-
The AmericanPolitical Science Review
VOL. LXIV DECEN{BER, 19?O NO' 4
CONCEPT MISFORMATION IN COMPAR,ATI\/E POLITICS-
Grovlwrvr SlntontUnittersit ! al Fbrewe
anything to share with the cruc.irl concern of"methodology,"
which is a conccrn with the log-ical structtiii and procedure of
scientific en-quiry. In a very ouiial sense there is no
meth-o.totigy rvithotrt /ogot, without thinking aboutthinkiirg. And
if a flrm distinetion is drarvn-asii shouid be-between methodology
and tech-nique, the latter is no substitute for the former'One may
be a wonderful researcher and ma-nipulator of data, and yet rems-in
an uncgT-scious thinker. The view presented in this articleis,
then, tliat the profession as a whole is griev-ously impaired by
methodological unawareness.The more we advance teehnically, the
more rveleave a vast,, uncharted terrilory behind ourbacks. And mv
underlyiug cornplaint is that po-lrtieal scientists eminenlly lack
(wlth excep-tions) a training in logic-indeed in
elcmentarylogic.
i stress "elementary'' becc.use I do not rvish toencourage in
the least the overconscious thinket,the man rvho reluses to discuss
heat unless he isgiven a thermometer. My sympathy goes, in-stead.
to the t'conscious thinker," Lhe man whorealizes the limitations of
not having a thernromctr and still manag$ to say a great
dealsimplit by saying hot snd cold, rvarmer and
-
voL. 641034 rIIE AI{ERICAN POLI'I'IIIAI, SCIENCE REVIEW
of philosophers and political theorists. To gomeextint,
therefore, the traditional political scien-tist could afford to be
an "unconscious thinker"-the thinking had already been done for
him.This is .rr"r, ,ior. the case with the counlry-by-country
legalistic institutional approach, whichdoes not particularly
require hard thinking-2However, the new political science engages
in re-conceptualiza.tion. And this is even more thecase,
necessarily, rvith the new comparative ex-pansion of the
discipline.3 There are many rea-sons for Lhis renntatio ab
iYnis.
One is the very "expansion on politics." Tosome exlenc politics
results objectiue&1 bigger onaccount of the fact that the
rvorld is becoruingmore and more politicized (more
par[icipation,more mobilization, and in any case trlore
sts.teintervention in formerly non-governmentalspheres)- In no
small measure, however, politicsii rubjectiaehl bigger in that we
have shifted thefocus of attention both toward the periphery
ofDoiiti(B (vis-i,-vis the governmental process),and toward its
input side. By now-4s Macridisputs it-we sludy everything lhat is
"potentiallypolitical,"' While this latttr aspeel of the erpan-sion
of politics is disturbing-it ultirnately leadsto the disappearance
of politics-it is not a pe-ctrliar concern for comparative
politics, in thesense that other segments of political science
areequally end even more deeply affected.5
'This is by no means a criticism of a compara-tive item by itrm
analysis, and even less of the"institutionsl-functional" approach.
On lhe lattersee the judicious remarkl of Ralph
Braibanti,"Comparative Political Analytics Reconsidered,"Thz
Journal ol Politirs,30 (February 1968),44-49.
: For the various pl-rases of the comparative ap-proach see
Eckstein's perceptive "Introductioo," inH. Eckstein and D. E. Apter
(eds.), CotnpantivePoli,tirs (Glencoe: Free Pres, 1963).
"'Comparative Polities and the Study of Gov-emment: The Search
for Focu6," ComparaL{oePobti.r.t, (October 1968), p. 81.
'On the "fulltcy of inputism" see again the re-marks of Roy C.
Macridis, l,ae. cit., pp- 84-87. hhis words, "The state of the
discipline can besummed up in one phrase: the gradual
disappea,r-ance of the pol i t ical." (p.86). A cogent steteruentof
the issue is Glenn D. Paige, "The Rediscoveryof Pol i t ics," in J.
D. MonLgomery and W. I. Sifnn(eds.), Approachas to DeueloTtment
(New York:McGraw Hil l , 1966), p. 49 ff . My essay "From
theSociologt, of Politics to Political Sociology," inS- M. Lipset
(ed.), Politics and the Social Sciences(New York: Oxford Uoiversi
ly Prees, 1969), pp.65-100, is also largely concerned with the
fallacyof inputism viewed as a sociological reduction
ofpolitice.
Aside from the exparuion of politics, e mgrespecidc aource of
conteptual and methodologicalcia.llenge for comparative politics is
what
.Brai-banti ci l ls the "lengtheningspectrum of polit
icalslstems."o We are. now engaged ln lvorlo-wloe'cross-area.
comparisons. And while there is anend to geographical size, there
is appa-rently noend tn ihe"proliferation of political units'
Therewere abouiE0 States in 1946; it is no wild gueesthat we mty
shortly arrive at I50- Sti l l moreimportant, the lengthening
spectrum of politicalsystems includes a variety of prlmltlve,
dlfiusepoiiti"r at very different stages of differentiationand
consolidation.
Now. the wider the world under investigation,the more lve need
conceptual tools that are ableto travel. It is equally clear that
the pre-1950vocabula.rv of poliiics wa^s not devised for
world-wide, crois-arel travelling. On the other hand,and in spite
of bold attempts ai drastic termino-logiea.l innovabion,? it is
hard to see how West-ern scholars could radically depart from the
po-litical experienee of the West, i.e', from the vo-cahulary of
politics which has been developedover millennia on the basis of
such erperience.Therefore, the 6rst quesiion is: how far, andhow,
can we travel with the help of the availablevocabulary of
politics?
By and large, so far we have followed (moreor less unwitingly)
the line of least resistance:broaden the meaning-and thereby lhe
range ofapplication----of the conceptualizations at hand.That, is
to say, the larger the world, the more wehave resorted to concephnl
stretching, or con-ceptual straining, i.e., to vague, amorphous
con-ceptualizations. To be sure, there is more to it'One may add,
for instance, that, conceptualstretching also represents a
deliberate ettemptto make our conceptualizations value free.
An-other concurrent explication is that conceptualstraitring is
largely a "boomerang cffect" of thedeveloping areas, i.e., a
feedback on the Westerncategories of the diffuse polities of the
Third
o "Comparative Poli t ical Analyt ics Reconsid-
ered,," lbc. cil., pp. 3G37.'The works of !'red W. Riggs are
perhaps the
best instnnce of such bold attemr:ts. For a recentpreseniat ion
see "The Comparison of Whole Po-litical Systems," in R. T. Flolt
and J' E. Turner(eds.), ?lr.e MethodoLogY oJ ComponLiue
Reseorch(New York: Free Press, 19?0), esp, pp. 95-115.While Riggs'
j trnovative stretegy has undeniablepractical drarvbacks, the
criticism o{ Martin Lan-dau ("A General Commentary," in Ralph
BraLb-anfi (ed.), Politiral and Ad.mtnxtraLi'ue Deuelop'zrsnl
(Durham: Duke University Press, 1969), pp.325-334.) appeare
somewhat uqfait.
-
10351970 CONCEPT MISFORMATION IN COMPARATTVE POLITICS
trVorld.8 These c
-
r036
mue.h as five million.".1 I find the estimatefrightening, for
computer technology and facili-ties are bound to flood us with
masses of datefor which no human mind can have any sub-stantive
grasp. But even if one shares the Lnthu-siasm of Deutsch, it cannot
be denied that wehave here a gigantic, unprecedented problem.
In the third place, our predecessors were farfrom being as
unguided as we are. They did notleave the decision about what was
homogenorn-i.e., comparable-and what was heterogen-ous-i.e.,
non-comparable-to each man's genialinsights. As indicaied by the
terminology,-theircomparisons *pplied to things belonging to
"thestme gentw." Thtt is to say, the br.ckground ofcomparabilitSr
rvas established by Lhe pcr genwet difierentiara mcde of analysis,
i.e., by ^.tuxe-nomical treatnxent. In this context,
comparablemcans something *'hich belongs to the same ge-nus,
species, or sub-species-in short to the samecla-qs. I{ence che
clags provides the "similarity el-ement" of comparability, while
the "difrerences,,enter as the species of a genus, or the
sub-spe-eies of r species-and so forth, depending onhow fine the
anrlysis needs lo be. However, andhere is the rrrb, the taxonomieal
requisites ofcomparability are currently neglected, if not
dis-orvned.
We are nolv better equippcd for a discussionof our initiil
query, namel),, why the travellingproblem of cornparative politics
has been meiwith the poor remedy cf "conceptual strelching,'instead
of being squarely conf ronted. Whilethere are mflny reasons for our
neglect to attackthe problem frontallv, a major reason is that
wehave been swayed by the suggesiion that ourdiltculties can be
overcome by switching from"what is" questions to "how much"
questions.The atgument runs, roughly, as follows. As longas
concepts point to differences of kind, i.e., aslong as we purslre
the either-or mode of analysis,we are in trouble; but if concepts
are under-stood es a matter of more-or-less, i.e,, as point-ing to
differencer in degree, then our difficultiescan be solved by
measurement, and the realproblem is precisely how to mea6ure.
Meanwhile-rvait.ing for the measur$-class concepts andtaxonomies
should be looked upon with suspi-cicn (if not rejected), since they
represeuL "8.nold fashioned logic of properties and attributesnot
rvell adapted tn study quantities and rela-t ions."12
'"Recent Trends in Researclr Methods," in J. C.Charlesworth
(cd.), ,4 Desiqn lor Political Science:Scope, Ob jeclitsct and, tr{
ethods (Philadclphia:American Academy of Political and .Social
Science,1966), p. 156.
" Carl F. Hetnpel, quoted in Don Mart indale,"Sociological
Theory and the Ideal Type," in
vol. 64
According to my previous analysis, a ta\o-nomic unfolding
represenis a requisite conditionfor compara-bility, and indeed a
backgroundwhich becomes all the more important the lesswe can rely
on a substantive familiarity withwhat is being cornpared. According
to the fore-going argument, instead, quantilication has noills of
its own; rather, it provides a remedy forthe ills and inadequacies
of the per genu eL dil^lerentiom mode of analysis. My own view is
thatrvhen we dismiss the so-called "old fashionedlogic" we are
plain lvrong, and indeed the victimsof poor logic-a vierv thet I
must now attemptto warrant.
