Professional Documents
Culture Documents
by Robert A. Dahl
IJepartment of Political Science, J'ale Cniversity
Homething like it, to Home Thing they be- had countleR~ profitable discuHsions during a year
lieve they have obseryed, one is tempted to we both spent as fellows at the Center for Ad-
vlillecd Study in the Behavioral Heiellces. I have
suppose that the Thing mmlt exist; and not drawn freel~: not only on Ollr joint work but on
only exist, but exiHt in a form capable of his own published and unpublished writings on the
201
"
of systematic and empirical import, then we some specific purposes, and the actual 01'
simply cannot say whether rigorous defi- possible results of these operational d, fi-
nitions of the concept of power are likely nitions.
to be useful in theoretical systems with a I should like to be permitted one liberty. ;
relatively large pay-off in the hard coin of There is a long and honorable history at- l'
scientific understanding. The evidence is tached to such words as power, influence, ~
not yet in, control, and authority. For a great many
I think it can be shown, however, that purposes, it is highly important that a
to define the concept "power" in a way that distinction should be made among them;
seems to catch the central intuitively under- thus to Max Weber, "Herrschaft ist . .. ein
stood meaning of the word must inevitably Sonderfall von Macht," Authority is a special
result in a formal definition that is not easy case of the first, and Legitimate Authority
" to apply in concrete research problems; a subtype of cardinal significance (11).
and therefore, operational equivalents of In this essay I am seeking to explicate the
Difficult to come up with
the formal definition, designed to meet the primitive notion that seems to lie behind
one definition because needs of a particular research problem, are all of these concepts. Some of my readers
different def.s suited to likely to diverge from one another in im- would doubtless prefer the term "influence," "
different questions portant ways. Thus we are not likely to while others may insist that I am talking
produce-certainly not for some consider- about control. I should like to be permitted
able time to come-anything like a single, to use these terms interchangeably when
consistent, coherent "Theory of Power." it is convenient to do so, without denying
Weare much more likely to produce a or seeming to deny that for many other
variety of theories of limited scope, each purposes distinctions are necessary and
of which employs some definition of power useful. Unfortunately, in the English lan-
that is useful in the context of the particular guage power is an awkward word, for unlike
>, piece of research or theory but different in "influence" and "control" it has no con-
important respects from the definitions of venient verb form, nor can the subject and
other studies. Thus we may never get object of the relation be supplied with noun
through the swamp. But it looks as if we forms without resort to barbaric neologisms.
might someday get around it. POWER AS A RELATION AMONG PEOPLE
With this in mind, I propose first to essay
a formal definition of power that will, I What is the intuitive idea we are trying
hope, catch something of one's intuitive to capture? Suppose I stand on a street
notions as to what the Thing is. By "formal" corner and say to myself, "I command all
I mean that the definition will presuppose automobile drivers on this street to drive
the existence of observations of a kind that on the right side of the road"; suppose
may not always or even frequently be further that all the drivers actually do as I
possible. Second, I should like to indicate "command" them to do; still, most people
>,
how operational definitions have been or will regard me as mentally ill if I insist that
might be modelled on the formal one for I have enough power over automobile
drivers to compel them to use the right side
subject. The comments of Jacob Marschak on this
of the road. On the other hand, suppose a
paper have also been most helpful. There are, of policeman is standing in the middle of an
course, approaches radically different from the intersection at which most traffic ordinarily
one employed here and in the works mentioned moves ahead; he orders all traffic to turn
above. John R. P. French, Jr. (2), has developed a right or left; the traffic moves as he orders I
model that assumes "a unidimensional continuum
of opinion which can be measured with a ratio it to do. Then it accords with what I con-
scale," and he defines "the power of A over B ceive to be the bedrock idea of power to
(with respect to a given opinion) [to bel equal to say that the policeman acting in this particu-
the maximum force which A can induce on B lar role evidently has the power to make
minus the maximum resisting force which B can
mobilize in the opposite direction." Game theory
automobile drivers turn right or left rather Definiton of
" provides still another approach. Cf. References 4, than go ahead. My intuitive idea of power, power
5,9. then, is something like this: A has power
over B to the extent that he can get B to patronage, his constitutional veto, the bases of power = formal decision
making powers
do something that B would not otherwise do. po:';Hihility of calling White House con-
H Hume and his intellectual successors ferences, his influence with the n'1tional
had never existed, the distinction between eleetorate, hi:,; charisma, his charm, and the
the two events above might be firmer than like.
