Limitations of Rational Choice Theory1
Raymond Boudon
CNRS
THE ATTRACTIVENESS OF RATIONAL CHOICE THEORY
The appeal of rational choice theory (RCT) has been well explained by
James Coleman (1986): rational choice theory has  a unique attrac-
tiveness as a basis for theory because it is such a complete conception of
action  that we need ask no more questions about it. Hollis (1977) has
expressed the same idea in other words:  rational action is its own explana-
tion (quoted by Goldthorpe 1996). It is true that, once we have explained
that subject X has done Y rather than Y2 because Y was more advanta-
geous, we need to know nothing more. Even if biology was able to describe
adequately the chemical or electrical processes going on in the brain when
a subject makes a decision, this would add nothing to the explanation as
to why the subject did Y. It would merely describe the same process in
a different language. But the biological explanation would be unable to
confirm or disconfirm the rational explanation. This  final aspect of ratio-
nal explanations, the fact that these explanations are without  black box
frustrations, is probably, as suggested by Coleman, the main source of
RCT s attractiveness.
IS RCT GENERAL?
Two questions should be raised before we give RCT the status of a general
theory. Being attractive does not necessarily imply that a theory is accept-
able, valid, or true in all circumstances. That  rational action is its own
explanation is one thing. Whether action can always be considered ratio-
nal in the very special sense RCT gives to the notion of rationality is
another. If all actions and, further, all social phenomena could be validly
explained by RCT, this would be fine. But can they? In the same way, it
would be just fine if I had lost my key bundle under the streetlight where
I may search more effectively than in the surrounding darkness. But a
preference for looking in an illuminated area does not mean that I will
1
Address correspondence to Raymond Boudon, CNRS, 54 Boulevard Raspail, 75270
Paris, Cedex 06, France.
� 1998 by The University of Chicago. All rights reserved.
0002-9602/99/10403-0006$02.50
AJS Volume 104 Number 3 (November 1998): 817 28 817
American Journal of Sociology
drop my keys under the streetlight instead of in a less accessible place.
So, an important question to ask is whether RCT is effectively, as con-
tended by its followers, general. If the answer is yes, the discussion can
stop at this point. If no, the next question is whether RCT can be revised
to make it more general.
Can we legitimately apply RCT to all research situations and to all
problems? The answer to this latter question is no for a simple reason:
RCT assumes that individual action is instrumental, namely that it has
to be explained by the actors will to reach certain goals. Now, action can
be noninstrumental, as most sociologists have recognized. Sch�tz, through
his distinction between Weil and Wozu motive, and Weber, through his
distinction between instrumental and axiological rationality, have stressed
that action is not always instrumental. If the instrumentality of action is
indeed limited, then RCT cannot claim to be a general theory of action.
Thus, one cannot apply RCT notably in the cases where an actor does X
because he believes in Z and that Z implies his doing X independently
of the consequences of X. And even when action is instrumental, it can
mobilize beliefs that need to be explained and that normally will not be
explained by RCT. Thus, the French authorities were, for many years,
more reluctant than the Dutch to use methadone to curb drug addiction.
Why? Because the Dutch thought methadone was an adequate means
while the French thought it was not. But why had the French and the
Dutch different beliefs on this point?2 RCT is of little help here.
Of course, these objections are not new and there are two traditional
ways to overcome them. First, one may promote the generality of RCT
by supposing that actions that appear to be noninstrumental are actually
instrumental at a deeper level. This conversion from noninstrumental to
instrumental is obtained by introducing the postulate that, contrary to
appearances, beliefs are the product of self-interest. This assumption con-
stitutes the core of some classical theories besides RCT. It was introduced
notably by Nietzsche, by Pareto, and by Marx: I believe X, because be-
lieving X serves my psychological interests or my class interests.