II. QUANTIFIC.LTION AND CL.{SStFIcATrONlVhat is verJ, confus'ng
in this matter is the
abuse of a quantitative idiorn which is nothingbut an idiom. All
too often, that is, we speak ofdegrees and of measurement "not
on'ly rvilhoutllny actual measurements having been per-formed, but
rvithor.rt any being projected, andeven rrithout anv epparent
arvarenegs of whatmrrst be done before such rneasurements can
becarried out."13 For instance, in most standardter-tbooks one
finds tha.t nominal scales are spo-ken of ae "scales of
measurement.''r4 Bub a. nom-inal scale is nothing else than a
qualitative clas-sification, and I fail to undersland rvhat it
isthat a nominal scale does, or can, measure- Tobe sure clnsses can
be given numbers; but this issimplv a coding device for identifying
items andhas nothing to dc with qunntification. Likewiscthe
incessant use of "it is a matter of degree"phraseology and of the
"continuum" image leaveus with qualitrtive-impressionistic
statementsrvhich do not advance us by a hair's breadth to-rvard
quantification- In a. similar vein we speakmore and more of
"yariables" which are notvariables in any proper sense, for they
are notattributes permitting gradations and implyingQross,
Syraposium on Sociologital Theory, p. 87.Marl indale aptly commenb
that "Hempel 's judg-.nents are made from the standpoiat of the
oaturalsciences." But the vein is not digsimilar when
thestatiatically trained scholar argues that "whereasit, is
dmittedly teehnically possible to think al-wrye in terms of
attribules and dichotoroies, oneu'ondels lrow pracl ical that is":
Hubert M. Bla-lock, Jr., Cav.tal Inleretces in
NonerperimenLalResearch (Chapel Hi l l : Univelsi iy of North
Caro-I ina Pre :s , 1964, p .33) .
"Abraham Kaplan, The Cond.uct ol Inquiry(San Francisco:
Chaudler, tS64), n. 213.
" Eg., L. Festingar and D. Xatz (.eds.), ResearchMetlrcds in the
Behauil trai ^ Sciarrces (New York:Dr; 'den Press, 1953) ; and Sell
t iz, Jahodz et aL,Research Methnds in Social Relatians (rev.
ed.,Neq, York: I{olt , Rinehart & Winston, 1959).
THR AITERICAN POLITIC.4, I , SCIENCI} NEVTCW
-
rs70
measurabilit]'- l.io harm necessarily follows if itpleases us to
use the rvord varirr,ble as a syn-onym for the rvord concept; buf
n'e are onll' ds-luding ourselves if rve realiy believe that by
ac,r1-ing variable we Jnue a variable.
All in all, ccquetting (if not cheatrng) with aquarrtitative
idiom grossly exaggerates the ex-tent to rvhich political science
is curlently amen-able tn quantifi.cation, and, still worae,
obfus-cxtes the very notion of quantification. The dividrng line
betrveen the jargcn and the substanceof quantilication can be drawn
very simpli,:quirntification begins with numbers, and wher.numbers
are used in relation to their arithmeti-cal properties. To
understand, however, the mul-tifaceted complexities of the notion
beyond thrsdividing line is a far less simple matter- Never-theless
one may usefullt' distinguish-in spite ofthe close
interconnections-among three broad&reas 0f meaning and
application, tltat is, be-tween quantifi.cation as i) measurement,
ii) sta-ttstical manipulution and, iii) formal mathemat-ieal
treatment.
In political science we generally refer to thefirst merr-ning.
That is to sry, far mcre oftentlian not the quantification of
political scienceconsists of (a) attaching numerical vslues toitems
(pure and simple measurement), (b) ue-ing numbers to indic&t
the rank order of items(ordinal scales) and (c) measuring
differencesor distances among items (interval scales).15
Beyond the stage of measurement we do own,in adaition,
poru&ruf
"t"ti.ii.rf l".rrriq".r- r.jonly for protecting ourselves
against samplingand measurement errorsr but also for establish-ing
signiflcant relationships among r.ariablesHowever, statistical
processing enters the sceneonly when sumcient numbers have been
pinnedon suffi.cient items, and becomes central tn thediscipline
only when we drspose of yariablesrvhich measure things that are
worth measuring,Both conditions-and especially the latter-are
'There ls som question as to whether it canreally be held that
ordinal scales are scales ofmeasuremnt,: moel of ottr rank ordering
occurstr i thout haviug recourse to numerical values, andshenever
r+'e do aasign number.s to our orderedeategories, these numbers are
arbitrary. However,there are good reasons for drawing the
thresholdof quautificatron between nominal and ordinalscales rather
Lhan betwecn ordinal and intervo"lscales (See Edward R. Tufte,
"Improving DataAnrlysis in Political Science," World Politirs,
2l(July 1969), esp. p. 645.) On the other hand, evenif the gap
belween ordinel scales and intervalmeasuremnt is not as wide in
practice as i f is inthqory, nonetheless from a mathematical point
ofr.iew the interesting scale-s are the interval andeven rnore, of
coursp-j the eardinal scale-s.
1037
hard to meet.lc Indeed, a cross-e\alnination ofour statistieal
findings in terms of their theoreti-cal significanee-and/or of a
"more releva.nt"political science--horvs an imnressive
dispro-portion between bravura and relel,ance. Unfor-tuna.telyl
rvhat rnakes a ststistical treatmenttheoretie-a.Lly significant has
nothing to do withstatistics.
As for the ultimate slage of quantlfication-{ormal mathematical
treatment-it is a factthat, so far, political science and
mathematicsluve engaged only "in a sporadic conversa-tion."r? It is
equally a fact that rve seldom, ifever, oblain isomorphic
correspondences betweenempirical relatiorrs alnong things and
formal rela-tiors a.mong tnrmbers.rs 'V[e may rvell disagreeabout
future prospects,le or as to whether it
'o Otherwise the compa,rative method wouldlargely consisU of the
atatistical method, for thelaitor surely is a stronger techuique of
control thenthe forrner. Tlie differeuce and the coqnections
srecogentlv discussed by Lijphart, "Comparative Pol-itics and and
the Comparative Method," op. cit.
"Oliver Eenscn, "The Methematieal Approachto Polltical Science,"
in J. C. Charlesworth (ed.),Conlemporary I'olitr,al Annlusis (New
York: FreePress, 1067), p- 132. The chapter usefully reviewsthe
literature. For an introductory treatment eeeIlt.3rvard R. Alker,
Jr., MalhemaLiis snd, PoliJ;rfoa(New York: Macmil lan, 1965). An i
l luminatingdiscuggion on how quanlification enters the
varioussocial sciences is in Daniel Leper (ed.), Qu"antityond,
Quality (Glencoe: Free Press, 196I), passim.
" A classie example is the (partial) mathematicaltranslation of
the tlteoretical system ol The Human Group of George C. Ilomaus by
llerbert A-Simon, i{od.:& ol Man (New York; Witey, t967),Chap.
7. No similar achievement exists in the po-Iitical science 6.eld.
To cite three aigniGcant in-stances, politieal science issuc,s are
eminentl.y lack-iug in Kenneth J. Arrow, "Mathematical Modelsin the
Social Sciences," in D. Lerner and I[. D.Lasswell (eds.), T'he
Polby Scizncec (Stanford:Stanford Unir.ersit5r Press, 1951), Chap.
8; in thecontributions collected in P. F. Laaarsfeld (ed.)
,Alathetnatical T'hinking in the Social Sclaru:es(Glencoe: Free
Press, 1954); in J. G. Kemeny andJ. L. Snetl, Mathem.atical Modek
in tha SocialSciences (Bostort: Ginn, 1962).
rc Perhaps the mathematical leap of the disci-pliue is just
around the corner waiting for non-quanti tat ive developments. I f
one is to judge, how-ever, from tlie "mathematics of mau" isue of
theInternoti.oral Social Scizrue Butbtin introducedby Claude
Levi-Strauss (IV, tS54), this l i teratureis very deceir ' ing.
More interesting is John C.Kemeuy, "Mathematics rvithout Numbers,"
inLerner, QunnLita ond QualiLgt, pp^ 35-51; and themodal logic
developed by the Bourbaki group.
CONCEPT MTSFOR\{A'TION IN CO}' IPARATIVE POLITICS
-
THE AMFRICAN POT,ITICAL SCTENCE REV(EW vo],. 64
makes sense to construct formalized systemsof quanlitatively
well defined relationships(mathematical models) so long as we
wander ina mist of qualitatively ill-dedned concepts- If weare to
learn, however, from the mathemaLicaldevelopment of economies, the
evidence is thatit "always lagged behind its qualitative and
con-ceptual improvement."'o And my point is, pre-cisely, that this
is not a ca$ual sequence. It is fora very good reason that the
progress of quantifi'cation ihould lag-in whatever
discipline-be-hind its qualitative and conceptuel progress-
In this messy controversy about quantifca-tion and its bearing
0n standard logical rules wesimply tend to forget that concept
lortnationstands prior to gwntificaLion. The proces$ ofthinking
inevitably hegins with a qualitative(natural) language, no matter
at which shore weshall subsequently iand. Correlatively, there isno
ultimate way of bypassing the fact that hu-mal understandin6-the
way in which our mindworks-requires cut-off points which
basicallycorrespond (in spite of all subsequent refine-ments) to
the slices into which a netural orqualitative language hrppens t'o
be divided.
There is a fantestic lack of perspective in theargument that
these cut-off points cen be ob-tained via statistical processing,
i.e., by lettingthe data themselves tell us where to draw them.For
this argument applies only uithin the frameof coneeptual ma.ppings
which have to tell us frstof what reality is composed. Let it be
stressed,therefore, that long before having data whichcan speak for
themselves the fundament^rl artic-uLation of ianguage and of
thinking is obtainedLogically-by cumulative conceptual
refinementand chains of coordinated defi.nitions-not bymeasurement,
Measurement of rvhat? We cannotmeasure unless we know first what it
is that rveare measuring. Nor can the degrees of somethingtell us
what a thing is. As Lazarsfeld and Bartonneatly phrase it, "before
we can investigate thepresence or absence o[ some attribute . . .
or be-fore we ca,n rank objects or measure them in
1038
E6menk de MathdmaLique, a.ppearing periodi-cal l .y (Paris:
Hermann). For a general trertmentsee J. G. Xemeny, J. L. Snell, Q.
L. Thompson,Introduction to Finite Mathematizs (EnglewoodClifs:
Prentice Halt, 1957).
'Joseph J. Speagler, "Quantifrcation iu Eco-nomics: Its
flis[ory," in L4rqer, QtnntiLy and'Qulita, p. 176. Spengler equally
points out that"the iuLrcduction of quantitative methodt
iueconomics did not re-:ult in striking discoveries"(i6rd.). While
formal ee.onomic theory is by nowhighly isomorphic with algebra,
matliematicaLeconomics has added liitle to the predictive powerof
the discipline and one oflen has fhe impressionthat we are
employing guns to kill mosquitos.
terms of some variable' u)e ftrust lorm the con-cept ol that
uafiahle."lr
The major premise rs, then, that quantifica-tion enterg the
scene after, and only after, hav-ing formed the concept. The minor
premise isthat the "stuff" of guantification-the thingsunderpinned
by the numbers-cannot be pro-vided by quantification itself. Hence
the rules ofconcept formation are independent of, and cax-not be
derived ftom, the rules which govern thetreatment of quantitiff and
quantilative rela-tions. Let us elaborate on this conclusion.