it is. But anyone who sees in the two cases In a sense, the ha:,;e i:,; inert, passive. It
the need to distinguish mere "association" must be exploited ill some fashion if the
from "cause" will realize that the attempt hehavior of othen, is to be altered. The
to define power could pmih UH into some means or in:,;truments of such exploitation
messy epistemological problems that do not are numerous; often they involve threats or
seem to have any generally accepted so- promises to employ the base in some way
lutions at the moment. I shall therefore and they may involve aetual use of the base.
quite deliberately steer dear of the possible In the case of the President, the mean:,;
identity of "power" with "cause," and the would inelude the prumise of patronage,
host of problems this identity might give the threat of veto, the hulding of a conference,
rise to. the threat of appeal to the eleetorate, the
power = Let us proceed in a different way. First, exercise of charm and charisma, etc.
relation among people let us agree that power iH a relation, and that Thus the means is a mediating activity hy
it is a relation among people. Although in A between A's base and B's response. The
common speech the term encompasses re- scope consists of B's response:,;. The :,;cope
lations among people and other animate or of the President's power might therefore
inanimate objects, we shall have our handH inelude such Congressional actions as
full if we confine the relationship to human passing or killing a bill, failing to override
beings. All of the social theory I mentioned a veto, holding hearings, etc.
earlier is interesting only when it deals with The amuunt of an actor's power can he
this limited kind of relatioIli-3hip. Let UH represented by a probability statement:
call the objects in the relationship of power, e.g., "the chance" are 9 out of 10 that if the
actors. Actors may be individuals, groups, President promises a judgeship to five key
roles, offices, governmentH, nation-states, Senators, the Senate will not override his
or other human aggregates. veto," etc. Clearly the amount can only be
To specify the actors in a power relation-- specified in conjunction with the means
A has power over B-is not very interesting, and scope.
informative, or even accurate. Although the Suppose now we should wish to make a
statement that the President has (Home) relatively complete and eoncise Htatement
power over Congress iH not empty, neither is about the power of individual A over indi-
it very useful. A much more complete vidual a (whom I shall call the respondent)
statement would include references to (a) with respect to some given scope of re-
Elements of power sponses. In order to introduce the basic
the source, domain, or base of the President's
power over Congress; (b) the means or ideas involved, let us restrict ourselves to
instruments used hy the President to exert the 2 by 2 ease, where the aetor A does or
power over CongresH; (c) the amount or does not perform some act and the re-
extent of his power over Congress; and (d) spondent a does or does not "respond."
the range or scope of his power over CongresH. Let us employ the following symbols:
The base of an actor's power consists of all
the reHources-opportunities, acts, objeets, (A, w) 11 does w. For example, the
etc.-that he can exploit in order to effect President makes a nation-
the behavior of another. i\Iuch of the best wide television appeal for
writing on power-Bertrand Russell is a tax increases.
good example-eonsiHts of an examination of (A, w) A does not do w.
the possible bases of power. A study of the (a, :r) a, the respondent, does :c. For
war potential of nations iH also a study of the example, the Senate votes to
bases of power. Some of the possible bases of increase taxes.
a Pre:,;ident's power over a Senator are his Ca, i) a does not do .r.
