Concept Misformation in Comparative Polities Giovanni Sartori The American Political Science Review, Vol. 64, No. 4 (Dee., 1970), 1033-1053. Stable URL: bttp//links jstor.org/sici?sic!=0003-0554%28 197012%2964% 3A4%3C 1033%3ACMICP%3E2.0,CO%3B2-A The American Political Science Review is currently published by American Political Science Association. ‘Your use of the ISTOR archive indicates your acceptance of JSTOR’s Terms and Conditions of Use, available at hhup:/www.jstororg/about/terms.hml. JSTOR’s Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at hup:/www jstor-org/journals/apsa.heml. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the sereen or printed page of such transmission, STOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals, For more information regarding JSTOR, please contact support @jstor.org. bupulwww jstor.org/ Mon Jun 28 11:37:15 2004The American Political Science Review VOL. LXIV DECEMBER, 1970 NO. 4 CONCEPT MISFORMATION IN COMPARATIVE POLITICS* Grovanw1 Sanrorr University of Florence “To have mastered ‘theory’ and ‘method’ is to have become a conscious thinker, a man at work and aware of the assumptions snd implications ‘of whatever he is about. ‘To be mastered by ‘method’ or ‘theory’ is simply to be kept from ‘working.”* The sentence applies nicely to the present plight of political science, The profession as a whole oscillates between two unsound ex- tremes. At the one end a large majority of polit- ical scientists qualify as pure and simple uncon scious thinkers, At the other end a sophisticated ‘minority qualify as overeonscious thinkers, in the sense that their standards of method and theory are drawn from the physical, “paradig- ‘The wide gap between the unconscious and the overconscious thinker is concealed by the rowing sophistication of statistical and research techniques. Most of the literature introduced by the title “Methods” (in the social, behavioral oF political sciences) actually deals with survey techniques and social statistics, and has little if parative Politics,” paper to the IPSA Torino Round Table of Sep= tember, 1960. I wish to thank, in this connection, the Agnelli Foundation which provided the for the Torino panel. T ain particularly indebted to David Apter, Harry Eckstein, Carl J. Friedrich, Joseph LaPalombara, Felix Oppenheim and Fred W. Rigge for their critical commente. T am also very much obliged to the Concilium on Intema- tional and Area Studies at Yale University, of which T was a fellow in 1986-67. This article is part of the work done under the auspices of the Conciius, *C. Wright Mills, ship," in Llewellyn’ Gross (ed), Symposium on Sociologieal Theory (New York) Harper & Rov, 1988) p.27 (My emphasis) anything to share with the erucial concern of “methodology,” which is a concern with the log ical structure and procedure of stientifie en- fquiry. In a very erucial sence there is no meth- ‘oxology without Zogos, without thinking about thinking. And if @ firm distinction is drawn—as it should be—between methodology and tech nique, the latter is no substitute for the former. One may be a wonderiul researcher and ma- nipulator of data, and yet remain an uncon- scious thinker. The view presented in this article is, then, that the profession as a whole is griev~ ously impaired by methodologieal unawareness ‘The more we advance technically, the more we Teave a vast, uncharted territory behind our Dacks, And my underlying complaint is that po- litical scientists eminently lack (with excep- tions) a training in logie—indeed in elementary logie. T stross “elementary” because I de not wish to ‘encourage in the least the overconseious thinker, the man who refuses to diseuss heat unless he is ven a thermometer. My sympathy goes, in- stead, to the “conscious thinker,” the man who realizes the limitations of not having a ther- mometer and still manages to say a great deal simply by saying hot and cold, warmer and fooler. Indeed I all upon the conscious thinker to steer a middle course between erude logical mishandling on the one hand, and logical perfeo- tionism (and paralysis) on the other band. Whether we realize it or not, we are still swim- ‘ming in a sea of naivete. And the study of com- parative polities is particularly vulnerable to, and illustrative of, this unfeliitous state of affairs ‘Traditional, or the more traditional, type of political science inherited a vast array of con cepts which had been previously defined and re- fined—for better and for worse—by generations 10331034 ‘rine AMER of philosophers and political theorists. To some extent, therefore, the traditional political scien tist could afford to be an “unconscious thinker” —the thinking had already been done for him. ‘This is even more the ease with the country-by~ country’ legalistic institutional approach, whieh does not particularly require hard thinking* However, the new political science engages in re= cconceptualization, And this is even more the case, necessarily, with the new comparative ex pansion of the discipline» There are many rea- sons fr this renovatio ab imis. ‘One is the very “expansion on polities.” To some extent polities results objectively bigger on ‘account of the fact that the world is becoming more and more politicized (more participation, more mobilization, and in any ease more state intervention in formerly non-governmental spheres). In no small measure, however, polities is subjectively bigger in that we have shifted the focus of attention both toward the periphery of politics (vis-i-vis the governmental process), ‘and toward its input side, By now—as Macridis its it—we study’ everything that is “potentially political” While this latter aspect of the expan- sion of politics is disturbing—it ultimately leads to the disappearance of polities—it is not a pe culiar concern for comparative polities, in the sense that other sogments of politcal scence are equally and even more deeply affeeted.