In the flrst place, if we never rer-lly have"how much"
findings-in the sense that the priorquestion elways is how much itt
whot, in whzlconceptual container-it follows from this thathow much
quantilative flndings ere ax intrnalelernent of "whz.t is"
qualitative questionsi theclaim that the latter should give way to
thefcrmer cannot be susiained. It equally (ollows,in the second
place, that "categoric concepts" ofthe eithcr-or type cannot give
way to "grada-tion concepts" of the more-than-les6-then type.
What is usually lost sight of is that lhe either-or type of
logic is the t.ery logic of classifi.cationbuilding. Classes are
required to be mutually ex-clusive, i.e., class concepts represent
characteris-tics which the object under consideration musleither
have or lack. Two items being comparedmust belong flrst to the same
class, end eitherhave or not have an attribute; a.nd only if
theyhave it, the two items can be matched in termsof rvhich has iL
m.ore or lz$s. Flence the lcgic ofgradation belongs to the logic of
classification.More precisely put, the switch from classifica-tion
to gradation basically consists of replacingthe signs
"same-different" rvith the signs "8ame-greater-lesser," i.".,
consists of introducing aquantitative di.ffermtiation within a
qualitativesa.rneness (of attributes). Clearly, then, the
sign"s!me" established by ihe logic of classificationis the
requisite condition of introducing thesigns "plus-minus."
The retort tends to be that this is true onl.v aslong as we
persist in thinking in terms of zLLri-butes and dichotomies. But
this rejoinder missesthe point theL-aside from classifl,ing-we
dis-pose of no other unfoLding tecfrnique. Indeed,the taxonomical
exercise "unpa,cks" concepls.and plays a non-replaceable role in
the processof thinking in that it decomposes mental com-pounds into
crderly and managea.ble sets ofcomponent units. Let it be added
that at nostage of the methodological argument does thetaxonomical
unpacking lose weight and impor-
" "Qualiletive Measurement in the Social Sci-ences:
Classifrcaticns, Typologies and Indices," inD. Lerner and E. D.
Lasswell (eds.), Thz PalicaSciznces, op. cit., p. 155 (my
emphasis).
-
19?0
tance. As u mztLer of fact, the more we enter thestage of
quantification, the more we need unidi-mensionel scales and
continua,; and dichotomouscategorizations serve precisely the
purpose of es-tablishing the ends, and thereby the
uni-dimen-siona"lity, of each continuum,
Having disposed of the fuzziness broughtabout by the abuse of a
quantitative idiom, at-tention should immediattly be called to the
fact-finding side of the coin. For my emphasis onconcepL formation
shouid not be misunderstoodto imply that my concern is more
theoreticalthan empirical. Thfu is not so, because the con-cepts of
&ny social science are not only the ele-ments of a. theoretical
systeln; they are equally,and just as much, data containers. Indeed
datais information which is distributed in, and pro-cesged by,
"conceptual containers." And since thenon-experimental scienees
basically depend onfact-finding, i.e., on reports about external
(notlabora.tory) observables, the empirieal questionbecomes what
turns a concept into a valuable,indeed a valid, fac,t finding
container.
The reply need not be far-fetched: the lowerthe discriminating
power of a conceptual con-tainer, the more the facts are
misgathered, i-e.,the greater the misinforma(ion. Conversely,
thehigher the djscrirninating power of a category,the better the
iniormation. Admittedly, in andby itseU this reply is not very
illuminating, forit only conveys the suggestion that for
fact-find-ing purposes it is morc prodtable to exaggeratein
over-differentiation than in over-assimilaiion.The point is,
however, that what establishes, orhelps eatablish, the
diecrim.iuating power of acategory is the taxonomieal infolding.
Since thelogical requirement of a clqqqification is that itsclasses
should be mutually exclusive and jointlyexhaustive, it follows from
this that the taxo-nomical exercise supplies an orderly series
ofwell sharpened categories, and thereby the basisfor collecting
adequately precise information.And this is indeed how we know
whether, and towhat extent, s. concept has a fact-gathering
va-lidity-
Once again, then, it appears that we havestarted to run before
having learned how towalk. Numbers mtst be attached-for our
pur-poges-to "things," to facts. How are thesethings, or facts,
identifled and collected? Ourultimate ambition rnay well be to pass
from a-science "of species" to a science of
"functionalco-relations."lt The question is whether we ztenot
repudiating a science of speciea in exchangefor nothing. And it
eeems to me that prematureha.ste combined with the abuse of a
quantitative
t-Earold D. Lasswell and Abraham Kaplan,Power and, SocieLu (New
Ilaven: Yale UniveraityPress, 1950), pp. XVI-XYII.
I039
idiom is largely responsible not only for the faclthat much of
our theorizing is muddled, but alsofor the fact that much of our
research is irivialand wasteful-
Graduate students ere being sent all over theworld-as
LaPalorcbara vividly puts it-an "in-diecriminate fishing
expeditions fot data.-"r3These fishing expeditions are
"indiscriminate"in that they lack taxonomical backing; which isthe
same as saying that they are fishing expedi-tions without adequate
nets. The researcher eetsout with a "che.cklist" which is, at bast,
an im-perfect net of his own. This xnay be an expedientway of
handling his private research problems,but remains a very
inconvenient strategy fromthe angle of the additivity and the
comparabil-ity of his findings. As a rcsult, the joint enter-prise
of comparative politics is menaced by agrowing potpourri of
disparate, non-cumulativeand-in the aggregate--misleading morass
ofinformation.
AII in all, and regardless of whether we relyon quantitative
da-ta or on more qualitative in-formaticn, in any case the problem
is the same,namelv, to construct fact-finding categories thatown
suffi.cient discriminating porver.2* If ourdata containers are
blurred, we never know towhal extent and on what grounds the
"unlike" ismade "alike." If 3o, quantitative analysis maywell
provide more misinformation than qualita-tive analysis, especially
on account of the aggre-vating circumstance that quantitative
misinfor-mation can be used without any substantiveknowledge of the
phenomena under consider-ation.
To recapitulate and conclude, I ha.ve arguedthat the logic of
either-or cannot be replacedby the logic of more-and-1ess. Actually
the twologics are complementary, and each has a legiti-mate fi.eld
of application. Correlatively, polaroppositions and dichotomous
conftontationscannot be dismissed: they are a necessary stepin the
process of concept formation. Equally,impatience with
classifica,tion is totaLly unjusti-fied. Rather, we often confuse a
mere enumere-
" "Macrotheories and Microapplicetions in Com-parative
Politics," C otnporatiu e Pobtics, (Octobert968), p. 66.
aft hardly needs to be emphasized that censusdala-and for that
matter mosL of lhe data pro-vided by exiprnal ageucies-are gathered
by con-ceptual containers which hopelessly la-ck discrimi-nation.
The question with our standard variebleson literacy, urbanization,
occupation, industrializa-tion, and the like, is whether they
really me&surecommon underlying phenomena. I t is prelty
ob-vioue that, across Lhe world, they do nctl and thisquite aside
from tJre reliability of Che darr gather-ing agencies.
CONCEPT MISFORMATION IN COMPARAqIT'O POI,ITICS
-
1040 THE A}INRIC^N POI- ITICAI. SCIENCT] RI | \ : I I i lV vol,,
64
tion (or c.hecklist) with a classifi.catjon, andmany so called
classifications fail to meet theminimal requirements for what they
claim to be.
The overconscious thinher takes the view tharif the study of
politics has to be a ,,science,,then it has-to be Nervton (or from
Newton allthe.rvay.up to Flempel). But the experimentalmethod is
hardly within the reach of politiealscience (beyond the format of
small grcup ex-perimentation) and the very extent to which
weare
.systemetically turning to the comparativemethod of verification
points Lo the ertent towhich no stronger me(hod-including the
statis-tical method-is a"vailable If so, oui distinctiveand major
problerrs begin rvhere the lesson ofthe more exact sciences leaves
off. This is tanta_mount to saying that a wholesale acceptance
ofthe logic and methodology of physics may rvellbe self-defeating,
and is lurelrr oi little uie forour distinctive needs- In
particrrlo"r, and what-ever their limits, classifications rcmain
the rea_uisite, if preliminary, condition for any
scientificdrscoune. As Hempel himself concedes, classifi_ca"tory
concepts do lend themselves to the de_seription of observational
findings and to theformulation of initial, if crude, empirieal
gener-alizations.l5
.Moreover, a classifrcaio ry r&,irrityremains the basic
instrument for introducing an-alytical
_clarity in whatever we are discu&ing,and leads us to
discuss one thing at a time anddifferent things at different
tim&. Finally, andespecially, we need taxonornical networki
forsolving_ our fact-fi.nding and fact-storing prob-lems. No
comparative science of pclitics iI it"u.-ible-cn a global
scale-unless we can d.raw onextensive inlormation which is
sufficiently pre_cise
.tn.be meaningfully compar ed. The requisitecondition for this
is rn adequate, relatively sta_ff9 ana, thereby,. oddittue
fiIing'qstem. Slch antlng system no lcnger is a wild dream,
thanksto computer technology and facilities---excentfor Lhe
paradoxical fact that the morc rve enterthe computer age, the lass
our fact-findrns andfact-storing methods abide bv anv
Iosicallvstandardized criterion. Therefore, m., con.e.nwith
taxonomies is also s concern n,ith t) thedata side of the question,
and 2) our failure toprovide a 6ling sydtem for computer
exnloita_tion. We haue enLered the computer ^ g;-butwith feet of
clay.
III. THE L-{DDER OF ABSTRACTIONIf qua"ntification eannoL solve
cur problems, in
that we eannot measure before conientual izine.rnd i f , on the
o ther hand, , , conqqp ix i l s t r . t c l i -ing" i s dangerous
ly conduc ive to the Hcge l ian
a Pundamenlats ol Coruept Formatinn in Em_pirical Science
(Chicagor Universit,v of ChicagoPrere, 1952), p. 54.
night in which all the cows look black (andeventually the
milkman is taken for a cow),then the issue must be joined from its
very be-ginning, that is, on the grounds of concepi for-mation.
A ferv preliminary cautions should be entered.Things conceived
or meaningfully perceived, i.e.,concepts, are the central elements
of proposi_tions, and--depending on how they are,r""rea-provide in
and by themselves Euidelines of inter-pretation and observation.
It-should be under-stood, therefore, that I shall implicitlv refer
tothe conceptual element problems which in amore extended treatment
actually and properlybelong to the rubric "propositions.', By
siyingconcept formation I impiicitly point to
"
nrono-sition-forming and problem-"olning activlty. ftshould also
be understood, in the
-second pl"ce,
that my focus will be on those concents ivhicha.re. c.rucia.l to
the discipline, that is, the conceptswhich Bendix describes as
,.generalizations indisguise.'tre In the third place,l !ropose to
con-centrate on the vertical members of i coneeptualstrrrcture,
that is, on l) obreruationol tcrms', tnd,2),the vertical
dispositicn of such terms along aladder ol absLrodion.