P (ulv) = Probability that u happens can be said to exist. I shall leave the con-
when v happens. cept of "connection" undefined, for I wish
'. only to call attention to the practical sig-
Then a relatively complete and concise nificance of this second condition. In looking
statement would be symbolized:
for a flow of influence, control, or power from
Pea, xlA, w) = Pl A to a, one must always find out whether ,.
there is a connection, or an opportunity for a ~
Pea, xlA, w) = P2 connection, and if there is not, then one need ~
Suppose now, that Pl = 0.4 and P2 = 0.1. proceed no further. The condition, obvious
Then one interpretation might be: "The as it is, thus has considerable practical
probability that the Senate will vote to importance for it enables one to screen out
increase taxes if the President makes a many possible relations quite early in an
nationwide television appeal for a tax inquiry.
increase is 0.4. The probability that the 3. In examining the intuitive view of the
Senate will vote to increase taxes if the power relation, I suggested that it seemed
'. to involve a successful attempt by A to
President does not make such an appeal is
0.1." get a to do something he would not other-
wise do. This hints at a way of stating a
PROPERTIES OF THE POWER RELATION
third necessary condition for the power l:
N ow let us specify some properties of the relation. Suppose the chances are about one r
power relation. out of a hundred that one of my students, ;
Causality/power 1. A necessary condition for the power .Jones, will read The Great Transformation '
relation is that there exists a time lag, during the holidays even if I do not mention
however small, from the actions of the the book to him. Suppose that if I mention
actor who is said to exert power to the re- the book to him and ask him to read it, the
sponses of the respondent. This require- chances that he will do so are still only one
ment merely accords with one's intuitive out of a hundred. Then it accords with my
belief that A can hardly be said to have intuitive notions of power to say that
'. power over a unless A's power attempts evidently I have no power over Jones with
precede a's responses. The condition, ob- respect to his reading The Great Trans-
vious as it is, is critically important in the formation during the holidays~at least
actual study of power relations. Who runs not if I restrict the basis of my action to
the XYZ Corporation? Whenever the mentioning the book and asking him (po-
president announces a new policy, he im- litely) to read it. Guessing this to be the
mediately secures the compliance of the top case, I tell Jones that if he does not read the
officials. But upon investigation it turns book over the holidays I shall fail him in
out that every new policy he announces has my course. Suppose now that the chances
first been put to him by the head of the sales he will read the book are about 99 out of
department. Or again, suppose we had a 100. Assume further that nothing else in
full record of the times at which each one of Jones's environment has changed, at least
the top Soviet leaders revealed his positions nothing relevant to his reading or not read-
'. on various issues; we could then deduce a ing the book. Then it fully accords with my
great deal about who is running the show intuitive notions of power to say that I
and who is not. A good bit of the mystery have some power over Jones's holiday read-
surrounding the role of White House figures ing habits. The basis of my power is the
like Sherman Adams and Harry Hopkins right to fail him in his course with me, and
would also be clarified by a record of this the means I employ is to invoke this threat.
kind. Let me now set down symbolically what
2. A second necessary condition is, like I have just said. Let
the first, obvious and nonetheless important
in research: there is no "action at a dis- (D, w) = my threat to fail Jones if he does
tance." Unless there is some "connection" not read The Great Trans-
between A and a, then no power relation formation during the holidays.
'.
scope of their power, i.e., in type of re- interested in the relative effectiveness of the
sponse evoked, (4) differences in the number threat of veto employed by different gover-
of comparable respondents, and (5) differ- nors.
ences in the change in probabilities, or M. In whatever fashion one chooses to define
The first two of these may be conveniently the relevant properties of the actors whose
"
thought of as differences in properties of the power he wishes to compare, strictly speak-
actors exercising power, and the last three ing one must compare them with respect to
may be thought of as differences in the the responses they are capable of evoking.
responses of the respondents. Now it is Ideally, it would be desirable to have a .,. ~
clear that the pay-off lies in the last three- single measure combining differences in '.
the responses. When we examine the first scope, number of comparable respondents
two in order to compare the power of in- controlled, and change in probabilities.
dividuals, rulers, or states, we do so on the But there seems to exist no intuitively
supposition that differences in bases and satisfying method for doing so. With an
means of actors are very likely to produce average probability approaching one, I
differences in the responses of those they can induce each of 10 students to come to
seek to control. class for an examination on a Friday after-
'. As I have already indicated, much of the noon when they would otherwise prefer to
most important and useful research and make off for New York or Northampton.