* This ia by no means a eritieism of a compar functional” approach. On the latter ee the judicious remarks of Ralph Braibant “Comparative Political Analytics Reconsidered,” The Journal of Politics, 80 (February 1968), 4-49. "For the various phases of the comparative ap proach see Eckstein's perceptive “Tatroduction,” in HE Eckstein and D. E. Apter (ods), Comparative Politics (Glencoe: Pree Press, 1963). “Comparative Polites and the Study of Gov- ferment: ‘The Search for Focus,” Comparative Politics, (October 198), p. 81 *On the “fallacy of inputisin” soc again the re- marks of Roy C. Macrdis, Ine. cit, pp. 84-87. Ia his words, “The atate of the discipline can be summed up in one phrase: the gradual disappear~ ance of the politcal.” (p. 88). A cogent statement of the issue is Glenn D. Paige, “The Rediscovery ‘of Politics” in J. D. Montgomery end W. I. Sifin (eis), Approaches to Development (New York McGraw Hill, 1968), p. 49 . My essay “From the Sociology’ of Politica to Political Sociology,” S.M. Lipset (od.), Politics and the Social Sciences (New York: Oxford University Press, 1960), pp. 15-109, is also lnrgely concerned with’ the fellacy of inputism viewed as a soriologiesl reduction of polities. POLITICAL. SCIENCE, REVIEW vou. 64 Aside from the expansion of polities, a more specific source of conceptual and methodological challenge for comparative politics is what Br banti calls the “lengthening spectrum of political systems.” We are now engaged in world-wide, ‘rose-area comparisons. And while there is an fend to geographical size, there is apparently no fend to the proliferation of politieal units. There ‘were about 80 States in 1945; itis no wild guess that we may shortly arrive’ at 150. Still more important, the lengthening spectrum of politieal systems includes a variety of primitive, diffuse polities at very different stages of differentiation ‘and consolidation ‘Now, the wider the world under investigation, the mote we need conceptual tools that are able to travel, It is equally clear that the pre-1950 vocabulary of polities was not devised for world wide, erose-area travelling. On the other hand, and in spite of bold attempts at drastic termino- Togical innovation’ it is hard to sec how West- ern scholars could radically depart from the po- litical experience of the West, ie, from the vo~ cabulary’ of polities which has been developed over millennia on the basis of such experience. ‘Therefore, the first question is: how far, and how, ean we travel with the help of the available vocabulary of polities? ‘By and large, 20 far we have followed (more or less unwitingly) the line of least resistance: broaden the mesning—and thereby the range of application—of the conceptualizations at hand. ‘That is to say, the larger the world, the more we have resorted to conceptual stretching, or con~ ‘eptual straining, ie., to vague, amorphous con- ceptualizations. ‘To be sure, there is more to it. ‘One may add, for instance, that conceptual stretching also ‘represents deliberate attempt to make our conceptualizations value free. An- ‘other concurrent explication is that eoneeptual straining is largely a “boomerang effect” of the developing areas, i, a feedback on the Western categories of the difuse polities of the Third "Comparative Political Analytics Revonside ered" Toe. eit, pp. 36-87. "The works of Fred W. Riggs are perhaps the best instance of such bold attempts. For a rocent presentation soe "The Comparison of Whole Po- Titieal Systems," in R. T, Holt and J. B. Tamer (eda), The Methodology of Comparative Research (New York: Free Press, 1070), esp. pp. 95-116. While Riggs’ innovative strategy has undeniable practical drawbacks, the crticim of Martin Lan- flaw ("A General Commentary,” in Ralph Braib- anti (ed), Political and Administrative Develop- ‘ment (Durham: Duke University Press, 1900), pp. 5325-834) appears somewhat1970 World® These considerations notwithstanding, conceptual stretching does represent, in compar ative polities, the line of least resistance. And the net result of conceptual straining is that our gains in extensional coverage tend to be matched bby losses in connotative precision. Tt appears that we ean cover more—in travelling terms— only by saying less, and by saying less in a far Jess preeise manner. ‘A major drawback of the comparative expan- sion of the discipline is, then, that it has been conducive to indofiniteness, to undelimited and largely undefined conceptualizations. We do need, ultimately, “universal” categories—eon- cepts which are applicable to any time and place. But nothing is gained if our universals tum out to be “no difference” categories lending to pseudo-equivalences. And even though we need universals, they must be empirical univer~ sals, that is, eategories which somehow are amenable, in’spite of their all-embracing very abstract nature, to empirical testing, Instead we seem to verge on the edge of philosophical uni- versuls, understood—as Croce defines them—as concepts which are by definition supra-empiri- cal? ‘That the comparative expansion of the disci pline would encounter the aforementioned stum- bling block was only to be expeeted. Tt was easy to infer, that is, that conceptual stretching ‘would produce indefiniteness and elusiveness, ‘and that the more we climb toward high-fown universals, the more tenuous the link with the tampirieal evidence. Tt ig pertinent to wonder, therefore, why the problem has seldom been squarely confronted. “Taking a step back, let us begin by asking whether it is really necessary to embark in haz- ardous world-wide comparisons. This question hinges, in turn, on the prior question, Why eom- pare? "The unconscious thinker does not ask himself why he is comparing; and this neglect oes to explain why so much comparative work provides extensions of knowledge, but hardly a strategy for acquiring and validating new knowledge. Tt is not intuitively evident that to compare is to control, and that the novelty, di tinetiveness and importance of comparative poli- tics consists of a cystematic testing, against as many cases as possible, of sets of hypotheses, generalizations and laws of the “if . . . then” "On the boomerang effect of the developing areas more in the final action "More precisely in B. Croce, Logiea come Sei- ‘enzo del Concetto Puro, (Bari: Laterzs, 1912), pp. 18-17, universals are defined wlirerappresenta tivi, as being above and beyond any conceivable empirical representobility. CONCEPT MISFORMATION IN COMPARATIVE POLITICS, 1035 types? But if comparative polities is conceived as. method of control, then its generalizations have to be checked against “all cases,” and therefore the enterprise must be—in prineiple— 4 global enterprise. So the reason for world-wide comparisons is not simply that we live i wider world; it is also a methodolo 'E tivo oF more items aro identical, we do not have a problem of comparability. On the other hhand, if two oF more items have nothing, or not enough in common, we rightly say that stones And rabbits cannot be campared. By and large, then, we obtain comparability when two oF more item’ appear “similar enough,” thot is, nether identical nor utterly diferent. But this asess- ment offer ite positive guidance. The problem