While the notion of abstrattion ladder is re_lated to the
problem of the levels of anal.vsis, thetwo-things do not coincide.
A highly absjractlevel of analysis may not result
'froin ,,ladder
climbing. Indeed a number of universal conceD_tualizations are
not abstracted from observa-bles: they are "theoretical terms,,
defined bvtheir systemic meaning.2T For instance th;meaning of
isorno rphism, homeostasis, feedback.entrophy, etc., is basically
defined bv the parithat each concept plays in the whole theo.y.
Inother instances, however, we deal rvith ,.observa_tional terms,"
that is. rve arrive at highly ab_
N Reinhard Fendix, ' ,Concepts and Generaliza_tions in
Comparative Sociolcgical Studies,', Amert_can Socinlngbal Reuieu,2g
(f9ffi), p. 533.
"'See Abraham Kaplan, The Conduct ol InquirE,pp. 56-57, 63-65.
According to Hempel theoreticalCerms "usually purport to nof
directlv observableentities and their cheracteristics.
. . - Th"v fun"-tioa . , . in eeientific theoriea inteaded to
exolaingeneralizatiooe,': .,The Theoretician's Dilemma.,'in Feigl,
Scriven and Maxwell (eda.), MinnedoittStudins in tha Philosophy o!
Science (Minnean_olie: University of Minnesota prese, lgsg). v;1.I
I , p. a2. While ir ie admittedly dif f icult to drawa neal
division between theoreticel and obeerva.tioaal terme, it is widely
recognized that the for_mer cannot bo reduced to, nor darived from.
theIatter. For a recent asse.ssment of the controversv.dee A- Meot
t i , , ,L 'E l im inaz ione de i Termin iTeorici ," in Riuista di
Fitotof ia,2 i lg6S), pp. 119_131.
-
1970
stract levela of conceptualization via ladderclimbing, via
abstractive inferences from observ-ables- For instance, terms such
as group, com-munication, conflict, and decision can either beused
in a very abstract or in a very concretemeaningr either in some
very distant relation toobservebles or with reference to direct
observa-tiorrs. In this ca.se we have, then, "empiricel con-cepts"
which can be located etr and movecalong, very di.fferen[ points of
a ladder of ab-stracfion. If so, we have the problem of
assessingthe level of abstraction at which observationalor (in this
sense) empirical concepts are located,and the rules of
transformation thus resulting.And (his seems to be the pertinent
focus for theisue under cousideration, for our fundamentalproblem
is how to make exterrsional gains (byclimbing the a"bstraction
ladder) without havingto suffer unnecessaly losses in precision and
em-pirical testability.
The problem can be neatly underpinned withreference to the
distinction, and relation, be-tweea the eztensi,on (denotation) and
intetxion(connobation) crf a term. A standard definition isas
follows: "The extension of a ryord is the classof things to which
the rvord applies; the inten-sion of a word is the collection of
propertieowhich determine the things to whieh the wordapplies."'8
Likewise, the denotation of a word isthe totality of objects
indicated by that word;and the connotation is tlre totality of
character-istics anything must possess to be in the denota.-tion of
that word.2s
Now, Chere are apparently two ways of climb-ing a ladder of
abstraction. One is to broadenthe extension of a concept by
diminishing ie at-tributas or properties, i.e., by reducing its
con-notation. In this case a more "general," or moreinclusive,
concept can be obtained without anyloss of precision. The larger
the class, the lesserits difterentia.e; but those differentiae that
re-main, remain precise. Moreover, following thisprocedure we
obtain conceptualizations whic,h,no ma0ter how all-embracing, still
bear a trace-able relation to a collection cf specifics,
&nd-outof being amenable to identiflable sets of specffica-lend
themselves to empirical testing.
On the other hand, this is hardly the proce-dure implied by
"concepLual stretching," whichadds up r,o being 2r attempt to
augment the ex-
d I quote from \4'esley C. Selmon, Logic (Engle-rvood Cliffs:
Preutiee-Hall, 1963), pp, 90-91. Thedislinction is more or lesa lhe
seme rD eny texi-book of logic.
t"'Connotalion'' is also applied, more broadly,to the
aasociat,ions, or agsociated conceptionsbrought to mind by the use
of a word. As iudi-cated by the text, I intend here the
narrowermean lng .
1041
tension without diminishing the intension: tledenotation it
ectended, bg obfu'scating the con-nntation. As a result we do not
obttin & moregeneral concept, but its counterfeit, a mere
gen-erality (where ihe pejorative "mere" is meant torestore the
distinction between correct and in-eorrect ways of subsuming a term
under abroader genus.) While a general concept can besaid to
represent a colleetion of speciics, a meregenerality cannot be
underpinned, out of its in-definiteness, by speciOcs. And while a
generalconcept is conducive to scientific "generzliza-tions," mere
generalities are eonducive only tovs,gueness and conceptual
obecurity.
The rulet for climbing and descending along aladder of
abstraction are thus very simple tules-in principle. We make &
concept more abstractand more general by lessening its properties
orattributee. Conversely, a concept is specified bythe a.ddition
(or unfolding) of qualificacions, i.e.,by augrnenting ils
attributes or properties. If so,let us pass on to consider a ladder
of abstractionas such. It is self-evident that along the
abstrac-tion ladder one obtains very different degrees
ofinclusiveness &nd, ccnversely, specffi city. Thesedifferences
can be usefully underpinned-for thepurposes of ccmparative
politics-by dietin-guishing three levels of abstraction, labeled,
inshorthand, HL (high level), ML (mediumlevel), and LL (low
level).
High level categorizations obtain universalconceptualizations:
whatever connotetion is gac-rificed to the requirement of global
denotation-either in spare, time, or even both,so HL con-cepts ean
also be visualized as the ultimate ge-nus which cancels all its
species. Dese,ending astep, medium level categorizations fall short
ofuniversalrty and thus can be aaid to obtain gen-eral classes: at
this level not all differentiae aresacrificed to exlensional
requirements. Noneihe-less, ML concepts are intended to stress
similad-tiee at the expense of uniqueness, for at thisleve1 of
abstraction we are typically dealing withgenera-lizations.
F'inally, low level categoriee ob-tain specific, indeed
configurative conceptualiza-tions: here denotaiion is sacridced to
accuracyof connotation. One may equally say that withLL categories
the differentiae of individual set-tings are stressed above their
similarities: somuch so that at this level definitione are
oftencontextual.
A couple of examples may be usefully entered,In a perceptive
essay which runs parallel to my
sThe epace and time dimeosions of conceptsare often aseociatd
with the geography versushislory debate. I would ra"ther see it as
the "whengocs with when?" question, that is, as a calendarLime
versus hit torical t i rne di lemma, But this l ineof development
cannot be pursued here
CONCEPT MTSFORMATION IN CO}{P,\RA'I'IVE POL(TICS
-
1042 THB AMEBICAN POLI,TICAL SCIENCE REVIEW vol,. 6{line of
thinking Neil J. Smelser makes the pointthat, for purposes of
comparability, .,staff ismore satisfactory than administra.tion . .
., andadministration is more satifactory than civii ser-vjgg."ar
This is so, according to Smeleer, becausethe concept of civil
service,.is literally uselqs inconnection with societies without a
formal stateor governmental apparatue.,, In this respect
t,theconcept of administration is somewhat auperior. . . but even
this term ia quite culture-bound.,,flence the more helpful term is
,,Weber's con-cept of stefr . . . since it can encompass
withoutembarassment various political alrangements. . -"!2 In my
own terms the argument would berephrased a.s follows. In the field
of so-celledcomparative public administration,,.staff" isthe high
level universal category. ,.Administra-tion" is still a good
travelling category, but, fallsshort of universal applicebility in
itrai-it retainssome of the attributes associa.ted with the
moreepecific notion of "bureaucracy.,t Descendins theladder of
abstraction further we then nnd tivitservice," which is qualified
by its a.saociationswith the modern State. Finally, and to
pursuethe argument all the way down to the low levelof abstraction,
a comparative study of, say,French and English siate employ..s wif
ais_cove:-their unique and distinguishing trai& andwould thus
provide contextual definiiions.
The example ouggested by Smelser is fortu-nate in that we are
oftered a choice of terms, aothat (whatever rhe choice) a different
levej ofabstraction can be identified by a d.ifferent
de_nomination. The nert sxqmple il illustrative, in_stead, of the
far less fortun4te situation in which*: T"y have to perform across
the whole ladd.eroI abstractlon wrth one and sa.me term. In
illua_trating his caution that many concepts are ..sen-eralizations
in disguise," Bendix .o-e rrios.such a_ simple concept as
,,village.,' yet he uotesthat.the term..village may be lnisleading
whenapplied to Indian society, where .,the rnini*umdegree of
cohesion commonly associated withthis term is absent.,'t3 Even in
such a simnlecase, then, a scholar is required tn plzce the
var-ious associations of ,.village,, along an abstrac_tion ladder
in accord with the trair"tting e;en_sion afrorded by each
connotation.-
Clearly, there is no hatd and fast divid.ine linebetween levels
of abstraction. Borders on-*tube drawn very lcosely; and the ou-n.,
of Ui.&
a "Notes on the Methodology of Comperative
Analysie of Economic Aclivity,', Transactions 6lthe Sizlh Woild,
Congress of Sociolnsy, 196i, Inter_national Sociological
Associalion, vol_ II, p. 103.e lbid,
"Bendix, "Concepts and Generalizationg. .,rtp . s 6 .
into which the ladder is divided largely dependeon how fine
one'B analysis needs to be. threeslices are suffcient, however, for
the purposes oflogical analysis. And my major enncern is, inthis
connection, with wha.t goes on at the upperend of the ladder, at
the crucial juncture- atwhich we cross the border between mediurn
levelgeneral concepts and high level universals. Theissue me.y- be
formulated as followsl how fer upcan an observational term be
pushed withoutself-denying results?
In principle the extension of a concept shouldnot be broadened
beyond the point at which atleast one relatively precise
connotetion (prop-erty or attribute) is retained. In practice,
how-ever, the requirement of positive iderrtificationmay be too
exacting. BuL even if no minimalpositive identification can be
afforded, I do notgee how we ca"n tenounce the requirement
ofnegative identif.cation. The crucial distinctionwouid thus be
between 1) concepts defined bynegatioa or ee ddercrao, i.e., by
saying what theyxte ,Lot, g.nd 2) concepts withouL nigaLion,
i.e.,no-opposite concepts, conceptions with out speci_fied
termination or boundaries. The loqical nrin-ciple involved in this
distinction is oinis i,eter-fr"lio est negatio, tha.t is, any
determinationlnvolves a negation. According to thia principlethe
former concepts ere, nc matter how Iroad,detertninate; whereas the
latter are indetermi_nate, Iiterally without terminntion.
If this principle ia applied to the climbincprocess along a
ladder oi abstraction. u"A-oilcisely
-to the pcint at which ML categories'areturned into IIL
universala, in the 6rsi instancewe ob-tain empidcal uniuersols,
whereas in thesecond instance we obtain universals which
lackempirical v a"Iue-pseudo-utitersak for an emnir_ical science.