analysis on the subject of power concerns With its existing resources and techniques,
the first two items, the properties of the the New Haven Police Department can
actors exercising power, and there is good prevent about half the students who park comparisons of
reason to suppose that studies of this kind along the streets near my office from staying power between
will be as indispensable in the future as they beyond the legal time limit. Which of us has actors must be
have been in the past. But since we are the more power? The question is, I believe, made in same
concerned at the moment with a formal incapable of being answered unless we are scope and domain
explication of the concept of power, and ready to treat my relationships with my
not with an investigation of research prob- students as in some sense comparable with
lems, (some of these will be taken up later the relations of the Police Department to
on) it is important to make clear that another group of students. Otherwise any
" analysis of the first two items does not, answer would be arbitrary, because there is
strictly speaking, provide us with a com- no valid way of combining the three vari-
parison of the power of two or more actors, ables-scope, number of respondents, and
except insofar as it permits us to make change in probabilities-into a single scale.
inferences about the last three items. If we Let us suppose, for a moment, that with
could make these inferences more directly, respect to two of the three variables the Diffilcult to
we should not be particularly interested in responses associated with the actions of compare power
the first two items-at least not for purposes two (or more) actors we wish to compare across actors
unless reposnes
of making comparisons of power. On the are identical. Then it is reasonable to define
exactly the same
other hand, given information about the the power of A as greater than the power
responses, we may be interested in comparing of B if, with respect to the remaining vari-
the efficiency of different bases or means; able, the responses associated with A's acts
in this case, evidently, we can make a are greater than the responses associated
" comparison only by holding one or both of with B's acts. It will be readily seen, how-
the first two factors constant, so to speak. ever, that we may have jumped from the
In general, the properties of the power frying pan into the fire, for the term "greater
wielder that we bring into the problem are than" is still to be defined. Let us take up
determined by the goals of one's specific our variables one by one.
research. For example, one might be in- To begin with, we may suppose that the
terested in the relative power of different probability of evoking the response being the
state governors to secure favorable legis- same for two actors and the numbers of
lative action on their proposals by means comparable persons in whom they can evoke
of patronage; or alternatively, one might be the response also being the same, then if the
"
Copyright (c) 2001 ProQuest Information and Learning Company
Copyright (c) General Systems Science Foundation
CONCEPT OF POWER 207
seope of responses evoked by A is greater tel' mined only in view of the nature and aims
than that evoked by 13, A's power is greater of the research at hand.
than 13's. But how can we decide whether The third n1riable is the only one of the
one scope is larger than another? Suppose three without this inherent limitation. If Comparing probability of
that I could induce my son to bathe every scope and numberH are identical, then there desired outcome
evening and to brush his teeth before going can be no doubt, I think, that it fully ac-
to bed and that my neighbor could induce cords with our intuitive and common-sense
Comparing power:his son to serve him breakfast in bed every notions of the meaning of power to say that
comparable morning. Are the two respOllses I can con- the actor with the highest probability of
scope trol to be counted as greater than the one securing the response is the more powerful.
response my neighbor can control? Evi- Take the set of Democratic Senators in the
dently what we are willing to regard as a Gnited States Senate. Suppose that the
"greater" or "lesser" scope of responses will chance that at least two-thirds of them will
be didated by the particular piece of re- support the Pref-iident's propoHals on federal
search at hand; it seems fruitless to at- aid to education is O.G. It is fair to say that
tempt to devise any single seale. At one no matter what I may do in behalf of federal
extreme we may wish to say that A's scope
aid to education, if there are no other
is greater than B's only if A's scope con-
changes in the situation except those brought
tains in it every response in 13's and at
about by my efforts the probability that
least one more; this would appear to be the
narrowest definition. At the other extreme, two-thirds of them will support federal
we may be prepared to treat a broad cate- aid will remain virtually at O.G. If, on the
gory of responses as comparable, and A's other hand, Senator .Johnson, as majority
scope is then said to be greater than B's leader, lends his full f-iUpport and all his skill
if the number of comparable responsef-i in of maneuver to the measure the probability
his seope if-i larger than the number in E'H. may rise, let UH say, to 0.8. vVe may then
There arc other pof-isible definitions. The eorwlude (what we already virtually kllow
important point is that the particular is the case, of eourse) that Senator .Johnson
definition one chooHes will evidently have to has more power over Democratic Senators
merge from considerations of the subf-itanee with respect to federal aid to education than
and objectiveH of a Hpe(~ifie piece of reseaf(~h, I have.