fs often outflanked by saying thet we make things comparable. In this perspective to com- pare is "to asimilate” ie, to discover deeper or fundamental similarities below the surface of secondary diversities. But this argument equally ‘fords Title milage and conveys, moreover, the risleading stagestion that the trick resides in taking the unlike look alike. Surely, then, we hhave here & major problem which eantot be dis- posed of with the argument that political theo- ts have performed decently with comparing nce the time of Aristotle, and therefore that we should not get bogaed by the question “What is Comparable?” any more then our predecesors ‘This argument will not do on aceount of three diferences, Tn the fist place if our predecessors were cul- ture botind this implied that they travelled only a far as their personal knowledge allowed them to travel. In the second place, our predecessors hardly disposed of quantitive data and were not quantitatively oriented. Under both of these lim ations they enjoyed the dstinet advantage of having a substantive understanding of the things they were comparing. This is hardly pos- sible on a world wide seale, and surely becomes impossible with the computer revolution. A few years ago Karl Deutsch predicted that by 1975 the informational requirements of political sci- ence would be satisfied by some “ity million card-equivalents [of IBM standard cards) « fd a total annual growth rate of perhaps 1s For the comparative method as « “method of control” see expecially Arend Lijphart, Compara- tive Politics and the Comparative Method, (mim- cographed) paper presented at the Torino IPSA Round Table, September, 1969. According to Lijp- hhart the comparative method is a “method of dis- covering empirical relationships among variables” (p. 2); and I fully concur, except that this defini- tion ean be entered only at a later stage of the ar- sgument.1036 ‘much as five million” T find the estimate frightening, for computer technology and facili- ties are bound to flood us with masses of data for which no human mind ean have any sub- stantive grasp. But even if one shares the enthu- siasim of Deutsch, it cannot be denied that we have here a gigantic, unprecedented problem, In the third place, our predecessors were far from being as unguided as we are. They did not Teave the decision about what was homogenous ie, comparable—and what was heterogen- ous—ie., non-romparable—to each man’s genial insights." As indicated by the terminology, their comparisons applied to things belonging to “the ssime gems.” That is to say, the background of comparability was established by the per genus et diferentiam mode of analysis, ic., by a taxo- nomical treatment, Tn this contest, ‘comparable ‘means something which belongs to the same ge- nus, species, or sub-species—in short to the same class, Hence the class provides the “similarity el~ ‘ement” of comparability, while the “diferences” enter as the species of a genus, or the subsspe- cies of a species—and so forth, depending on how fine the antlysis needs to be. However, and here is the rub, the taxonomical requisites of comparability are currently neglerted, if not dis- owned. We are now better equipped for a discussion of our initial query, namely, why the travelling problem of comparative polities has been met with the poor remedy of “conceptual stretching” instead of being squarely confronted. While there are many reasons for our neglect to attack the problem frontally, a major reason is that we have been swayed by the suggestion that, our dificuties ean be overcome by switching from “‘uhat is” questions to “how much” questions, ‘The argument runs, roughly, as follows. As long as concepts point to differences of kind, Le, a8 long as we pursue the either-or mode of analysis, we are in trouble; but if concepts are under- stood as a matter of more-or-less, i.e, a8 point ing to differences in degree, then our difficulties canbe solved by measurement, and the real problem is precisely how to measure. Meanwhile —waiting for the measures—class concepts and taxonomies should be looked upon with suspi cion (if not rejected), since they represent “an old fashioned logic of properties and attributes not well adapted to study quantities and rela tions.” * Recent Trends in Research Methods," in J.C. Charlesworth (ed.), A Design fr Politica! Science Scope, Objectives end Methods (Philadelphia ‘American Academy of Political and Social Science, 1960), p. 156. "Carl P. Hempel, quoted in Don Martind “Socjologieal Thoory and the Ideal Type,” THE AMERICAN POLITICAL SCIENCE, REVIEW vou. 64 According to my previous analysis, a taxo- nomie unfolding represents a requisite condition for comparability, and indeed » background which becomes all’ the more important the less ‘we can rely on a substantive familiarity with what is being compared. According to the fore- going argument, instead, quantification has no ills of its own; rather, it provides a remedy for ‘he ills and inadequacies of the per genus et dif- ferentiam mode of analysis. My own view is that ‘when we dismiss the so-called “old fashioned logie” we are plain wrong, and indeed the vietims ‘of poor logie—a view that T must now attempt to warrant. 1, QUANTIFICWTION AND CLASSIFIONTION What is very confusing in this matter is the abuse of a quantitative idiom whieh is nothing but an idiom. Al! too often, that is, we speak of degrees and of measurement “not only without ny actual measurements having been per- formed, but without any being projected, and even without any apparent awareness of ‘what rust be done before such measurements ean be carried out.” For instance, in most standard textbooks one finds that nominal scales are spo- keen of as “seales of measurement." But a nom- inal seale is nothing else than a qualitative clas sification, and I fail to understand what it is that a nominal seale does, or ean, measure, To be sure classes ean be given numbers; but this is simply a coding device for identifying items and has nothing to do with quantifiation. Likewise the incessant use of “it is a matter of degree” phrascology and of the “continuum” image leave us with qualitative-impressionistic statements which do not advance us by a hair's breadth to- ward quantifiation. In a similar vein we speak more and more of “variables” which are not variables in any proper sense, for they are not attributes permitting gradations and implying Gross, Symposium on Sociological Theory, p. 87 Martindale aptly comments that “Hempel’s judg- ments are made from the standpoint of the natural sviences.” But the