The reason for this is that a .oo.lptqualffied by a negation may,
or may not, 6eIound to
-apply to the real world; wherea"s Inon-bounded concept always
applies by defini_tlon: hevrng no specified termination, there is
noway of ascertaining whether it applies to thereal world ar not.
An empirical universel is su"f,because it still points Lo
somnthing; *t".""-*non-empirical universal indiscriminately
pointsto eue_rEthin4 (as any researcher on it
"
nAasoon drccovers),.
The.group concept tends itself nicely as an il-lustration of tbe
foregoing (other exomnler willbe diecussed in greafer detail
later), and is uerumuch to the polnt in that it represents the
firsflarge scale attempt to meet the travelline Drob-lem of
comparative polit ics. In the qrouplh"o-of politics (Bentley, David
Trumrn,
"na ea.tLathan being the obvioue references) it is clear
enough that "group,' becomes an ali_embracinec&tegory: not
only an analytical construct (as tbi
-
ls70 CONCEPT MISFORMATION IN COMPARATIVE POLITICS 1043
distinguish such obieet from all the other speciasof the sarne
genus. When Apter complains ihaiour "analytieal categories are too
general whenthey are theoretical, and too descriptive wherethey are
notr"ss I understand thie complaint toapply io our disorderly leaps
from observationalfindings all the way up to universal
categories-and vice versr--by-passing as it were the stageof
definition by analysis. Apter is quite right tnpleading for "belter
intermediate ana.lytical cat-egories." But these intermediate
categories czn-not be constructed, I fear, as long as our
con-tetnpt for the taxonomieal exercise leeves uawith an atrophied
medium level of abstraxtion.
The low level of abstraction may appear unin-teresting to the
comparative scholar. He wouldbo wrong, however, on two counts.
First, whenthe comparative scholar is engaged in field work,the
more his fact-finding categories are broughtdcwn to this level, ihe
better his research. Sec-ond, it is the evidence obtained
nation-by-nation,or region-by-region (or whatever the unit of
an-alysis may be) that helps us decide which classi-fication works.
or which uew criterion of classifi-cation ahould be developed.
Virhile classifying must abide by logical rules,logic has
nothing to do wilh the usefulness of aclassificatory system.
Botanists, rnineralogistsand zoologists have not created their
taxonomi-cal trees as a matter of mere logica.l unfolding;tha.i is,
they have not imposed their ''classes"upon their animals, any more
than their animals(flowers or minerals) have irnposed
themselvesupon their classifiers. Let ii be added that
theinformaticn requirements of such an unsettledscience as a
science of politics can hardly be sat-isfied by single-purpose
cla-ssificaticns (not tomention single-purpose checklists). As I
havestressed, we desperately need standard fact-find-ing and
fact-storing ccntainers (concepts). Butthis standardizaLion is
onl-v pcssible and fruitfulon the basis of "multi-purpose" and, at
thelimit, all-purpose classifications, Now, whether aclassification
may serve multiple purposes, andwhich classification fits this
requirement best,this is something we discover inductively, thatis,
starting from the bottom of the ladder of ab-sfrzction.
The overall discussion is rec*pitulated in Ta-ble I with respect
to its bearing on the proble-msof comparative poiitics. A few
additional com-menti are in order. In the f,rst place, rcferenceto
three levels of absLraction brings out the in-adequacy of merely
distingulshing between
s David E- Apter, "Political Studiea a:rd theSearch for a
Framework," (pp. 15-16 mns.) to bepublislred in C. Allen, W.
Johnson (eds.), AlricnnPercpectioes, Cambridge University
Press.
queer and unclear terminclogy of the disciplinewould have it),
but definitely a universal con-struct. However, we ole never really
told whatgroup is zol. Not only "group" eppliet
.euery-wher:e, as any universal should; it equally ap-plies to
euergthin4, that is, never and nowhereihali rue encounter
non-groups.3' If so, how
'ls itthat rhe group theory of politics has been fol-lowed-in
the fifties-by a great deal of empiri-cal research? The reply is
that the rescarch wasnot guided by the universal construct but by
in-tuitive concrele conceptualizations' I{ence the"inde6.nite
group" of the theory, and the "con-crete groupi" of the research,
fall wide epartThe unfortunate c.onsequences are not only thatthe
research lacks theoretical berking (for wantof medium level
categories, end especially of ataxonomic frameworlc) , but tiat the
vaguenessof the theory has no flt for the specificity of
thefindings, We are thus left with a body of lit'era-ture that
gives the frustrating feeling of disman-tling theoretically
whatever it discovere empiri-cally.
There is, then, a break-off point in the searchfor universal
inc.lusiveness beyond which wehave, theoretically, a "nullification
of the prob-lem" and, empirically, what may be called an"empirical
vaporization." This is the point atwhich a concept is not even
determind ec ad-terso.By saying that no-opposite universels areof
no empirical use I do uot imply that they areutterly useless. But t
do wish to say that wlien-ever notions such as groups or-a.s in my
subse-quent examples-pluralism, integration, partici-pation, and
mobilization, obtain no tcrmination,i.e., remaiu indeterrninate,
they provide onlytags, chapter headings, i.e., the main entries of
a,filing system. From an empirical point of viewpseudo-universals
are only funnels of approachand can only perform, so to speak, an
allusivefunction.
Turning to the middle slice-the fat slice ofthe medium level
categories-it wiil sufnce tonote that at this level we are required
to per-fcrm the rvhole set of operations that some art-thors call
"deflnition by anall'sis," that is, theprocess of defining a term
by finding the genusto which the object designated by the word
be-longs, and then specifying the attrlbutes which
" This criticism is perhaps unfair Lo David Tru-mzn'a The
Gauernmenlal Process (New York:Knorrf, 1951). f lowever, in spite
of its penetratinganetomy Lhe pace of Lhe enquiry is set by the
sen-tence thaL "an excessive preoccupation with defini-t ion wil l
only prove a handicap" (p. 23)- For z-developmenC of this l ine of
ct i t ic iem eee G. Sartori ,"Gruppi di Preesione o Gruppi di
Interease?," 1lMulino, 1959, pp. 7-42.
-
r044 THE ^MERICAN POLITICAL SCTENCE REVIEW
TTTALU I . L^DTJER Or. ABSTR^CTION
vol,. 64
Maximll extensionI{ inimal intensionDednit ion by
negaiionBalsnce of denotation with con-no ta,tto rll . tef ini t
iorr !v enalysis, i .e. pergenus e t d tne fen t lam
L e v e | s o ( A b s t r a c t i o n | } { ' l , - " : 9 " * 1
: a r a t i v e S c o p e a n < . l j p " * @__l Pu.oo""_ l
F."-pi,iti"J
"r c;;iil;;
IIL: H,ig.h LeuI CategoriestJ nrvereal con ceD tual izr-t
ions
ML: Medilnt Letel CategoriesGeneral
-concep tual lza-t lcng snd t,axonomles
LL; Lou Leuel CaleoorieaCondgurative "concep-tual iz,at ions
Cross-ares comparisons a.monglifi:i5""t"8 contexts
(globalfntra-area comparisons anrongrelat ive ly homoqeneous
con-texts (middle range theory)
Country by country analysis(na.rrow-geuge theory) Vl lximal
intensionIr{ inimal extensionContevtual dednit ion
"broad" and "nanol" meanings of a term.36For this does not
clarif1,, whenevlr lhis is neces-sary, whether rve distinguish, l)
between FLuniversal and IUL general conceptualizations, or2)
between ML genuses and specias or, 3) be-trveen ML and LL
categories, or even 4) be-tween HL universal and LL configurative
con_ceptualizations.
In the second place, and more important, ref_erence to tbe
ladder of abstraction f6rcibly high_lights the drastic loss of
logical articulation,'in-deed the gig_antic leap, implied by the
argumeniLhAr oll differences Lre .'a- matter of dicree."This cannot
be conceded, to begin with, a"t thelevel of universal categories.
But all difierencescannot be considered a matter of more_or_less
atthe medium level either. At the top we inevita-bly begin rvirh
opposite pairs, rvitli polar oppo-sites, and this is tantamount to
saying tha[ ihelqq Ml, categories definitely and
onhiestablishdifferences in kind. From here dorvnwards defi-nttiong
are obtained via the logic of classifica-tion, and this impties
that a logic of gra-dadoncannot be applied as long as we-establLh
differ-ences between species. Diferences in degree ob_tain only
after having established thaitrvo ormore objects have the same
attributes or prop_erties, i.e., ielong to the same species.
tndeed, itis only uithtn the same class thit we are entitied-arrd
indeed required-to ask which object hasmore or less of r.n
attribute or propertv..
h principle, then, it is s. fallaci to apply thelogic of
gradation r.vhenever ladder clim'bini 1o.descending) is involved.
If we are reminded-thatalong the ladder we augment the extension
bvdiminishing the denotaiion (and vi"e u".s"i,what is at stake here
is the presence or absenceof a given property; and this is nol a
matter of
-The same caution applies to the distinctionsbetween micro and
macro, or beLween molecularend molar. Thee distinctions are
iosufficient forthe purpose of rrndcrpinning the level of
analysis.
degree, but a mabter of establishing the level ofabstraction.
Ifence it is only afte; having set-tled at a given level of
abstraction that coniid.er-ations of more-and-less correctly apply.
And therule of thumb seems to be that the hiEher thelevel of
abstraction, the tess a. degree l"ogurg"applies (as anything but a
metaphor); whereasthe lower level of abstraction, the more a
degreeoptics correctly and necessa_rily applies, and themore we
proft from graduation concepts.
In the third place, and equally important, ref-erence to the
ladder of abstraction casts manydoubts on the optimistic
view-largely shareiby the methodological literature-lthat .Themore
universal a_ proposition, i.e., the greater theuumber of events a,
proposition ac"ouots for, themore potential falsifers ca"n be
found, and themore informative is the proposition.,,r? The
sen-tence suggestE a simultaneous and sometvhat nat-ural
progression of universality, falsifiers and in-formative content.
It seems to me, instead. thatreference to the correct technique of
tadderclimbing (and descending) confronts us at allpoints wilh
choosing between range of explana-tion (thereby including the
explanation of rherelationships among the iGms under
investisa-tion), and accuracy of description (or informa-tive
accuracy). By saying thit the ,,informztivecolrtent" of a
proposition grows by climbing the,rbstraction ladder, we should nol
be misleJintounderstanding Lhat rve are supplying more
de.-qeriptir.e information. Hence
-
ii is dubiousivfgt-her lve are really supplying more
potentialfalsifiers (let alone the dangei oi,,overly univer-sal"
propositions o( no informative value forwhich (alsifiers cannot be
found).