and not from general theoretical conf-iidera- Earlier in defining the amount of power by
tions. the measure, i1f, I had already anticipated
Much the same argument applies to the this concluf-iion. What I have just said is
Comparing power:second variable. It is dear, I think, that we
precisely equivalent to saying that the power
comparable cannot compare A 'H pmyer with resped to
objects of power the respondentH ai, a2 ... an and E's power
of A with respeet to some set of ref-ipondents
and responses i" greater than the power of B
with respect to the respondents bl , U2 ••• lin
with resped to an equivalent set if and only
unlef-is we are prepared to regard the two
sets of individuals as comparable. This is a if the measure J1:[ assoeiated with A is
disagreeable requirement, but obviouf-ily a greater than the measure il[ associated with
sensible one. If I can induce 49 under- B. To recapitulate:
graduates to support or oppose federal aid
to education, you will scarcely regard thiH as
equivalent to the power I would have if I
;1[(: :w, :r) = PI - P2, where
could induce 49 Senators to support or PI = Pea, xlA, w)
oppose federal aid. Again, whether or not
the probability that a will do x, gIven
we wish to treat Senatorf-i af-i comparable to
aetion w by /1
students, rich men as comparable to poor
men, soldiers as comparable to civilians, pz = PCa, :ej.4,w)
enlisted men as comparable to officers,
military officen; as eomparable to civil ser- the probability that a will do x, gIven
vants, etc., iH a matter that can be de- no action w by A.
But this is not a very helpful definition. cause the use of undiseriminating questions
For the important question is whether we produced results of very limited theoretical
can specify some properties that will insure significance. By chom;ing a relatively weak
comparability among aetors, respondents, criterion of powm' comparability, the author
means, and scopeH. The answer, alas, is no. inevitably robbed his inquiry of much of
So far as an explication of the term "power" its potential richness. Considerations of
is concerned, power comparability must be comparability are, therefore, critic'al. But
taken as an undefined term. That is, power the criteria employed depend upon the
comparability will have to be interpreted in problem at hand and the general state of
the light of the speeifie requirements of relevant theory. The only way to avoid an
researeh and theory, in the same way that arbitrary and useless definition of "power
the decision as to whether to regard any comparability" i~ to eonsider carefully the
two objeets-animals, plants, atomH, or goals and substance of a particular piece of
whatnot-as comparable depends upon re~earch in view of the theoretical con-
general considerations of classification and structs one has in mind. ThUD in the ea:-;e of
theoretical import. To this extent, and to the Senate, it may be :-;atisfaetory for one
this extent only, the deeiHion is "arbitrary"; piece of research to define all Senate roll-call
but it is not more "arbitrary" than other votes on all issue" as comparable; for an-
decisions that establish the criteria for a other, only votee; on foreign policy iDsues
class of objects. will be comparable; and for still another,
To political scientists it might seem far- only votes on foreign policy issues involving
fetched to compare the power of a British large appropriations; ete. In a word, the
prime minister over tax legiHlation in the researcher himself must define what he
House of Commons with the power of the means by comparability and he must do so
President of the t;"nited States over foreign in view of the purpose of the ranking he is
policy decision:,; in the Senate. It would seem seeking to arrive at, the information avail-
farfetched because the theoretical ad- able, and the relevant theoretical constructs
vantage:,; of sueh a compari:,;on are not at governing the research.