A second way of salvaging the generality of RCT is to appeal to Milton
Friedman s epistemology and to treat the causes of behavior, by principle,
as unknowable. In this view, one set of assumptions about the causes of
behavior is as good as any other, and it can legitimately be assumed that
self-interest explains any behavior. By virtue of this  positivistic episte-
mology, this assumption cannot be discussed: the only thing that matters
2
The answer is given by Bergeron (1999) who, rather than RCT, uses the theory of
rationality I advocate here.
818
Symposium: Boudon
is whether or not theories incorporating this postulate reproduce correctly
the observed data.
Assumptions of type 1 can be acceptable in some cases. Thus, I can
believe that social inequalities are unfair because such a belief makes my
poverty undeserved and thus more acceptable to me; alternatively, I be-
lieve that social inequalities are fair because I would then perceive my
opulence as well deserved. It is hard, though, to accept the view that all
beliefs are generated by their psychological or social function. This objec-
tion was explicitly raised by Weber ([1920] 1986, p. 241) against Nietzsche
and Marx:  ressentiment theory applies exclusively to particular cases.
In the general case, psychological or social interests may draw my atten-
tion to a theory and eventually create in my mind a positive or negative
disposition toward the theory in question. But interests alone are generally
unable to explain conviction.
Assumption of type 2 rests upon a very debatable epistemology. Why
such a view as Friedman s  positivism was developed can be understood:
it derives from the reluctance of the positivistic tradition toward taking
into account subjective factors. But such an epistemology is ungrounded:
I can check whether this man whom I see cutting wood in his yard wants
his room to get warmer. If he puts the piece of wood in his chimney, my
interpretation of his behavior will be confirmed. If the weather is hot or
if he starts carving the piece of wood, my interpretation will be falsified.
Even if I cannot perceive directly his reasons, I can reconstruct them.
This reconstruction has the status of a theory that can be confronted with
data. That the reasons motivating people are not directly observable does
not imply that their reconstruction is doomed to be arbitrary. Now, Kiser
and Hecter reject both Friedmanian positivism and also the view that
determining the reasons explaining actions would be an empirical ques-
tion. Instead, they see self-interest as the ultimate real cause of any action.
But this raises a difficult question. The rational choice theorist who en-
counters a voter who tells him that he votes because he considers voting
to be a civic duty will reject the interpretation of the subject himself and,
by application of his own RCT, assume that voting maximizes for the
voter some costs-benefits balance. At this point, the rational choice theo-
rist should explain the  false consciousness he attributes to the actor: Why
does the subject think he votes for one reason while he really votes for
another? But how does the rational choice theorist know that the con-
sciousness of his voter is  false ? I am not saying here that what the actor
himself thinks and says of his own motivation is the ultimate truth. It is
rather one piece of information among others. What I am saying is (1)
that actor s statements about his motivation are facts that, as any fact,
should be taken into consideration and explained, (2) that the Nietzschean,
819
American Journal of Sociology
the Marxian, or the rational choice theorist who attributes to the behavior
of the observed subject causes that the latter does not endorse should
explain his  false consciousness, notably if he rejects the Friedmanian
epistemology.
On the whole, none of the strategies currently used to make noninstru-
mental actions instrumental appears very convincing. Each raises more
questions than it answers. In other words, noninstrumental actions cannot
easily be converted into instrumental actions. Consequently, RCT cannot
be held as general.
Beside this basic objection,  empirical objections can be raised against
RCT, namely that it has never succeeded in explaining satisfactorily im-
portant classes of phenomena. Voting is the best known of these classes.
People vote, though any individual vote has a practically zero probability
of having an effect on the outcome of an election. In this case, the antici-
pated consequences of individual action cannot easily be taken for the
cause of the action. An enormous literature has been devoted to this prob-
lem; it tries to reconcile RCT with the hard facts of voting (see, e.g., Over-
bye 1995). Many other puzzles in the same style could be evoked: Why
do people appear so easily upset by political corruption and so sensitive
to it? This familiar observation cannot be easily explained by RCT, since
political corruption has, in the case of Western democracies at least, a
negligible and invisible effect on the well-being of citizens. In other words,
the rejection here is implausibly the effect of the consideration by social
actors of the consequences of corruption on themselves. Allais (1953) and
a number of authors after him have revealed another Achilles heel of
RCT (if I may make Achilles heels plural) and shown that people do not
behave effectively according to RCT predictions: when they have to
choose between lotteries, in given experimental circumstances, they do
not behave as maximizers.