vein is not dissimilar when the Statistically trained scholar argues that “whereas it is admittedly technically possible to think raya in terms of attributes and dichotomies, one wonders how practical that is": Hubert M. Bla- Tock, Jr, Causal Inferences in Nonezperimental Rescarch (Chapel Hill: University of North Caro- Jina Pres, 1964, p. 32) "Abraham Kaplan, The Conduct of Inquiry (San Francisco: Chandler, 1068), p. 212 Fg. [. Festinger and D. Kate (eds.), Research Methods in the Behavioral Scionces (New York: Dryden Press, 1958); and Selliz, Jahoda et al, Rescarch Methods in Social Relations (rev. New York: Holt, Rinehart & Winston, 1959)1970 ‘measurability. No harm necessarily follows if it pleases us to use the word variable as a syn- ‘onym for the word coneept; but we are only de- Inding ourselves if wre really believe that by aay ing variable we have a variable. All in all, coquetting (if not cheating) with a quantitative idiom grossly exaggerates the ex- tent to which political science is currently amen able to quantification, and, still worse, obfus- cates the very notion of quantification, The di- viding Hine between the jargon and the substance ‘of quantification can be drawn very simply ‘quantification begins with numbers, and when ‘umbers are used in relation to their arithmeti- ‘al properties. To understand, however, the mul- tifaected complexities of the notion beyond this dividing line is a far less simple matter. Never= theless one may usefully distinguish—in spite of the close interconnections—among three. broad areas of meaning and application, that is, be- tween quantification as i) measurement, i)’ sta- tistical manipulation and, i) formal mathemat- ‘eal treatment, In political science we generally refer to the first meaning. That is to say, far more often than not the quantification of political science consists of (a) attaching numerical values to items (pure and simple measurement), (b) us ‘ng numbers to indicate the rank order of items (ordinal seales) and (c) measuring differences or distances among items (interval seales) Beyond the stage of measurement we do own, in addition, powerful statistical techniques not only for protecting ourselves against sampling and measurement errors, but also for establish- ing significant relationships among variables, However, statistieal processing enters the scene only when sufficient numbers have been pinned ‘on sufficient items, and becomes central to the discipline only when we dispose of ‘variables which measure things that are worth measuring, Both conditions—and especially the latter—are "There is some question as ¢o whether it can really be held that ordinal scales are scales of rmensuroment: most of onr rank ordering occurs nithout having recourse to numerical values, and ‘whenever we do assign numbers to our ordered categories, these numbers are arbitrary. However, there are good reasons for drawing the threshold fof quantification between nominal and ordinal scales rather than between ordinal and interval scales. (See Edward R. Tufte, “Improving Data Analysis in Political Science)” World Politics, 21 (July 1969), esp. p. 645.) On the other hand, even if the gap between ordinal sealee and interval measurement is not as wide in practice aa it isin theory, nonetheless from a mathematical point of view the interesting scales are the interval and even more, of course, the exrdinal scales, CONCEPT MISFORMATION IN. COMPARATIVE, POLITICS 1037 hhard to meet. Indeed, a crossexamination of ‘or statistical findings in terms of their theoreti- cal signifanee—and/or of « “more relevant” political. science—shows an impressive dispro- portion between bravura and relevance. Unfor- tunately, what makes a statistical treatment theoretically signifieant has nothing to do with statistics ‘As for the ultimate stage of quantifieation— formal mathematical treatment—it is a fact that, so far, political seience and mathematics have engaged only “in a sporadic conversa tion.”"" Tis equally a fact that we seldom, if ever, obtain isomorphie correspondences between empirical relations among things and formal rela- tions among mumbers.* We may well disagree about future prospects? or as to whether it “Otherwise the comparative method would largely consist of the statistical method, for the latter surely is a stronger technique of control than the former. The difference and the connections are cogently discussed by Lijphart, “Comparative Pol- itics and and the Comparative Method,” op. cit. * Oliver Benson, “The Mathematical Appronch to Political Science,” in J. C. Charlesworth (ed.), Contemporary Political Analysis (New York: Free Prose, 1967), p. 182. The chapter usefully reviews the literature, For an introductory treatment see Hayward R. Alker, Jr, Mathematice and Politice (New York: Macmillan, 1965). An illuminating dliseussion on how quantification enters the varios socialsciences is in Daniel Lerner (ed), Quantity ‘and Quality (Glencoe: Pree Press, 1961), passim. "A classic exemple i the (partial) mathematical transition of the theoretical system of The Hu- ‘man Group of George C. Homans by Herbert A. Simon, Models of Man (New York: Wiley, 1967), Chap.'7. No similar achievement exits in the po litieal stience field. To cite three significant in- stances, political svience issues are eminently lack ing in Kenneth J, Arrow, “Mathematical Modele in the Social Sciences,” in D. Lemer and H. D. Lasswell (eds), The Policy Sciences (Stanford: Stanford University Press, 1951), Chap. 8; in the contributions collected in’ P. F, Lagarefeld (ed), Mathematient Thinking in the Social Sciences Glencoe: Free Press, 195) ; in J.C. Kemeny and 5. L. Snell, Mathematical Models in the Social Sciences (Boston: Ginn, 1962) Perhaps the mathematical leap of the dissie pline js just around the comer waiting for non- ‘quantitative developments. I one is to judge, how ever, from the “mathematics of man” issue of the Indernational Social Science Bulletin introduced bby Claude LevieStrauss (IV, 1954), thie Heerature is very deceiving. More interesting is John G. Kemeny, “Mathematics without Numbers” in Lerner, Quantity and Quality, pp. 95-81; and the ‘modal logic developed by the Bourbaki group,1038 makes sense to construct, formalized systems of quantitatively well defined relationships (mathematical models) so long as we wander in mist of qualitatively ill-defined eoncepts. If we fare to lear, however, from the mathematical development of economies, the evidence is that it “always lagged behind its qualitative and con- ceptual improvement.”2° And my point is, pre