Before coneluding it should not pass unno-o'I quote Erik
Allardt, ,,The Merger of Ameri-
can and European Trrditione of Soeiological Re-e earch:
Contextual Analysis,,, Social Sci^ence In-lormalinn, I (1368), p.
l6E. But the sentence isil lustrative of a current mood.
-
1970 CONCEPT M(FTORTTATION JN CO.\ IPAR.{ , ITVI i POI,T ' I '
ICS r045ticed that in this section I have never used thervord
"variable,', nor mentioned operational de-finitions, nor invoked
indicators. Equall5,, myreference to gradation concepts rnd to
.onrid.._atione of more-or-lesn has been, so far,
entirelypre-quantitative. ltrIhat is noteworthy, then, iithe length
that hss been trar,elied beior. enier_ing the. problems which seem
to monopolize ourmethodological awerenees. There ij nothin,,t
l9lg,to besure, in taking up oo r.gu*uni-riwhichever point rue feel
that wle have
-somedrir.,sto.say-except that t l ie tail of the
rnethodoloqilcal argument should not be mistaken for its
b-e_ginrfng. Since I have trken up the i".u" al ..,S.."rlV qt"g.,
I-cannot possibly carr-v it through torL6 end. lf behooves me,
nonetheless, to indicatehow. I would plug what I have said into
u,hatshall have to remain unsaid.3t-
Fgqotl. thing, it should be understood that bycongldnng
concepts-the genus_I have not ex_cluded the consideration of
variables, which area species. That is, a variable is still a
concent.but a. concept is not nece$arily a variabl.. ii^rllconcepts
could be turned into variables, the dif-ference could be considered
provisionai. ilf*-tunately, as a scholar tvell veised in
quantitativeanalysis puts il,
_,,all the most interesll"g ""riAblpq x1s nominal."3o \Vhich is
the same;. ;ittg thaf al1 the most interesting eoncepts ,i
"itr-ariables.in,the proper, strictiense o? impg,lng"the
possibil i ty of measurement in the mosi ex_act senge of the
rvord.,,.o
A closell, Iinked end similar lrgument :rnnliesto the
operationist requirement.
.lust ^ , .o,i..ui"are not necessarily variables, dcfinitions
are notnecessarily operational. The defirrit ional require_ment.
for. a concept is that its meaning is de_clared, while operational
definitions are requiredto state the conditions, indeed the
operations, bymcans of which a concept c,an be
^uerifted antd,,
ultimately, measured. Accordinglv we may use-fully distinguish
betrveen definiiion of mi"ninnand operationai definit ion. And
rvhile it i .
"n"iorrs that, an operational definition still is a
decla_ra.tion oI meanirrg, the reverse is uot true_
"In thi.s latter connecticn an exeellent readerst i l l is P. l
. Lazarsfeld alrd M. Rosenberg (eds.),The Larquage ol Socia|
Re.search (Cl"oeie, ,IheFree Press, l05S), See also its largel-v
revised ancupdated revision, R. Boudon and p. F.
Lazarsfeld,Md,tlndes de ta Sociologiz, 2 yols. (paris rrnd LaHaye ;
Mouton, lg65-1966).
" Richard Rose, . ,Social Measure and publiePolicy in Bri
tain-The Empir icizing pr.ocess,,, mns.p . 8 .
* Lazarsfeld and Barton in Lerner and Lasswell .The Palicy
Scienres, p. l?0. This noiabl-,- excludes,for the authors, the
alrpl icat ion of . ,variable', toi lems that cen be r lnke d but
not mea,"ured
The contntion often is that dednition ofrneaning
.represents a pre-scientific age of de.fini_tion, which should
be superseded t scientificdiscourse by operationat de6nitions.
However,this contention can hardly meet the proble-ms ofconcept
formation, and indeed appeals to ignorethern. As the Iedder of
abstraction scheme irelpsto underline, a-mong the many possible
,u,u,s
"ndprocedures of defning Lhe ez aduerso definitionsand taxonomic
un(oldings (or definition bv anal_ysis) some correepond to
different levels of anal_ysis and play, at each level, a
non-renlaceablerole...Moreover opera-tional defr ni tions
generallyentail a drastic curtailment of meaninglor thcycan only
maintain those meanings that complyq'ith the operationist
requiremeni. Now. rue areslrely required to reduce ambiguitl, b.y
euttingdown the range of mea.nings of conccpts, Bur theoperationrl
criterion of reducing tmbiguil.y en_toils drastic losses in
conceDtuel richness and inexplanetory po$:e(. Take, ftr instance,
the sug_gestion that "social class, should be dismisseaand replaced
by a set of operational statemen(srelating to income, occupation,
edueational level,etc. If the suggestion were adopted wholesa)e,the
loss of conceptual substance rvould. be notonly considerable, but
unjustified. The same ap_plies, to cite another irntanc.e, to
.,power.,,To beconcerned rvith the measurement of porver
doesnot
-imply thab the meaning of the conceptshould be reduced to q.hat
can be *.".ui.aabout power-the latter view ,.vould make hrr_nran
behavior in rvhatever collective spherc al_most inexplicable.
It should be understood, therefore. that oner-ational
.
d^efinif ions implement, btrt do not re_nlace, dehnltlons of
meaning. Indeed lhere mr.rslbe a conceptualizaLion befoie we engaq
in oner_ationalization. As Hempel recomm-en-d., oolo_tional
definitions should not be ,.emphasized loLhe neglect of the
requirement of sysiematic im_port."rr This is also to say that
definit ions ofryean_ing of theoretical import, hardly
opera_tion-al defi.nitions, account ior itre dynamil ofintellecbual
disco'ery and stimrrlation. r.Uattv itshould be urrderstood that
empirical testine"oc_curs before, and a lso rv i thou{ . , o l
tcrat ronal d i6ni_tious.. ' lcsting.is an..- method oi .t l".Ling
co...-spondence with reality b;-the use of perti ient
cb_servatrons; hence the decisive difference brought
^\ I,'unAarncnLals ol Conccpt Formation in Enrpirital $cie.nce,
p. 60. At p. 47 Hempcl writes: ,,itis preci.*ely the discor,ery of
concep(s with theorel_ical impor'u rvhich adr-ancas scicntific
understand_ing; and such discover,y reqrrirc.s scientif ic
inven-tiveness and ceunot be replaced by
the_certain15,indispensnble, but cl-.o dr:6nitcl-r,
insufficient_operat ionis t or emprr ic is t l .cqui remcnt of
emnir ica iimpo r f a l one . "
'
-
1046 THE AMERICAN POLITICAL SCIENCE BEVIEW vol,. 64
about by operationalization is verification, orfalsifi cation,
by measuretnent.oz
Speaking of testing, indicators are indeed pre-cious "testing
helpers." 3s a matter of fact it isdifficult to see how theoretieal
terms could beempiricized and tested otherwise, that is, with-out
having recourse to indieators. Indicators arealso expedient
shortcuts for the empirica.l check-ing of obsetvational terms. Yet
the question re-mains: Indicalors of whzt? If. we have
fuzzyconcepts, the fuzziness will remain as it ie. Thatis to say
that indicators cannot, in and by them-selves, sharpen our concepts
and relieve us fromcomposing aud decomposing them along a ladderof
abstraction.
rV. COMPARATIVE FALLACIES: AN ILLUSTRATION
We may norv confront in more detail how theladder of abstraction
scheme brings out thesneres and the fauLts of our current way of
han-dling the travelling problern of comparative poti-tics. For we
m^ay now settle at a less rarifledlevel of discussion and proceed
on the basis ofexamples. It is pretty obvious that my line
ofanalysis Largely cuts across the various theoriesand schools that
propose themselves for adop-tion in comparative politics, for my
basic prenc-cupation is with the ongoing work of the "nor-mal
science," i.e., with the common conceptualproblems of the
discipline. Nonetheless it will beuseful to enter here a somewhat
self-enntainedillustration which bears not only on
discreteconcepts, but equelly on a theoretical frame-rvork. I have
thus selected for my first de[aileqdiscttssion the categories of
"gtructure" and"funclion," end this precisely on account oftheir
crucial role in establishing the struciurat-functional approach in
the political science set-ting.a3
In introducing his pioneering comparative vol-ume, Almond boldly
aseerts: "What we have
"This is not to say tha-t operationalization al-lows co ipso for
quantitative mcasurements, bulto suggesh that either operational
defrnitions areultimately conducive lo meaaurement, or may notbe
worthwhile.
'r I specify political scicnce aetting to avoid theunnecessary
regression to Malinowski and Radcliff-Brown. This is also Lo
explain why I set aside thecontributions of TalcoLb Parsons and of
Marion J.Levy. Flanigan and Fogelmou distinguish beLweenlhree major
6treams, labeled 1) eclectic function.ahsm, 2) empir ical
functional ism (Me.rton), and3) sLructura-l-funclional anaLysis.
("FunctionalAn alysis, " in Ch arleswo r Lh, C ont empo r ary P o
Liti'calAnalgs;s, pp. 72-75). My discusaion exclusively ap-pl ier
to parl of the latber.
done is lo separale political function from politi-cal
structure."l{ This separaticn is indeed cru-ciaL. But ten years
have gone by and the assign-ment remeins la.rgely unfulfilted.
Indeed thestructural-functional school of thought is
stillgrappling-with clear s_vmptoms o f f rust ration-with the
prelininary difficulty of defi.ning "func-tion"-bolh taken by
itself and in its relation to"sf,ructu(e."rB
Whether function ca.n be simply conceived asan "activity"
performed by structures; orrvhelher it is more proper io construe
functionas an "effect";t6 or whether function should beconceived
only as a "relation" among structures{?-this controversy turns out
to be largely im-material in the light of our substantive
perfor-mance. That is to say, if our attention turns tothe
functional vocabulary in actual use, a pe-rusal of the literature
ouiclclv revea.ls two
sGabriel A- Almond and James S, Coleman,The Politics ol Lhe
Deuelnping Areas (Pr'inceton:Princeton University Press, 1960),
p.59.
s It, ahould be understood that by now lhe
struc-tural-fuoctional label applies to a widely scatteredgroup
operaling on premises which are largely atvarieuce.
{This focus was suggested by R. K. Merton,whose concer!. was to
separate function-defined asan "observable obj ective
cousequeice"-f rom "sub-jective dispo.sitiou," i.e., aims, motivcs
and purposee(Social Theory and Social Struclure, Glencoe ; TheFree
Presc, rev. ed., 1957, p. 24 tnd,passim., pp. l9-81.) In attempting
to meet the difficulties raisedby the Mertouian focus, Robert T.
Eolt construesfunclions as "sub-lypes" of effecta, and preciselyas
the "eystem-relevant eflecls o{ sLructures";utderetanding
system-televaoce a6 the "system-requirednees" which ic determined,
in turn, by the"fuuctional requisites" of a given system- ("A
Pro-posed Stnrctural-Functioual tr'ramework," inCharlesworth,
ConLemporary Pohtical Alo,IUsk,pp. 88-90). My own position is that
Merton ovr-s[ated his case thereby creating for his
followersunnecssary and unsettled complicationa.