Rational choice theorists have tried to meet these objections by auxil-
iary assumptions. For instance, nonvoting would include a high social cost
because it would be disapproved of socially; cognitive biases make people
overestimate the weight of a single vote; biases would have the effect that
people see in an erroneous fashion the mathematical expectation of lotter-
ies; cognitive  frames predict that people do not see the world as it is. I
cannot discuss these theories in detail here (I have done so elsewhere; see
Boudon 1996, 1997a). In fact, none of the rational choice theories proposed
to explain why people vote is satisfactory. As to the theories that propose
to reconcile RCT with observed data by introducing the notions of
 frames,  biases, and so on, they appear as empty ( frame,  bias are
mere words); moreover, they purchase the reconciliation with observation
at a high price, since they lose the  unique attractiveness of RCT. Once
one has introduced the assumption that people see the world in such and
820
Symposium: Boudon
such fashion because a bias or a frame affects their perception, the next
question is, namely, Where do these biases come from? As RCT has no
answer to this question, the explanation is no more final; it generates, on
the contrary, large black boxes.
The two categories of objections merge into a general objection. Some
actions are purely instrumental. Among the purely instrumental actions,
some are egoistic. Some actions are not purely instrumental in the sense
that they include a cognitive dimension: the actor wants to reach a goal,
G; he has the impression that M is a good way of reaching G, but the
relation between M and G is not trivial. In that case, the noninstrumental
cognitive dimension of action is the focus of the analysis. Some actions
are not instrumental at all, as when an actor does X not because he wants
to generate some outcome, but because X is a consequence of the princi-
ples he endorses. In that case, the main point in the analysis is to explain
why the actor endorses the principles. Endorsing principles, endorsing a
theory or a viewpoint is also an action, but of the noninstrumental type.
This diversity cannot be forgotten or reduced except by two controversial
strategies: considering the noninstrumental aspects of actions as uninter-
esting and being content with saying that the actors are subject to biases,
frames, and so forth; or assuming that all actions would be at a deeper
level of a unique type: not only instrumental, but egoistic. I agree with
Somers that such an assumption has a metaphysical flavor.
ALTERNATIVE WAY
Instead of trying to salvage RCT against these objections, a more fruitful
move is to question the basic postulates of RCT. Somers says rightly that
the weak side of RCT is not its individualistic approach but its definition
of  rationality : intentionality, self-interest, maximization. Do we need to
endorse this very special view about rationality? Do we really need to
accept the idea that all actions are not only consequential but egoistic?
Is this version of rationality the only one representing the uniqueness of
providing explanations without black boxes? The greatest of classical his-
torical sociologists, Tocqueville and Weber, have implicitly answered this
question. Yes, they said, action should be considered as meaningful; yes,
the meaning of his action to the actor should be perceived as its cause;
yes, in most cases, the meaning to the actor of his action resides in the
reasons he perceives as strong to adopt it: in other words, they introduced
the general postulate that the causes of an action reside in the reasons the
actor has of adopting this action. And they added that, depending on the
situation the actor is involved in, these reasons can take the form of cost-
benefit considerations but also other forms. Thus, endorsing a theory is in
most cases an action caused by the fact that one sees strong reasons of
821
American Journal of Sociology
endorsing it. Priestley endorsed the phlogiston theory, not only because
he had a strong interest defending it, but because he was convinced it
explained many facts better than did alternative theories.
These classical writers should be heard: they have solved implicitly by
anticipation some of the crucial questions raised by the present discussion
and indicated a path worth following today. They start from the view
that (what we call) RCT can be useful notably in the situations where the
actor is invited to apply a cost-benefit analysis by the very nature of the
situation. But, in other circumstances, the notion of  rationality must be
given another content. In modern words, RCT is not a general theory
because it uses a much too rigid and narrow conception of rationality. To
illustrate, I will evoke briefly two examples from Tocqueville.