cisely, that this isnot a easual sequence. It is for ‘a very good reason that the progress of quantifi- cation should lag—in whatever diseipline—be- hind its qualitative and eoneeptual progress, Tn this messy controversy about quantifiea- tion and its bearing on standard logieal rules we simply tend to forget that concept formation stands prior to quantification. The process of thinking inevitably begins with a qualitative (natural) language, no matter at which shore we shall subsequently land. Correlatively, there is zo ultimate way of bypassing the fact that bu- ‘man understanding—the way in which our mind wworks—requires cut-off points which basically correspond (in spite of all subsequent refine ments) to the slices into which a natural or ‘qualitative language happens to be divided, ‘There is a fantastic lack of perspective in the argument that these cut-off points ean be ob- ‘tained vin statistieal processing, Le, by letting the data themselves tell us where to draw them. For this argument applies only within the frame ‘of eonceptual mappings which have to tell us frst of what reality is composed. Let it be stressed, ‘therefore, that, long before having data which can speak for themselves the fimdamental artic~ ulation of language and of thinking is obtained logieally—by cumulative conceptual refinement ‘and chains of coordinated definitions—not by ‘measurement. Measurement of what? We cannot measure unless we know first what it is that we fare measuring. Nor ean the degrees of something tell us what a thing is. As Lazarsfeld and Barton neatly phrase it, “before we can investigate the presence or absence of some attribute ... or be- fore we can rank objects or measure them in Flimente de Mathématique, appearing periodi- cally (Paris: Hermann). For 2 general treatment sce J. G. Kemeny, J. L. Snell, G. L. Thompson, Introduction to Finite Mothematice (Englewood Clits: Prentice Hall, 1957). "Joseph J. Spengler, “Quantifcation in Eeo- nomics: Its History," in Lemer, Quantity ond Quality, p. 176. Spengler equally points out that “the introduction of quantitative methods in economics did not result in striking discoveries” (ibid). While format economic theory is by now highly’ isomorphic with algebra, mathematioal economics has added litle to the predictive power of the discipline and one often has the impression that we are employing guns to Kill mosquitos. ‘THE AMERICAN POLITICAL SCIENCE REVIEW vou. 64 terms of some variable, te must form the con- tent of that variable” ‘The major premise is, then, that quantifca- tion enters the scene after, and only ater, hav ing formed the concept. ‘the minor premise is that the “stuf of -quantieation—the things tnderpinned by the tumbers—eannot be pro- Vided by quantifeation itself. Hence the rules of Concept formation are independent of, and an~ not be derived from, the rules which govern the treatment of quantities and. quantitative reas tions. Let us elaborate on this conclusion. Tn the fst pce, if we never realy “how much” fndings—in the sense that the prior question always is how mich in what, in what Conceptual container—it follows from this that how much quantitative findings ate an internal element of “what is” qualitative questions: the tlaim that the latter should give way to the former cannot be sustained. Tt equally feos in the second place, that “categorie concepts” of the eitheroe type cannot give ray” to "arid tion coneepte” ofthe moresthan-ess-than type. ‘What is usually lost sight of is that the either- or type of loge i the very loge of easication Tring, Classes are requied to he mutually ex- lisive, i, laa concepts represent charaterie- tics which the object tinder consideration must tither have of lack. Two items bring compared must belong fist to the same class and either Have or not have an atsbute; and only if they have ft the two items ean be matched in terms af which has it more or la. Fence the loge of sradation belongs {0 the logi of elassifeaton Moce precisely pt, the switch from clssifiea: tion to gradation basicaly consists of replacing the signs “same-diflernt” with the signs “same= sveaterleser,” ie, consists of introducing ‘hantitative diferentation within a. qualitative Sameness (of atrbutes). Clearly, then, the sen “same” established by the loge of esiSeaton is the reaisite condition of introducing. the signs “plus-minus.” “The Flor tends tobe that this trae only a Jong as we persist in thinking interns of afte Stes and dlcotemis, But this rejoinder misses the point thataaide from claifying—we dive ese of 0 other folding technique, Indeed, {he taxonomicalevereise “unpacks” concepts, tnd plays-a-non-replaceable roe in the process of thinking in that it decomposes mental com pounds into order’ and. manageable sets of Component nits, Let it be added that at no Stage of the methodologial argument does the tasonomical tnpacking lose weight and impor- Qualitative Measurement in the Social Sei- ‘ences: Classifications, Typologies and Indies,” in D. Lemer and H. D. Lasewell (eds), The Policy Sciences, op. cit, p- 185 (my emphasis).1970 tance, As a matter of fact, the more we enter the stage of quantification, the more we need unidi- ‘mensional geales and continua; and dichotomous categorizations serve precisely the purpose of es ‘ablishing the ends, and thereby the uni-dimen- sonality, of erch continzum, Having disposed of the fuzziness brought about by the abuse of a quantitative idiom, at= tention should immediately be called to the fact~ finding side of the coin. For my emphasis on concept formation should not be misunderstood to imply that my concer is more theoretical than empirical. This is not s0, because the con~ cepts of any social seience are not only the ele- ments of a theoretical system; they are equally, and just as much, data containers, Indeed data is information which is distributed in, and pro~ cessed by, “conceptual containers.” And sinee the non-expetimental sciences basically depend on fact-finding, ie., on reports about extemal (not laboratory)’ observables, the empirical question becomes what tums a concept into a valuable, indeed a valid, fact finding container. ‘The reply need not be far-fetched: the lower the discriminating power of a eonceptual eon- tainer, the more the facts are misgathered, ie, the greater the misinformation. Conversely, the higher the diseriminating power of a eategory, the better the information. Admittedly, in and by itself this reply is not very illuminating, for it only conveys the suggestion that for fact-find~ ing purposes it is more profitable to exaggerate in over-differentiation than in over-assimilation ‘The point is, however, that what establishes, or helps establish, the discriminating power of a category is the taxonomical infolding. Since the logical Fequirement of a classification is that its classes should he mutually exclusive and jointly exhaustive, it follows from this that the taxo- romical exercise supplies an onderly series of ‘yell sharpened eategories, and thereby the basis for collecting adequately precise information. ‘And this is indeed how we know whether, and to ‘what extent, a concept has a fact-gathering va- lity. Once again, then, it appears that we have started to rim before having learned how to ‘walk, Numbers must be attached—for our pur- posee—to “things,” to facts. How are. these things, or facts, identified and collected? Our ultimate ambition may well be to pass from a science “of species” to a science of “functional co-relations.”