" This is the mathematical meaning of f rrnction.E.C- according
to Fred \{. Riggs in systems theoryfunction refers to "a telation
between struc-tures." ("Some Problems wit.h Systems Theorl'-The
lmportance of Slructure," mimeographed p.8. A redrafted version is
scheduled for publicationin Michael Ha,as and Elenry Kariel (eds.),
,{.p-proaches to the Study ol Political Science, (Chaud-ler
Publishing Co,) There are problems, however,ako with this
dcfinition. Iu particular, while themathematical meaning of
function is srrited forwhole rystems aoalyais, iL hardly suits the
needsof segmented systems analysis.
-
1970 CONCEPT MTSFORMATION IN COMPARATIVE POI,ITICS LA47ihin6s: a
Lantalizing anarchy (on this morelater)
, and, second, that the functional terminol_ogy emp-loyed most
of the time by most practi-tioners definitely carrieg a purposive
or teimlogi-crl connotr.tion, Skillful verbal camouff.age maywell
push the teleologicel implication "in thibar,kground. Yet it is
fiard io frnd a functionelargumentation which really escapes, in
the finalanalysis, ZweckratinnaJitiit, whzt Max Webercelled
rationality of ends.a8 We may well quarrelabout the definition;ne
yet the sudstance of tt"malter remains that the definitional
controversyhat little bearing on our subsequent proceedings.rI 80,
lt sults my purposas to settle for the wayin which most people uee
,,function,, in practic-e(regardless of how they theorize about
it), andthereby to settle for tbe corrmon eer,ru,'unro_phisticated
meaning.
'When we say, somewhat naively, that struc_
tureg "have functions," \r'e ere interested in the1gaso1 fo.r
being of structures: we are implying,Ln4t 16, that struetures exist
/or some end, pur_pose, de.stination or assignment.so This is
ianta-mount to saying that ,,function,' noints to
amea.ns-end.relationship (which becomes, from asystemic viewpoint,
also a part-whole ielation_ghip),- i.e., that function G the
activity per_formed by a structure-the means-vis-i-vis its
{ Rationality of ends should not be confusedwilh
WertratinnaliLiit, value rationality, amongother reasons because in
the former perspective allconeeivable ends can be hypothesized as
beurtequally valid. Elence id Lhe Zueckratbnalit
-
t048 T}IE A\{EEICAN POLITICAL $CIENCE REV(E\Y \:oL. 64
tional purposes. And here we enter a, sornewhatvieious rvhirl
which leads the appros.ch io con-clusions which, if true, wouLd be
self-denying.
Whatever else the structural-functionalscholar may have failed
to discover, he feelspretty sure about three points: first, no
st,ruc-ture is uni(unctional, i.e., performs only onefunction;
second, the same structure can bemultifunctional, i.e., can perform
aeross di.fferentcounlries rvidely different functions; third,
andtherefore, the same function has structural al-ternatives, i.e.,
can be performed by very differ-ent structures. Now, to some extent
these pointsere undeniable--but only to the extent sensd atany time
by any perceptive comparr.tive scholarMy quarrel is with the
emphasis, which is un-wArranted and positively misleading.
Is it realLv the same 8tructure that functionsdifterently? Or is
the functional performancedifferent because the structure is not
the sarne?The thesis generally lacks adequate evidence onthe
structural side. For instance, "elections" aremultifunctional (they
may well serve the pur-pose of legitimizing a despot), but
"freeelections" are not.63 That is to say! as soon esthe electoral
process obtains o structural under-pinning-the minute and multiple
structuralconditions that make for free
voting-eLectoralmultifunctionality rapidly comes to an end. Ifthe
voter rs offered dternatives, if the candi-dates are f ree to
compete, if fraudulent eounlingrs impossible, then free elections
do serve^---ev-erywhere--the purpose of allowing an electorateto
select and dismiss office holders. In view ofthis primary,
fundamental purpose Lhe mtneelectoral structure (same in providing
all thenep,essary safeties) either approachee uni-func-tionality,
or leaves us with non-functionality,e,g., rvith the finding that
illitpra"te voters areunable to use electoml mechanisms rvhich
pre-suppose literacy.
While the most serious problem and default isthal the structures
are inadequately pinpointedand described, let me hasten to add that
lve arenot perforrning much better from the functionslend of the
argument. For our functional catego-ries algo generally lrck
adequate underpinning.Surprisingiy enough-if one considers the
fargrezLer ease rvith which the functional side of[he problern can
be attacked-orrr functionstend lo be as unhelpful as our
structures.
For instance, if one asks, "Why a party sys-tem?" the least
challengeable and most incluaivereply might be that parties perform
e communi-
" I cite the tit le of W. J. M. MaeKenzje's bookFree
Ele.cti,orn, (London: Allen & Unwin, 1958)to imply lhet a real
structura) underpinning maywell presuppose a hundred-page
descriptioo,
cation function. And if the problem is lefL aithat, it easily
follows that the authorities andthe citizens "communicate," in some
sense, inany polity, i,e., even when no party system ex-ists.
Ilence palty systems have structural alter-nLtilres-*qrLod erat
d,em,onstra:ndum. But theproblem cannot be left at that, i.e., with
4n un-bounded, no-difference notion of communicationrvhich
nullifies the problem. And the underpin-ning of communication
brings otLt, first, thatthere is an essential difference betrveen
up-goingand descending commr-rnication, and, second,thai it is
equally important to distingttish be-trveen "communication-in f
ormation" and "com-munication-pressure." If so, to de6ne a
partysystent es an instrument for "communicating"demands and
conveying "information" to theauthorities, is to miss the point. A
party systemis, in reality, a mechanism for su-etoining
de-mands-and pressin4 demands-all the waythrough to policy
implementation. What is atstake, then, is the passage from a
two-way (re-versible) communication-information to a pre-valence of
up-going communica.tion-firessure.And for this Iatter purpose we
have not devisedso far, any structural alternative. A pa,rtv
svstemturns out to be, therefore, a. non-replaceable,unique
strurture as soon as we spell out its dis-tinctive, crucial reason
for being.
A more careful scrutiny goes to show, then,that the multi-f
unctiona L, multi-structurel argu-ment has been pushed far too far,
indeed to thepoint cf becoming erroneous. Aside from the er-ror,
the irony of the situation is that, as itstands, the thesis appears
self-defeating, If thesame 6tructure perforns utterlv different
func-tions in different countries, and if we can always6nd
structural alternatives for rvhatever [unc-tion, what is the use of
structural-functionalanalysis ?
Pulling the threads together, I need not spendmuch time in
arguing that the sta"lemate a"nd themishandlings of the
structural-functional ap-proach have a lot to do with the ladder of
ab-straction.
On the functional side of the coin we are en-cumbered by a.
rvealth of haphazard functionalcatcgcries which are merely
enumerated (hardlyclassified a.ccording to some criterion, and
evenless according to the logical requirements of ataxonomical
tree-type unfolding) , and defi.niteli'provide no clues as to the
level and type of angl-ysis (e.9., total versus partial systems
analysis)to which they apply.s' As r result the global
" A sheer liet oI the fuuctioual denominatione,roles or
attributions Bcsttered throughout thelilerature on political
parties suffices to illustratethe point, and would be as follows:
participation,
-
t970 CONCEPT I I ISFOIiMATION IN COTVTpARATIVE POLITICS
r049functional argument developed by a number
ofstructural-functionalists remains suspended inmid-a.ir-for lack
of a coordinated medium leveltaxonomic support-and is left to play
rvith ov-erstretched, if not contentless, functiona.l
univer-sals.
On t.he structural side of the coin we a.reconfronted, instead,
with little more than noth-ing. Structures qualified on their own
righthardly exist-a.t least in the Almond line ofthinking.ro This
is all the more regrettable inview of the fact that rvhile
functions are meantto be (at least in global coreparative
polities)broad explanetory c&tegories whieh do not re-quire a
low level specifi.cation, structures bear,instead, a. closer
relation to observablee, and def-initely need under-pinning all the
way dorvn theladder, With structures understood as organiza-tional
sttucturs we are required, in fact, to de-scend the ladder all the
way down to low levelcontgu ra,tive-descriptive accounts,_
Starting from the top, one can identify-withthe help of minor
terminological deviees.-atleestr Lrr different tevels of analysis:
1) struc-tura"l principles (e.g., pluralism), 2)
structuralconditions (e.g., the class or the economic struc-ture),
3) organizational patterns (with relationto membership systems), 4)
specidc orgzniza-tjonal structures (e.g. eonstitutions).
Bylavins"structural principles" I mean that a.s ^ "iic.^Legary the
notion of structure can onlv nointto the principles according to
which the comno-nent part of polities, or of societies, are
relaiedto each other. With reference, instead, to the lorvlevel of
abstraction it should be cleai that con-
elecLioneering, mobilization, ex lraction, regula iion,conuol,
integration, cohesive function, moderatiugfuoction, consetrsus
maintenance, simplidcaUou ofalternatives, reconciliation,
edaptetion, aggrega-tion, mediatiou, couflict resolutiou,
brokerage, re-cruitment, policy making, etpression, commuuica-lion,
linkage, channelmert, conversion, legitimizingfunction,
demoeratization, labelliug function.
*I make specific reference to Almond becauee Ibelieve that hjs
very conception of structure islargely respousible for this
outcome. Fot iuslance,"By structure lye maD the observable
activitieswhich make up the pofitical system. To refer toLhese
activitiee as having a structure simply im-plies that there is a
certain regularity to them.,,(Almoud and Powell, Comparatiue
politics: A De-uelnpntental Approach, Bostton: Little, Brown,
lg6,p. 2l). In the eubsequent paragraph one readsr"lVe refer to
perticular eetr of ro]es whir:h ar.e re-lated to one another as
stnrctures.,, Uncler suchporoua and excessively sociclogical cr i
ter ia, . .struc-ture" becomes et'rneacent_
stitutions and statutes are not the "re&l struc-ture.
Nonetheless behavior under wrltten rules iseasier to pin down than
behavior under diffuseroles, and excessive anti-formalism leads us
tnneglect organizational iheory and the extent towhich. legally
enforced regulations do moldDenavl0r.
In surnrning up, not only has the structural-functional scholar
ignored the ladder of abstrac-tion, but he has inadvertently
destroyed, duringhis reekless climbing, his own ladder.tc So
mucliso that the approach encounters exactly thesame perplesity as,
say, general systems theory,namely, "Why has no scholar succeeded
in pre-senting I structural -functional formulation wfiichmeets the
requirements of empirical analysis."rrNow, it is hardly surprising
tha.t the geneial svs-tems theorist should encounter great
djfficuliiesin dcriving testable propositions about politics,since
he is required to proceed deductively onthe basis of theoreticel
primitives.ds But this isnot the cose with the
structural-functional an_prorch, which is not necessarily
cornmitted iorvhole sl,sfqms analysis and enjoys the distinc-tive
empirical advantage of leaning extensively-especiallj, with
segmented systems a.nalysis-onobserva.tional terms.so So, why
should the struc-
'This complaiut ie ad hoc, but could be ex-panded at length. On
the general lack of logicalaud methodological etatus of the
approach twostrong critical atatements are: R. E. Dowse.