all clear. On the other hand, it would not
seem quite :';0 farfetched to compare the APPLICATIONS OF THE CONCEPT OF
two institutional positions with re:,;pect to POWER COMPARABILITY
the "same" kind of polieY-Hay tax legis- Assuming that one has power compa-
lation or foreign policy; indeed, political rability, the next problem is b rank every
:,;cientists do make compari:,;om; of this aetor whose rank is relevant to the research.
kind. Yet the decision to regard tax legis- Here we run into practical problpllll' of
lation in the House of Commons as com- great magnitude.
parable in Home sense to tax legiHlation in Suppose we wish to rank a number of
the Senate is "arbitrary." Even the decision Senators ,yith respect to their infiuenc'e over
to treat as comparable two revenue measures the Senate 011 que:-;tic)]ls of foreign affairs.
pa:,;sed at different times in the United Specifically, the respondent and response
States Senatei:,; "arbitrary." 'Yhat :,;ave:,; a are defined as "all Senate roll-call voteH on
comparison from being genuinely arbitrary measures that have beell referred to the
is, in the end, its scientific utility. Some kinds Foreign Hdatiom; Committee." To begin
of comparisons will seem more artificial with, ]pt Uti take two Senators. What we
than others; Home will be theoretically more wish to find out is the relative infiuenee OIl
interesting and more productive than others. the Senate vote of the activities of the two
But these are criteria derived from theoreti- Senator" for or against a measure prior to
cal and empirical conHiderations independent the roll call. "For" and "against" must be
of the fundamental meaning of the term defined by referellee to some standard
power. "directioll." Passage of the measure is one
On what grounds, then, can one criticize possible "direction" in the sense that a
the study mentioned a moment ago? Be- Senator can he for passing the measure,
against it, or without a position for or against defeats the bills he works against; it is at a
passage. This is not, however, a particularly minimum of -1 when the Senate always
significant or meaningful direction, and one defeats the bills he works for and always
might wish to determine the direction of a passes the bills he works against; and it is
measure by reference to the President's at 0 when there is no change in the outcome,
position, or by content, or by some other no matter what he does.
standard. For this discussion, I shall assume In addition, there is one clear advantage
that "for" and "against" are defined by to M*. It is easily shown that it reduces to
reference to the first standard, i.e., passing
the measure. M* = PI - P2.
>,
Let us now assume that a Senator does
In a moment we shall see how advantageous
one of three things prior to a roll-call vote.
such a simple measure is.
He works for the measure, he works against
The theoretical problem, then, is clear-cut
it, or he does nothing. (The assumption,
and a solution seems reasonably well defined.
although a simplification of reality, is by no
It is at this point, however, that practical
means an unreasonable simplification). Let
research procedures begin to alter the
us further assume (what is generally true)
significance of a solution, for the particular
that the Senate either passes the measure or
operational means selected to breathe life
defeats it. With respect to a particular into the relatively simple formal concepts
Senator, we have the following conditional
outlined so far can produce rather different
probabilities: and even conflicting results.
The Senator Let me illustrate this point by drawing
>, Works Works Does on a paper by Dahl, March, and Nasatir (1)
For Against Nothing on influence ranking in the United States
Senate. The aim of the authors was to rank
Passes PI P2 pa thirty-four Senators according to their
The Se~~:eats -~=-;I-I-~=-;- -~=-;- influence on the Senate with respect to two
different areas, foreign policy and tax and
economic policy. The 34 Senators were all
Since the bottom row provides no addi- those who had held office continuously from
tional information we shall, in future, ignore early 1946 through late 1954, a long enough
it. Following the earlier discussion of the period, it was thought, to insure a reasonably
concept M, the measure of power, it is large number of roll-call votes. The class-
reasonable to define ification of measures to the two areas was
taken from the Congressional Quarterly
>, MI = PI - P3. Almanac, as were the votes themselves.
Thus the subject was well defined and the
lJ1 2 = P3 - P2.
necessary data were available.