TWO EXAMPLES FROM TOCQUEVILLE
Example 1
Using implicitly  methodological individualism and RCT. At one point
in his Old Regime Tocqueville ([1856] 1955) wonders why, at the end of
the 18th century, French agriculture remains stagnant at a time when
agriculture is flourishing in England. This is particularly puzzling since
the physiocrats, who develop the view that modernizing agriculture is the
main path to growth, are politically very influential in France at the time.
Tocqueville s explanation: administrative centralization is the cause of
the fact that positions of  civil servants are more numerous and hence
more easily available in France than in England. Also, the French central-
ization makes serving the king in France a unique source of prestige, in-
fluence, and power; consequently, other things equal, landlords are more
easily incited in France than in England to leave their lands and buy a
royal position. In England by contrast, being an innovative landowner
not only produces local respect and prestige, it may also open the way to
Westminster. This macroscopic difference between England and France,
summarized by Tocqueville by his notion of  administrative centraliza-
tion, explains why landlord absenteeism is much larger in France than
in England. Further, landlord absenteeism is the cause of a low rate of
innovation: since their interests are at the court, the landlords themselves
have little motivation to innovate; as to the farmers who run the landown-
erships, they would have a motivation to innovate, but hardly the capacity
of doing so. Finally, the low rate of innovation is responsible for the stag-
nation of agricultural development in France.
In this discussion, Tocqueville uses  methodological individualism
(MI). The macroscopic difference between France and England is ex-
plained as the effect of individual decisions taken by the landlords. The
individual decisions are analyzed as taken, not by  angels, but by men
822
Symposium: Boudon
belonging to social contexts. The parameters characterizing the French
and British contexts are themselves the products of a long history. This
point gives me the opportunity of stressing that MI does not imply solip-
sism as soon as individual decisions are analyzed, as here by Tocqueville,
as affected by the parameters characterizing the context.3 Finally, Tocque-
ville uses here what we call RCT: by leaving their land and serving the
king, the landlords gain in influence, prestige, and so on. In England by
contrast, it is a better strategy to appear locally as a modern and efficient
landlord. The macroscopic statement  centralization is a cause of agricul-
tural underdevelopment appears as entirely acceptable, because it is sup-
ported by this individualistic analysis. Though  centralization is a com-
plex factor, it is identified with precise  parameters that affect the
situation of decision making of the actors, here the landlords.  Centraliza-
tion is a construct. But it is not a mere word. In summary, Tocqueville
uses MI and, moreover, he uses the basic behavioral axiomatics of RCT;
the individuals are analyzed as selfish, goal-oriented, and maximizers. It
can be noted incidentally that Tocqueville s path has been literally fol-
lowed by Root (1994) in his illuminating book on the comparative devel-
opment of the modern state in Britain and France.
Example 2
Using implicitly MI, but rejecting RCT. In other circumstances,
Tocqueville uses MI but not RCT. Thus, he wonders, again in his Old
Regime, why the cult of Reason became immensely popular in France at
the end of the 18th century, but not in England. His answer is that tradi-
tional institutions, and hence  Tradition with a capital T, were totally
disqualified in France but not in England. Thus, the British aristocracy
fulfilled important social and economic roles. Consequently, its higher sta-
tus was considered by people as grounded and legitimate. In France, by
contrast, the gentry had no visible social and economic function except
sitting in Versailles. Those members of the gentry who were not able to
buy a royal position remained on their land. Poor and bitter, they stuck
ritualistically to their privileges. Their officially higher rank was perceived
by the peasants as illegitimate. As it was the product of tradition, the
peasants came to the idea that institutions deriving their strength from
tradition were bad. So, when the philosophes proposed to substitute for
institutions grounded on Tradition, a society grounded on what they pre-
sented as the opposite term, namely Reason with a capital R, they had
3
Bunge (1996) stresses rightly that MI (especially as I have applied it; see Boudon
1974) has nothing to do with atomism or solipsism and is perfectly compatible with
his systemism.