** The question is whether we are not repudiating a science of species in exchange for nothing. And it scems to me that premature haste combined with the abuse of a quantitative "Herold D. Lasswell and Abraham Kaplat Power ond Society (New Haven: Yale Us ross, 1950), pp. XVI-XVII. CONCEPT MISFORMATION IN COMPARATIVE, POLITICS 1039 idiom is largely responsible not only for the fact ‘that much of our theorising is muddled, but also for the fact that much of our research is trivial and wasteful ‘Graduate students are being sent all over the vworld—as LaPalombara vividly puts it—on “in- Giseriminate fishing expeditions for data.”*> ‘These fishing expeditions are “indiscriminate” in that they luck taxonomieal backing; which is the same as saying that they are fishing expedi- tions without adequate nets, The researcher sets out with a “checklist” which is, at best, an im- perfect net of his own, This may be an expedient ‘way of handling his private research problems, but remains a very inconvenient strategy from the angle of the additivity and the comparabil ity of his findings. As a result, the joint enter prise of comparative polities is menaced by a rowing potpourri of disparate, non-cumulative fand—in the aggregate—misleading morass of information, ‘AIL in all, and regardless of whether we rely fon quantitative data or on more qualitative in- formation, in any ease the problem is the same, namely, to construct fact-finding extegories that own sufficient diseriminating power." If our data containers are blurred, we never know to ‘what extent and on what grounds the “unlike” is made “alike.” If s0, quantitative analysis may ‘yell provide more misinformation than qualita- ‘tive analysis, especially on account of the aggra~ wating circumstance that quantitative misinfor- mation ean be used without any substantive knowledge of the phenomena under consider- ation, ‘To recapitulate and conclude, T have argued that. the logic of either-or cannot be replaced by the logic of more-and-less, Actually the two Jogies are complementary, and each has a legiti- mate field of application. Corrdatively, polar oppositions and dichotomous confrontations feannot be dismissed: they are necessary step jn the process of concept formation. Equally, ‘impatience with classification is totally unjustie fied, Rather, we often confuse a mere enumera- *--Nfacrotheories and Mieroapplications in Com= parative Politi,” Comparative Politics, (October 1908), p. 06 ™It hardly needs to be emphasized that census data—and for that matler most of the data pro- vided by external agencies—are gathered by con- feptual containers which hopelessly lack diserimi- nation. The question with our standard variables fon literacy, urbanization, occupation, industrial tion, and the like, ie whether they really measure common underlying phenomena. Tt is pretty ob- ‘vious that, erose the world, they do not; and this ‘ite aside from the reliability of the dats gathor- ing agencies.1040 tion (or checklist) with a classification, and many so called classifications fail to meet the ‘minimal requirements for what they claim to be. ‘The overconscious thinker takes the view that if the stady of polities has to be a “science,” then it has to be Newton (or from Newton all the way up to Hempel). But the experimental method is hardly within the reach of politieal science (beyond the format of small group ex- perimentation) and the very extent to which we are systematically tuming to the comparative method of verification points to the extent to which no stronger methori—inchuding the statis- tical method—is available. If so, our distinetive tnd major problems begin where the lesson of the more exact sciences leaves off. This is tanta- ‘mount to saying that a wholesale acceptance of the logie and methodology of physies may well be seli-defeating, and is surely of little use for our distinctive needs. In particular, and what ever their limits, clasfications remain the req- uisite, if preliminary, condition for any seientine discourse. As Hempel himeelf concedes, clasifi- eatory concepts do lend themselves to the de- scription of observational findings and to the formulation of initial, if erude, empirical gener- alizations** Moreover, a classficatory activity remains the basic instrument for introducing an- lytical clatity in whatever we are discussing, and leads us to discuss one thing at a time and different things at different times. Finally, and especially, we need taxonomieal networks for solving our fact-finding and factstoring prob- lems. No comparative science of polities is plaus- ‘ible—on a global seale—unless we ean draw on extensive information which is suficiently pre- cise to be meaningfully compared. The requisite condition for this is am adequate, relatively sta ble and, thereby, additive fling system. Such a filing system no longer is a wild dream, than to computer technology and facilities except for the paradoxical fact that the more we enter the computer age, the less our fact-finding and fact-storing methods abide by any. losically standardized criterion. ‘Therefore, my concern with taxonomies ig also a eoncem with 1) the data side of the question, and 2) out frilure to provide a filing system for computer exploita- tion. We have entered the computer age—but ith feet of clay If quantifestion eannot solve our problems, in ‘that we cannot measure before conceptualizing, and if, on the other hand, “conceptual stretch ing” is dangerously conducive to the Hegelian * Fundamentals of Concept Formation in B= pirical Science (Chicago: University of Chieego Pres, 1952), p. 5 THE AMERICAN PoLtTicsy SCIENCE REVIEW vou. 64 night in which all the cows look black (and eventually the milkman is taken for a cow), then the issue must be joined from its very be: ginning, that is, on the grounds of concept for- mation, A few preliminary cautions should be entered. ‘Things conceived or meaningfully perceived, ie, concepts, are the central elements of proposi- tions, and—depending on how they are named— provide in and by themselves guidelines of inter- pretation and observation. It should be under- stood, therefore, that I shall implicitly refer to the conceptual ‘element probleme which in a more extended treatment actually and properly belong to the rubric “propositions.” By saying concept formation T implicitly point to a propo- sition-forming and problem-solving activity. Tt should also be understood, in the second place, that my focus will be on those concepts which are crucial to the discipline, that is, the concepts which Bendix describes as “generalizations in disguise.”