,,Atr'uncticnsliet's Logic,,, World Politha, (July lg66),ffi7-622;
and A. L. Kalleberg, ,,The Logic o( Com_parison," World Pakitins,
lg (October 1966), 69_82.While the two authors are or,rconscioua
thinkers,I would certainly agree n'ith Dor!,se's
coucludingaentence, namely, thal .,to ignore trivial loeiealpoints
is to risk being not even trivially trtre;1p.622).
n'Flanigan and Fogelman in CharlesworLh, Con-Lemporary Political
Analyis pp. g2-gl.
"On general 6ystems theory one may uaefullycousulL Orau R.
Young, Sgstems oJ patiLiral ScLence (Englewood CIif fs:
Prcntice-Hall , l96S),Chzp, 2. See also Giuliano Urbani, ',Geueral
Sys-tems Theory: Un Nuovo Stnrmento per l ,Anatisidei Sistemi Pol i
t ic i?," I I Poti t ico,4 (1963), ?g5_8r9-
o'IVhile there is some eontroversy on the respec_tive merits and
shortcomings of the two nirateiies,the strucLural-functional
approach is oot inher_ently tied tc either one. For the partial
vercuswhole syetems controversy lhe two stances are wel,represented
by J. LaPeJombara, who favors thesegmeuted approach. (cf. esp,
,,parsimony audEmpiricism in Comparative pol i t ics: An
Anti_Scholastic View,' , in R. T. Holt gnd J. E. Turner(eds.), ?Ae
I{elhodologE ol Comparaliue Re_
-
1050 TITE AMERICAN POI,ITICAL SCIENCII REVIDW vol,. 64
tural-functional scholar remain tied to "a leve!of analysis
which [does] not permit empiricaltesting?"do According to my
diagnosis there isno intrinsic reason for this. Quite to the
con-trary, we uray expect very rewarding retulns,and the empirica-l
promise (and dietinctiveness)of the approach may weli near
fulfillment, if weonly lea.rn how to manerlver along a ladder
ofabstraction,
Let us now pass on to a more loose discussion-the second part of
this illustration-for whichI ha.ve selected a. somewhat different
family ofeategories : pluralism, integration, participationand
rnobilization-6l While one may think ofmany other examples that
rvould suit my pur-poses just as rvell, the four categories in
questionare representa.tive in that they are ueed (or sig-nidcant
theoretical developments not onlv undera variety of different
fr4meworks, but also bythe non-afrliated scholar, thereby
including-inthe case of participation atrd mobilization-thescholar
who happens to be interested only instatistical manipulations.
Given the fact tha[ piuralisar, integration,participation and
mobilization are culture-bouncconcepts which may reflect-2s far as
we knowat the outset-a distinctive Weatern experience,the
rnethodological caveat here is that the refer-ence &rea should
make for the starting point ofthe investigation. So to speaft, we
are tequired toelaborate our culture-bound concepts in a "we-they"
clockwise direction. It is proper, therefore,to start rvith the
question: How do we under-sfand pluraLism, integration,
participa"tion andmobilization in their domee[ic, original
context?
At home "pluralism" does not z"pply to societaland/or political
strueture, nor to interplay be-tween a plurality of actors.
Pluralism came to beused, in the.Western literature, to convey the
ideathat a pluralistic socieiy is a society whose struc-tural
configuration is shaped by pluralistic be-liefs, namely, that all
kinds of e.utonomous sub-units should develop at all levels, that
interestsare recognized in their legitimate diversity, and
searcb, op. cir., pp. 125-14S); and, for the contreryview, Fred
W- Riggs (cf. especially his for[hcom-ing essey in Eaas aod
Keriel,. Approaehea Lo the1nr.du ol Politb.ol Scizrne.)
'F lan igan and Foge lman, oP. c i t .'The relevent "family
difterence' is Lhat struo-
ture and fttnction are not culture-bound concepts,while lhe four
other categorieE are. This is alsoto note thet the travelling
problem of comparativepolitica cannot be reduced to the
construction of"non-culture bound" conceptualizations. Eow touse
those conceptualization"s which cannot helpbeing culture bound is
equally a problem.
that dissent, not unanimity, represents the basisof civility.
Pluralism is indeed-as already noted-a highly abstmct structural
principle. Yet theterm points to a partiru)ar soeietal
structure-not merely to a developed atage of difterentiationand
specialization-and does retain a wealth ofcharacterizing
connotatiolts whenever we discuss,in the Western dernocracies, our
internal policiesand probLems.
"Integration" can be conceived as an end-6l,tte, trs a process,
or as a, function performedby integrating a4encies (parties,
interest groups,etc.). In eny case, in the Western polities
inte-gration is not applied to wlwteuer kind of "put-ting
together," Lo uhateuer state of amalgama-tion. tr'or instance, when
America.n scholars dis-cues their own domestic problems, they
havedefinite ideas of what is, arrd what is not, inLe-gra.tion-
They would deny that integration hasanything to do with "enforcing
uniformity."They are likely to e-csume, instead, that integra-tion
both presupposes and generates a pluralisticscciety (as qualified
above)- Aad, surely, an inte-grative agency is required to obtain a
maximumof coaleseence and solidarity with a ruinunum
ofcoercion-62
Simila.r points can be made with regard toperticipation and
mobilization. Regardless ofwhether "participatiout' is used
normatively (aspointing to a basic tenet of the democraticideal) or
descriptively (as reflecting a derro-cratic experience), in either
case in our domes-tic discussiorx participation it not any such
kindof "taking part." Thus the advocates of a partici-patory
democracy are hardly satisfled by anykind of involvement in
politics. To them partici-pation means sell-motion; it does not
meen be-ing manipulated ot coerced into motion. Andsurely the
original definite meaning of the termconveys the idea of a free
citizen who acts andintervenes-if he so wishe*-according to hlsbest
judgement. So conceived, participation isthe very opposite, or the
very reverse, of mobili-zation. Mobilizetion does not convey the
idea ofindividual aelf-motion, but the idea of a mallea-ble,
passive collectivity which is beng put intomotion at the whim of
persuasive-and morethan persuesive-authorities. We say that
indi-
'Since we are discrrsqing here macro-problem,aand macro-theory I
need not follow the conceptsunder inveatigation all the way down
the ladder ofabstraction. I should not let pass unnoticed,
how-ever, that "integration" also belonga to the vocab-ulary of
eociology and psychology, thereby lendingitself to very fine lower
level disdnctions. See e.g.,W. S. Landecker, "Typee of Integration
and theirMeaaurements," in The Language o! Eocinl Re-search, op.
ciL., pp. l9-?7.
-
1970 CONCEPT MISFORMATION IN COMPARATIVE POLITICS l05l
viduals "participate," but we cannot say a,boutthe same
individuals that they "mobilize"-theyan mnbilized..
It is quite clear, then, that pluralism, integra-tion,
participa.tion and mobilization all have spe-cific connotations
which can be pinned down,and are in lact retained-no matter how
implic-itly-in our Western euquiries and controversies.However, in
tfie conlexi of global comparativeplilics the specificity of these
notione gets lost:therg is no end to pluralism; integration is
ap-plied indifferently to pluralistic and non-plural-istic
settings; a-nd participation and mobilira-tion are turned into
largely overlapping notiotx.There is no end to pluralism, for we
are nevertold what is non-pluralism. Since pluralism ex-iste
somewhere, the assumption appears to bethat "to a different
degree,' pturalism will befound tn exist everywhere. However, a
differentdegree oJ what! This is indeed the ironv of us,ing
a.degtee language-intended when used ap-nropriately to convey
precision-for conveyingelusiveness. Likewise the mu.ning of
integraiionchanges, and eventually evapora.tes, en route. Fi-nally,
and similarly, the distinction between oar-ticipation and
mobilization onlv holds al home.With rnost comparative orierrted
echolars mobili-zation comes to mean whatever process of
socialactivation; and partieipa"tion is currentlv appliedby the
discipline at large both to democratic andmobiiizational techniques
of political activation.
At this stage of the argument I need not laborat explaining why
and how we obtain these dras-tic losses of specifity, They result,
as we know,from conceptual stretching, which resulfs, inLurn, from
incorrect ladder climbing: the clumsyattempt to arrive at
"travelling universals,' atthe expense of precision, irutead of at
the ex-pense of connolation (i.e., by reducing the num-ber of
qualifying attributes). What iemains tobe highlighted are t}e
consequences of thfu stateof affairs.
Ta.ke, for instance, the formidable errorr irrinterpretation and
prediction which are sug-gested by the universal, unspecified
applica,tionof "pluralismt' and "integration." If weiay thatAfriam
societies are not pluralistic but "tribal-istic," the argument is
likely to be that a situa-tion of tribalistic fragmentalion hardly
providesthe structural basis not only for integrativeprocesses to
occut, but also for bringing in-tegrativc agencies to the fore.
Indeed my argu-ment rvould be that the functional needs,
orfeedbacks, of a fragmented society are at oddswith the functional
feedbacks, or needs, of a plu-ralistic sociely. In Europe, for
instance, medi-eval fragmcntation generated monarchieal
abso-lutism. However, if pluralism is vapori.zed intoan empty
generality, then we are auLhorized to
call African societies pluralistic, and the unfor-tunate
implication may welL be that we expectAfricans to solve their
problems as if they hadto deal with Western-type societies.63
"Mobilization" is also a worthwhile example inthat it confronts
us with a problem that hasonly been mentjoned, so far, in passing.
Whilepluralism, integration and participation are de-rived from our
experience with democracl--i.e.,lrom the context of the democratic
polities-wealso dispose of a limited set of terms which orig-inate
from a tota-literian context. This is thecase of the term
mobilizs-tion, which derivesfrom military terminology-especially
the Ger-man total mobilization of World War l-s,ndenters the
vocabulary of polities via the militiatype of party (as Duverger
calls it), and speci6-cally via the experience of fascism znd oi
nzz-ism.dl Nonetheless the Grm is currently appliedalso to the
democratic politiee-and this meansthet we have drawn r, "reversed
extrapole-tion,'(t.e., a, counter-clockwise extrapoletion).
Andsjnce we often complain that our terminology
isdemocracy-centered, ny 6rst complaint is t}r.twe fail to take
adva.ntage of the fact Lhat we dohave terme which esez-pe the
democratic bias.flowever, the inconvenience resultinq from
re-veraed extrapolationa are seen best, on a broaderscale, aad with
particular reference to what Icall the "boomerang effect" of the
developingares.65
Western schol"a,rs travelling across Africa orSouth-East Asia
discover that our categorieshardly apply, which is hardly
surprising- Fromthis t