823
American Journal of Sociology
immediate success. After all, the notion translated widespread feelings.
Tocqueville makes clear that this success cannot be analyzed as the prod-
uct of interpersonal influence, since it was immediate. So, the macroscopic
phenomenon under examination, namely the fulgurant success of the idea
of Reason, is analyzed by Tocqueville as the effect of the fact that individ-
ual French peasants, lawyers, and so on, accepted easily the theory that
good institutions should be the effect of social engineering (in our lan-
guage), be the product of Reason (in 18th-century parlance). This analysis
follows MI, but not RCT. Here again, a social fact, in this case a difference
between France and England, is analyzed as the product of reasons, but
not of the RCT type. Individual peasants tend to endorse the political
theory proposed by the philosophes because this theory appears to them
as valid. Evidently, they expected returns for themselves from the applica-
tion of this theory. Most of them were probably convinced that their condi-
tion would become better if the ideas of the philosophes were applied.
But this does not explain why they saw  social engineering as a good
political philosophy. Tocqueville uses here implicitly a view of rationality
that I have proposed as  cognitive rationality. The peasants endorse the
political theory of the philosophes because they have strong cognitive rea-
sons for seeing it as valid. This type of rationality is typically at work in
the case of the scientist who chooses theory T against a theory T2 . Cogni-
tive rationality in my language overlaps with  rationality in the sense
that historians and philosophers of science use this word. Tocqueville s
analysis provides a powerful hint here: that the type of rationality at work
in the endorsement by scientists of an idea or theory is also at work in
 ordinary knowledge.
COGNITIVE RATIONALITY
Cognitive rationality should be distinguished from instrumental rational-
ity. First, because endorsing a theory is a noninstrumental action. Second,
because the question the actor is confronted with here is not to maximize
any cost-benefit balance, but to check whether, to the best of his knowl-
edge, an idea is acceptable. Radnitzki (1987) has tried to reduce this cogni-
tive rationality to RCT. When more and more facts appeared easily ex-
plainable by the theory that the earth was spherical, it became more and
more difficult to develop alternative arguments supporting the theory that
the earth was flat. Radnitzki proposes to substitute  costly for  difficult
and makes then the point that the choice between alternative scientific
theories can also be analyzed in RCT terms. But the main point is that
the arguments supporting the theory that the earth is spherical appeared,
after a while, much stronger than the arguments supporting the alterna-
tive theories. Therefore, little is gained by substituting  costly for  diffi-
824
Symposium: Boudon
cult.  Cost is namely a consequence of  difficulty. So, what needs to be
explained is why a set of arguments appears as defensible or not.
AXIOLOGICAL RATIONALITY
Although I cannot go very far in a short note on this, another important
point, as suggested by Weber, is  axiological rationality, which should be
distinguished from  instrumental rationality. This notion has been much
discussed, and what Weber meant by it is not clear. My own interpretation
(see Boudon 1997a) is that he wanted to introduce the idea that in some
circumstances actors do X not because they expect any desirable conse-
quence, but because they are convinced that X is good, since it is grounded
on strong reasons. Thus, Weber would probably never have considered
voting as a paradox. It is a paradox as long as one assumes that rational
action is always consequentialist, as RCT assumes. In that case, voting
is well a paradox, since my vote has with quasi certainty no consequence
on the outcome of the election. People vote, though. Why? Because they
have strong reasons to believe that democracy is better than alternative
regimes, they see that elections constitute a major institution of democ-
racy, they understand the principle  one man, one vote, they see that this
principle is an expression of a basic value, and so forth. In other words,
they vote because one should vote if one believes in the value of democ-
racy. This explanation mobilizes what, following my interpretation, We-
ber called  axiological rationality. I vote because I think I should vote.
I think I should vote because I have strong reasons to believe in democ-
racy. Of course, I will have strong reasons to refrain from voting if none
of the candidates convinces me, if I do not know how I should vote, for
example.