=* In the third place, I propose to eon centeate on the vertical members of a conceptual structure, that is, on 1) observational terms, and 2) the vertical disposition of such terms slong a ladder of abstraction While the notion of abstraction ladder is re- lated to the problem of the levels of analysis, the ‘wo things do not coincide. A highly’ abstract level of analysis may not result from “ladder climbing. Indeed a number of universal concep tualizations are not abstracted from observa bles: they are “theoretical terms” defined by ‘their systemic meaning? For instance the ‘meaning of isomorphism, homeostasis, feedback, gqurphy, et» ie ascly defined by the par that each concept plays in the whole theory. In other instances, however, we deal with “observa tional terms,” that is, we arrive at highly ab “Reinhard Bendix, “Concepts and Generalien- tions in Comparative Sociological Studies,” Ameri- can Sociolopicel Review, 28 (1963), p. 888. *Sce Abraham Keplan, Phe Conduct of Inquiry, pp. 60-87, 68-65. According to Hempel theoretical tems “usually purport to not directly observable entities and their characteristics... . They funo- tion ... in scientific theories intended to explain ations”; “The Theoretician’s Diem,” igh, Scriven and Maxwell (eds), Minnesota Studies in the Philosophy of Science’ (Minneap- lis: University of Minnesota Prese, 1958), vol UL, p. 42. While itis admittedly dificult to’ draw ‘neat division between theoreticel end observ tional terms, itis widely recognized that the fore mer cannot be reduced to, nor derived from, the latter, For a recent arsessment of the controversy, see A. Meotti, ‘“L/Eliminesione dei Termini ‘Teorici” in Rivista ai Filosofia, 2 (1960), pp. 119 13,1970 tract levels of conceptualization via ladder climbing, via abstractive inferences from observ ables, For instance, terms such as group, com- ‘munication, eonfict, and decision can either be used in a very abstract or in a very concrete ‘meaning, either in some very distant relation to observables or with reference to direct observa~ tions, In this ease we have, then, “empirical eon- cepts” which can be located ‘at, and moved along, very different points of a ladder of ab- straction, If 0, we have the problem of assessing the level of abstraction at which observational ‘or (in this sense) empirical concepts are located, and the rules of transformation thus resulting. ‘And this seems to be the pertinent foeus for the issue under consideration, for our fundamental problem is how to make’extensional gains (by climbing the abstraction ladder) without having to suffer unnecessary losses in precision and em- pirial testability. ‘The problem can be neatly underpinned with reference to the distinction, and relation, be- tyween the extension (denotation) and intension (connotation) of a term. A standard definition is 1g follows: “The extension of a word is the class fof things to which the word applies; the inten- sion of a word is the collection of properties which determine the things to which the word applies." Likewise, the denotation of a word is the totality of objects indicated by that word; and the connotation is the totality of character- ties anyehing must possess to be in the denota- tion of that word ** ‘Now, there are apparently two ways of climb- ing a ladder of abstraction. One is to broaden the extension of concept by diminishing its at- tributes or properties, ie., by reducing its con notation. In this case a more “general,” or more inclusive, concept ean be obtained without any Joss of precision. The larger the class, the lesser its differentine; but those differentiae that re- ‘main, remain precise. Moreover, following this procedure we obtain conceptualizations which, ‘ho matter how all-embracing, still bear a trace able relation to a colleetion of specifies, and—out of being amenable to identifiable sets of specifies —lend themselves to empirical testing, ‘On the other hand, this is hardly the proce- dure implied by “conceptual stretching,” which adds up to being an attempt to augment the ex- *T quote from Wesley C. Salmon, Logie (Bngle- wood Clifs: Prentice-tall, 1963), pp. 90-01. The distinction is more or less the same in any’ text= book of logic. Connotation” is also applied, more broudly, to the associations, oF associated conceptions brought to mind by'the use of a word. As indi- cated by the text, I intend here the narrower meaning. CONCEPT MISFORMATION IN COMPARATIVE POLITICS 1041 tension without diminishing the intension: the denotation is extended by obfuscating the con- notation, As a result we do not obtain a more general concept, but its counterfeit, a mere gen~ erality (where the pejorative mere” is meant to restore the distinetion between correct ant correct ways of subsuming a term under a broader genus.) While a general concept can be snid to represent a colleetion of specifics, a mere ‘generality cannot be underpinned, out of its in- definiteness, by specifies. And while » general foncept ig conducive to scientific “‘generaliza- tions," mere generalities are conducive only to vagueness and conceptual obscurity. ‘The rules for climbing and descending along ladder of abstraction are thus very simple rules —in principle, We make a concept more abstract ‘and more general by lessening its properties or attributes. Conversely, a concept is speeified by the addition (or unfolding) of qualifications, ie, by augmenting its attributes or properties. If so, let us pass on to eonsider a ladder of abstraction 1s such, Tt is self-evident that along the abstrac- tion ladder one obtains very different degrees of inclusiveness and, conversely, specificity. These differences can be usefully underpinned—for the purposes of comparative polities—by distin- fishing three levels of abstraction, labeled, in shorthand, HL (high level), Mi, (medium level), and LL (low level) High level categorizations obtain universal coneeptualizations: whatever connotation i sac- rificed to the requirement of global denotation— cither in space, time, or even both.