MI is a measure of the Senator's power when No such systematic record is maintained
he works for a measure and lJ1 2 a measure of course, for the positions or activities of
of his power when he works against a meas- Senators prior to a roll-call vote, and what
ure; in both cases a comparison is made is more it would be exceptionally difficult
with how the Senate will act if the Senator to reconstruct the historical record even
does nothing. There are various ways in over one session, not to say over an eight-
which we might combine M 1 and M 2 into year period. Faced with this apparently
a single measure, but the most useful would insuperable obstacle, it was necessary to
appear to be simply the sum of MI and M 2 • adopt a rather drastic alternative, n'1mely
To avoid confusion with the earlier and to take the recorded roll-call vote of a
>,
slightly different measure which we are now Senator as an indication of his position and
approximating, let us call the sum of lJf l activities prior to the roll-call. While this
and M 2 , M*. Like M, it is at a maximum is not unreasonable, it does pose one major
of 1 when the Senate always passes the bills difficulty: a vote is necessarily cast either
a given Senator works for and always for or against a measure and hence the rol1=
call provides no way of determining when a With ;)4 Senators, i56l possible pairs of this
Senator does nothing prior to the roll-call. kind exist; but only 158 pain.; were tabulated
But the very essence of the formal concept for foreign policy and 20G for tax and
of power outlined earlier hinges on a com- economie policy over the whole period.
parison of the difference between what the The measure used to enable comparisons to
Senate will do when a Senator takes a given be made between the two Senators in each
position and what it does when he takes no pair might be regarded as an alternative to
position. M *. This measure-let us call it ill" -rests
It is at this point that the advantages of upon the same basic assumption, namely
the measure 1Y[* reveal themselves. For that we can measure a Senator's influence
provided only that one is prepared to take by the difference between the probability
the Senator's recorded vote as a fair indi- that the Senate will pass a measure the
cation of his prior position and activities, Senator opposes and the probability that
the data permit us to estimate the following it will pass a measure he supports. How-
probabilities, and hence M * ever, there are two important differences.
The Senator First, the authors deeided not to distinguish
Works Works between "negative" and "positive" power;
For Against consequently they used absolute values only.
Second, in estimating the probability of a
The Senate Passes P_l_-'--__P_'_...J
1_ _ measure passing the Senate, the positions of
two Senators were simultaneously compared
One could, therefore, estimate 111* for in the manner shown in the table. Thus the
each of the 34 Senators and rank all of them. influence of 8 1 over the Senate was measured
The validity of this method ranking as the difference between the probability
would appear to be greatest, however, when that a bill will pass the Senate when 8 1
all Senators are ranked on precisely the favors it and the probability that it will
same set of bills before the Senate. To the pass when 8 1 opposes it. However, this
extent that they vote on different (although difference in probabilities was measured
mostly overlapping) sets of bills, the com- twice: (1) when 8 2 favors the motions
parability of 1~[* from one Senator to an- before the Senate; and (2) when 8 2 opposes
other will be reduced, conceivably to the the motions. In the same way, 8 2 's influence
vanishing point. was memmred twice. Thus:
For a number of reasons, including a
slightly different interpretation of the M~ (8 1) = IPH - P121 ,
characteristics of an ideal measure, the that is, the change in probabilities,
authors chose a rather different approach. given 8 2 in favor of the bill.
They decided to pair every Senator against
every other Senator in the following way. 1Y[~ (8 1 ) = Ip21 - P2z1 ,
The number in each cell is an estimate of that is, the change in probabilities,
the probability that the Senate will pass a gIVen 8 2 in opposition to the bill.
proposal, given the positions of the two I,ikewise,
Senators as indicated; the number is in
fact the proportion of times that the Senate jll~(82) IPH - P211
passed a foreign policy (or tax) measure in .M~ (8 2) Ip12 - pni .