As I said earlier, many efforts have been made to explain voting by RCT
theories. But these theories are all unconvincing and moreover assume
implicitly without explaining it the existence of a  false consciousness of
the voter: this is, then, a huge black box.
CONCLUSION
Finally, we come to the idea that Tocqueville and Weber among others
have sketched a model, which I have proposed to call the  cognitivist
model (CM), resting on the following postulates (see Boudon 1994, 1996a,
1997b, 1998).
1. Until the proof to the contrary is given, social actors should be consid-
ered as rational in the sense that they have strong reasons of believing
what they believe, of doing what they do, and so forth.
2. In particular cases, these reasons can be realistically treated as deal-
825
American Journal of Sociology
ing with the difference between costs and benefits of alternative lines of
action. In other cases, they cannot: in particular when a decision or an
action rests upon normative or cognitive beliefs, the reasons will generally
not belong exclusively to this type. This results from the fact that beliefs
are unintentional, and that normative beliefs are not always consequen-
tially grounded. Also, in many circumstances, a social actor can be person-
ally distant from an issue, yet have strong feelings about it (a good example
of this is the use of the death penalty in the United States). This point is
crucial: it suggests that RCT is of limited use in the analysis of public
opinion.
3. In some circumstances, the core of some action is constituted by  cog-
nitive reasons: he did X because he believed Z is likely or is true, and
because he had strong reasons of believing so.
4. In some circumstances, the core of some action is constituted by axio-
logical reasons: he did X because he believed that Z is fair, good, unfair,
and so on, and had strong  nonconsequential reasons of believing so.
It follows from these postulates that RCT is a particular case of CM.
When the reasons in CM are restricted to belong to the benefits-minus-
costs type, we get RCT. Reciprocally, when the restriction that reasons
should belong to the benefits-minus-costs type is lifted in RCT, we get
CM. Again, RCT is a powerful model; it cannot be held as a general
theory.4
CM supposes that actions, decisions, and beliefs are meaningful to the
actor in the sense that they are perceived by him as grounded on reasons.
Even though he cannot be able to identify these reasons clearly, he has
the intuitive impression that they are grounded on reasons.
Two important remarks can be introduced here. Although it is tautolog-
ical to define  rationality by the notion of  strong reasons, it is the only
way of getting rid of the discussions as to  what rationality really means,
where the discussants expose generally what they mean. As to the postu-
late that beliefs and actions are grounded on reasons, it is not tautological.
Many traditions start, on the contrary, from the assumption that actions
and beliefs are not the effect of reasons. As to finding out those reasons,
reconstructing them, this can be a hard job: Why were the French land-
lords less innovative than their British counterparts? Why did Priestly
believe in the phlogiston? Why did Englishmen of the 18th century believe
that miners should be paid more than soldiers? (On this, see Smith 1976.)
Why did the people of London not try to exert pressure on political power
by street gatherings as frequently as did the Parisians? (See Root 1994.)
4
I defended the two points already (Boudon [1977] 1982, chap. 7) but explored else-
where the non-RCT dimensions of rationality (see Boudon [1994] and later publica-
tions).
826
Symposium: Boudon
Why are magical beliefs more likely to found in some societies than in
others? (See Boudon 1994.) Why was methadone used much earlier in
Holland than in France? (See Bergeron 1999.) All of these questions have
been convincingly answered in works that use MI and an open theory of
rationality rather than the special figure of rationality used by RCT. In
particular it can be noted that, in his Wealth of Nations, Adam Smith,
RCT s spiritual father, solves the above question about miners and sol-
diers using what I call the CM rather than RCT (Boudon 1996a, 1998).
The second remark, which I cannot develop, is that CM excluded radi-
cally solipsism: I cannot perceive as strong the reasons leading me to en-
dorse a statement  X is good, legitimate, right, true, and so forth without
conceiving these reasons as grounded and hence as intersubjectively valid
(Boudon 1995).
Finally, a crucial question is raised by Somers: What is a good theory?