*° HL con- cepts can also be visualized ag the ultimate ge- nus which cancels all its species. Descending a step, medium level categorizations fall short of universality and thus can be said to obtain gen- eral classes: at this level not all differentine are sacrificed to extensional requirements, Nonethe- Tess, ML concepts are intended to stress similari- ties’ at the expense of uniqueness, for at this lovel of abstraction we are typically dealing with generalizations. Finally, low level categories ob- tain specific, indeed eonfigurative eonceptualiza- tions: here denotation is sacrificed to accuracy of connotation. One may equally say that with LL categories the differentine of individual set- tings are stressed above their similarities: so much so that at this level definitions are often contextual. A couple of examples may be usefully entered. Ina pereeptive essay which runs parallel to my "The space and time dimensions of concepts are often associsted with the geography versus history debate. I would rather see it a5 the “when goes with when?” quostion, that is, as « calendar ‘ime versus biatorial time dilemma, But this Tine of development cannot be pursued here1082 line of thinking Neil J. Smelser makes the point that, for purposes of comparability, “staff is ‘more satisfactory than administration ... , and administration is more satifaetory than civil ser~ viee."** This is so, according to Smelser, beestse the concept of civil service “is literally useless in connection with societies without a formal state ‘or governmental apparatus.” In this respect “the concept of administration is somewhat superior - but even this term is quite eulture-bound.” Hence the more helpful term is “Weber's con- cept of staff... since it ean encompass without embarassment various political arrangements Tn my own terms the argument would be rephrased as follows. In the field of so-called comparative public administration, “stall” is the high level universal category. “Administra tion” is still » good travelling eategory, but falls short of universal applicability in that it retains some of the attributes associnted with the more specific notion of “bureaucracy.” Descending the ladder of abstraction further we then find “civil service,” which is qualified by its associations with the modern State. Finally, and to pursue the argument all the way down to the low level of abstraction, a comparative study of, say, French and English state employees will dis cover their unique and distinguishing traits and would thus provide contextual definitions. ‘The example suggested by Smelser is fortu- nate in that we are offered a choice of terms, so that (whatever the choice) a different level of abstraction can be identified by a different de- nomination. The next example is illustrative, in stead, of the far less fortunate situation in which We may have to perform across the whole ladder of abstraction with one and sume term. Tn ilhis- ‘rating his caution that many concepts are “gen cralizations in disguise,” Bendix comes across such # simple eoneept as “village.” Yet he notes ‘that the term village may be misleading when applied to Indian society, where “the minimum degree of cohesion commonly associated with this term is absent.”"* Even in such a simple case, then, a scholar is required to place the var~ fous associations of “village” along an abstrac- tion ladder in accord with the travelling exten sion afforded by each connotation, Clearly, there is no hard and fast dividing line between levels of abstraction, Borders ean only. be drawn very loosely; and the number of slices "Notes on the Methodology of Comparative Analysis of Economie Activity,” Transactions of the Sizth World Congress of Sociology, 196, Inter- national Sociological Association, vol. TI, p. 109, * Ibid. "Bendix, “Concepts and Generalizations... .” p. 536, ‘THE AMERICAN POLITICAL SCIENCE REVIEW vou. 64 into which the ladder is divided largely depends on how fine one's analysis nods to be- Pires Sls ae sufiient, however, for ths purport fogical analye And ty Tanfor toners Ch connections vith what goes on atthe Upper td of the ladder, at thovervcal funebre at ‘hie we cos the border between medi level fener concepts and high fevel univers, The Sie may be formulated as fellows: ew far up can an ubereatonal term be pushed webu sat. denying reale? Ta pincpe the extension ofa concept should sot bo broadened beyond the pot a whieh at Test one relatively precise tomnetaton (op. erty of attribute is Wetaned Tn pence home over, the requirement of pute ientienton tay’ be too exacting. But even if no stoal Tostive identifestion ean be alfred donot tee how we etn renounce the requvenest af Agate Heniieation The rca dtnetin would ths be between 1) concepts defied by negation or ex adverso, i.e., by saying what they tre nat and 2) conept taihou negation, e-pposte cone, coneepions witht fp fed termination of boundries. The logeal Pn- Chl involved In ths ditincon te ont doer Iinatio est nepati, that i any determination Involves neption, Aeedag tthe ple the former concepts are, no tater hol Dead determinate; wheres the Iter are indeerai, "ate, Herly without termination TE tha pines apple tothe linking proce along a fader ef abstain, and poe uly to the point a which ME etgune ae turned oto HL univer inthe fe etanee we obtain empirical uve. whereas i te ‘econ instance we obtain univers whieh Lae expla alue—poeudosunivrs for tea eal scence. The reason for thas tat a concep tualifed by a negtion taps or may nt found to apply to the rel world whereas non-bounded concept always apples by def tion: having no special raingtom here way of ascertaining whether it apples Tote fel world or not An empirical univers fateh Tecate ie til points fo tpmctings where 6 non-mpiia univeral iadacreistay reins teeing (ar sy racer on th Bl “The group concept lends taf nicely ag an lusttn of the foregoing (other Gast Be diese in greater delat ater) and ven such tothe pot in that repeat lige se sap to meet the tneling pro tem of comparative pote In the sro then of polities (Bentiey, David ‘Truman, sada Ladin being the dbviou referenosy i's Se enough that "group" becomes an slLemraing category’ not ly an analyte coetrt (as