between power concentration and systemic alignment; expected utililty
supports an inverse association. Newman finds a strong inverse rela-
tionship, with his results being highly significant substantively and
statistically. The expected utility-based hypothesis explains over 50% of
the variance in systemic alignment. Again in a direct test of expected
utility against a plausible alternative (the balance of power in this case),
the Lakatos criteria favor expected utility.
In essence BdM posits that the decision makers are indifferent to or assign the same
expected utility to both the gamble (the lottery: going to war and either winning or
losing) and the certainty equivalent (not going to war) for all possible results from a
standard lottery. Thus, scaling of policy positions is unaltered by risk.... the
certainty equivalent (no change in policy) is set equal to zero. As we have pointed
out above, this implies that all nations obtain zero utility from no change.
In effect, many of his basic premises arefalse. Now, consider what would happen,
given these incorrect premises, if Bueno de Mesquita's assumptions (qua back-
ground conditions) were realistic. What we would expect, quite simply, would be
that any deductions from these premises would also be false. For example, one of
Bueno de Mesquita's premises is that the gain of winning a war is -1 times the loss
of losing it. This view we have already criticized as seriously reductionistic and
incorrect; but what if it were combined with the realistic assumption that decision
makers are unable to make fine distinctions among alternative courses of action-
say between expected utilities of 0 and -.001 ... ? Under that assumption, the
deduction that calculation of the expected utility of war would give definitive
guidance for fighting or not fighting it would be incorrect, since decision makers
would have to move within ranges of utilities. This would probably eradicate as
meaningful expected utilities hovering just above or below zero. Thus, under
realistic assumptions, Bueno de Mesquita's premises would yield deductions that
are empirically false.
First, at no time have they done more than assert that my assump-
tions are dubious or wrong. They provide no systematic evidence for
these claims. Second, and far more importantly, they seem completely
to deny a consequentialist approach to evidence and the acquisition of
knowledge. They have, we must remember, conceded that my theory is
internally consistent. They concede in the section from which I just
quoted (in which they argue against the widely utilized "as if" principle)
that my theory possesses predictive power. Yet, internal consistency and
predictive power are not, according to them, warrants for viewing the
theory as a useful vehicle for explaining international conflict. Why not?
Because we know of past instances where such consistency and predic-
tive power proved incorrect-Ptolemy's theory of astronomical motion
being the prime example. But here we must look at exactly what they
say.
The key phrase in the quotation cited above is "no one today would
say Ptolemaic theory is therefore correct." Quite right-no one today.
Why not? Because today we have a theory that is more parsimonious
and that renders more accurate predictions. That is, in exactly the
Lakatos sense, the Ptolemaic theory has been supplanted by a better
theory. The Newtonian revolution in science yields better predictions
and explanations than did the Ptolemaic theory. Of course, in the same
sense, Einstein's theory of relativity has supplanted Newtonian theory.
This reminds us that scientific knowledge is a growing, vibrant entity, in
which new perspectives from time to time replace older ones. But, at any
moment, we must rely on the best knowledge available. It does us no
good to say that the future will hold a better explanation than we now
possess. Surely this is always true. One builds on earlier knowledge. One
does not begin any endeavor de novo, as if there were a blank slate. I
have no doubt that future research will modify the efforts set out in
TWT, probably making the original formulation unrecognizable or
even irrelevant. I doubt my theory can survive the challenges of rival
approachs for more than a millenium, as did Ptolemy's. For the sake of
world peace, I pray it does not.
LESSER CRITICISMS
BdM commits the error of reification: measuring phenomena that quite possibly do
not exist. We make no claim that expected utility for war cannot or does not exist ...
but merely that its existence in this case is unproven and, in light of our criticisms in
the previous section, problematic. Here, as elsewhere in The War Trap, BdM fails
to meet the burden of proof.
and empirical support than any other subject in the social sciences.
Furthermore, expected utility theory as an explanation of war-related
phenomena is not original with me. The research on deterrence that
grew out of work by Brodie, Schelling, and others is very much
grounded in an expected utility framework. Even as early a student of
war as Thucydides seemed to have a solid grasp of expected utility as an
explanation for war. Thus, Thucydides, quoting Hermocrates of Syra-
cuse, wrote,
That war is an evil is a proposition so familiar to every one that it would be tedious
to develop it. No one is forced to engage in it by ignorance, or kept out of it by fear,
if he fancies there is anything to be gained by it. To the former the gain appears
greater than the danger, while the latter would rather stand the risk than put up
with any immediate sacrifice. But if both should happen to have chosen the wrong
moment for acting in this way, advice to make peace would not be unserviceable ...
I suppose that no one will dispute that we went to war at first, in order to serve our
own several interests, that we are now, in view of the same interests, debating how
we can make peace; and that if we separate without having as we think our rights,
we shall go to war again [1951: Book IV; 59-60].
REFERENCES