Sometimes, good scientific theories use mathematical language, are de-
rived from a general theory, and so on. But such attributes are not compo-
nents of a good theory generally. Otherwise, physical theories would be
good, but biological theories bad, since the latter make little use of mathe-
matics and are hardly deducted from a general theory. Celestial mechanics
is not the model to be followed by all disciplines. A good scientific explana-
tion of a phenomenon P is rather a set {S} of statements meeting three
requirements: (1) that all s " {S} are acceptable, (2) that {S} �! P, (3) that
relevant facts are not arbitrarily ignored. They are satisfied, for instance,
in Tocqueville s above examples: all statements are acceptable (simple
psychological statements, empirical statements congruent with observa-
tion, no  black box concepts in the components of the statements, etc.);
moreover, {S} �! P (P stagnation of French agriculture, enthusiasm of
the peasants for the idea of Reason); finally the theory explains not only
the observed behavior but the verbal statements of actors as they have
reached us.
REFERENCES
Allais, M. 1953.  Le comportement de l homme rationnel devant le risque: Critique
des postulats de l �cole am�ricaine. Econometrica 21 (4): 503 46.
Bergeron, H. 1999. Soigner ou prendre soin des toxicomanes: Anatomie d une croyance
collective. Paris: Presses Universitaires de France.
Boudon, R. 1974. Education, Opportunity and Social Inequality. New York: Wiley.
   . (1977) 1982. The Unintended Consequences of Social Action. London: Mac-
millan.
   . 1994. The Art of Self-Persuasion. London: Polity Press.
   . 1995. Le juste et le vrai. Paris: Fayard.
   . 1996.  The  Rational Choice Model : A Particular Case of the  Cognitive
Model.  Rationality and Society 8 (2): 123 50.
   . 1997a.  Le  paradoxe du vote et la th�orie de la rationalit�. Revue Fran�aise
de Sociologie 38 (2): 217 27.
827
American Journal of Sociology
   . 1997b.  The Present Relevance of Max Weber s Wertrationalit�t (value ratio-
nality). In Methodology of the Social Sciences, Ethics, and Economics in the Newer
Historical School: From Max Weber and Rickert to Sombart and Rothacker, edited
by P. Koslowski. New York: Springer.
   . 1998.  Social Mechanisms without Black Boxes. In Social Mechanisms: An
Analytical Approach to Social Theory, edited by R. Swedberg and Peter Hedstrom.
New York: Cambridge University Press.
Bunge, M. 1996. Finding Philosophy in Social Science. New Haven, Conn.: Yale
University Press.
Coleman, J. 1986. Individual Interests and Collective Action: Selected Essays. Cam-
bridge: Cambridge University Press.
Goldthorpe, J. 1996.  Class Analysis and the Reorientation of Class Theory: The Case
of the Persisting Differentials in Educational Attainment. Manuscript. Nuffield
College, Oxford University.
Hollis, M. 1977. Models of Man: Philosophical Thoughts on Social Action. Cambridge:
Cambridge University Press.
Overbye, E. 1995.  Making a Case for the Rational, Self-Regarding  Ethical Voter
. . . and Solving the  Paradox of Not Voting in the Process. European Journal of
Political Research 27:369 96.
Radnitzky, G. 1987.  La perspective �conomique sur le progrŁs scientifique: Applica-
tion en philosophie de la science de l analyse co�t-b�n�fice. Archives de Philoso-
phie 50 (avril juin): 177 98.
Root, H. L. 1994. The Fountain of Privilege: Political Foundations of Economic Mar-
kets in Old Regime France and England. Berkeley and Los Angeles: University of
California Press.
Smith, A. (1776) 1976. An Inquiry into the Nature and Causes of the Wealth of Nations.
Oxford: Clarendon Press.
Tocqueville, A., De. (1856) 1955. The Old Regime and the French Revolution. New
York: Doubleday.
Weber, M. (1920) 1986. Gesammelte Aufs�tze zur Religionssoziologie, vol. 1. Munich:
Mohr.
828