the period 1946-54, given the recorded
votes of the two Senators as indicated. The influence of 8 1 was said to be greater
Sl than the influence of 8 2 only if M~ (8 1) >
Favors Opposes iV!;' (8 2 ) and jvI~ (8 1 ) > jl[~ (8 2 ). That is, if
the the
motion motion IPn - P121 > IPH - P211 and
TABLE 3
THIRTY-FOUR U. S. SEXATORS CLAo;SIFIED ACCORDING TO 'POWER" OVER SE;\[ATE ])ECISIOX'; O;\[ FOR-
EIGN POLICY AND TAX POLICY, 1946-54
Foreign Policy
Tax and Economie Medium Smith (1\ ..J.)** Hickenlooper** East land
Policy influence Aiken* Knowland* H.ussell
Hayden Johnston Capehart*
Chavez Bridges*
Low Ferguson *
influence Wilev** Murray*
Hill *' Fulbright** Butler (Xebr.)tt
Magnuson Morse Langer*
Green** Kilgore
'.
Read at the annual meeting of the American 7. l\Tarch, .J. G . .:\Ieaslirement concepts in the
Political Science Association, Washington, theory of influence. J. Politics. (In press).
D. C. September 1956 (mimeo). 8. March, J. G. Influence measurement in experi-
2. Frendl, .J. H. P ..Jr. A formal (.heorv of social mental and scmi-expcrimental groups. Soci-
power. Psychoi. Rev., lB5(i, 6:3, Isi-Hl4. ollletry. H)56 , l!), 2tlO-271.
3. Lasswell, H. D., & Kaplan, A. Power and so- \J. Shapley, L. S. & Shubik, :\1. A method for
ciety.Xe\\' Haven: Yale Univ. Press, IH50. evaluating (he distribntion of power in a
4. Luce, R. D. Further comments on power dis- committee system. Amcr. pol. Sci. Rev.,
tribution for a stable (\"o-party Congress. 1954,48, 71l7-7\)2.
I!J5tl (September) (mimeo). 10. Simon, H. Notes on the observa.tion and mC~lH
5. Luce, R. D., & Rogow, A. A. A game theoretic urement of political power . .!. Politics, IH53,
analysis of Congressional power distribu- 15, 500-516.
tions for a stahle two-party system. Behav. 11. Weber,::\1. Wirtschaft lind Gesel/schafl. Tubin-
Sci., HJ5tl, I, 1l3-\)5. gen:.1. C. B. Mohr, H12.5, 2 vob. (Grllnririss
6. March, J. G. An introduction to the the or.,' del' Sozialekonofllik, Vol. :i).
and measurement. of influence. Amer. pol.
Sci. Rev., 1\)55, 5(), 431-451. (Manw.;cript received April 3, 1957.)
Get rid of the old liberals, then; get rid of the soldier in polities;
and put the ,,'orld into the hands of the seientists, the industrial
captains and the artists. For the new society was to be organized,
not, like Babeuf's, on the principle of equality, but aecording to a
hierarehy of merit. Saint-Simon divided mankind into three classes:
the savants, the propertied, and the unpropertied. The savants were
to exercise the "spiritual pmvcr" and to supply the personnel of the
supreme body, which was to be known as the Council of Newton-
since it had becn revealed to Saint-Simon in a vision that it was
Newton and not the Pope whom God had elected to sit beside Him
and to transmit to humanity His purposes. This council, according
to one of Saint-Simon's prospectuses, was to be made up of three
mathematieialls, three physicians, three chemishl, three physiol-
ogistf', three litterateurs, three painters and three musicians; and it
was to occupy itself with devising new inventions and works of art
for the general improvement of humanity, and in especial ,\'ith
discovering a new law of gravitation applicable to the behavior of
social bodies ,yhich would keep people in equilibrium with one
another. (So the eighteenth-century eommunist philmlOpher lVlorel-
let, in a book called The Code of Nature, had asserted that the law of
self-love ,yaH to play the same role in the moral tlphere as the law of
gravitation in the physical.) The salaries of the Council of Ke\yton
were to be paid by general subscription, because it waH obvioutlly
to everybody'i:i advantage that human destinies flhould be controlled
by mell of genius; the subscription would be international, beeaw.;e
it would of course 1)(' to the advantage of all peoples to prm'ent
inteI'llational warR.
-EDJ\n:ND WILSON, To The Finland Station