Download as pdf
Download as pdf
You are on page 1of 81
The Logic of Comparative Social Inquiry ADAM. PRZEWORSKI ne fone HENRY TEUNE Unie of Ponoia & KRIEGER PUBLISHING COMPANY MALABAR, FLORIDA BOTS ne See “Ares rorel Ne per of he bok may be probe ety Joomer yay amon ormechon mene cag hfaoon Sto andra sso ib pemason ming om Be ‘Libeary of Congres Cataloging a Publication Date “he pe ef compre ln. it: aay ptt New Yo: Way Saees hase. sw soma To our collaborators in the International Studies of Values in Politics PREFACE ‘The ideas presented in tht book developed during the course of et participation in comparative research project, the International Studies (of Values in Polite, which was a study ofthe relationship between cha ‘acteristics of local politcal leadership and the behavior of local gover ‘ments in India. Poland, the United States, and Yugoshva. It ito our alleges inthis research Sear, wth whom we works Facetnfice for 4 numberof year, tit we data thi book "Among the fist problems we coafroted were thse of measurement In the fall of 1965 the fst pretest of valu-sale items was under may ‘The goal of the pretest was to develop valid and reliable sakes of wales Jin four national samples Before the data were gathered we ancpated exremedifcalty i Menying st of items tat would fore satisfactory ‘cals ofthe same aloes in all four countries. In our dlicusions we for- mulated the idea of eonstcting sales from a combination of intrzationl tems which would ensure crow-national relablify, and national item, Which would contribute to valty i each country. Using diferent item for each county, however, could be jsied only ifthe tems were to be “equivalent.” Tas procedure was subsequcutly developed and put lished as “Equivalence in Crose National Research" (Public Opinion Quar= tery, Winter 1966). At that tine we cat our arguments in terms of con ‘ruc sales in croseaationa research. We did ot fully compretend the generality of our arguments for any type of comparative research and ‘the epistemological asumptios of such measurement procedures. Te cia poten we fa fc ro tom he med fo “comparable” measures of commits “actives in four countries. The dit hat coud be wod to ame "activener™ were cud the contol of the researchers. In fact there was not a sige em of data Gut was the sme in all four counts. Our previous arguments had to be modiied and generalized x Preece But the mest important problem was to traslte the data tht were being {theo nto a form amenable to theoretical statements, Were we 10 be ‘content with satements such as "ia Inia..y but in Poland"? ‘The Ata were acumulated at Owes fves: indivi! politcal Tears, heal government, and countries. The question was one of what phenomena Should be expleined by what. We fek tht comparative rserch should focus on withinsystems relationships rather than atibaes of systems ‘Thi idea hd far-reaching implications for ovr thiakng end perhaps was the key to assembling our thoughts fr writing this book. ‘Although these ideas would never have been developed without an ‘opportunity to confront problems of comparative research the Inteaaticl Staies of Values in Poldesshouid ot be identified withthe views peo- ‘Pounded in this book. We itd a geoerl treatment ofthe lgl of com Farative research fce from the operations and fadings o any particule ‘esearch program. The intellectual orientation ofthe took i based Oat {lef that methodological procedures shoul rely on explicit epstemology nue vew te weakness of books on loge of lngtiy is that they ae uly writen with ite understanding ofthe paeticalprober and limitations of ‘esearch, If such philosophy of science books ae understood and taken literally, eseatch is Ikely to be paralyzed. By contrat, many books on techniques of resarch ae so involved in presenting specie procedures and techniques that nether their juticadons nor thie implestons are realy of comparative research Adam Proeworski August 1969 Henry Teane UrRDQUCTION An Overview of Problems ant ont: Theory currex ont Comparative Research and Social Science Theory Explanation and Theory and Socal Since, Theory and Spatiotemporal Parameter: The Poste of Subsittbty. The Sus of Proper Names of Sytem Summary. curren two Research Designs * “Most Simic Sytem” Designs, "Moe Dieret Systems" Design Univarnte Comparisons. Comparing Relationships CHAFTER THABE System Level Variables: Changing the Level of Analysis, Diferng Kelatinsbips. Comparative Stody and eves of Ana SysemLevel Variables: Ditusion Pater, Stings, and Coens Level of Analysis and Inference: Inerpreing Ecological Correlations, Tnfereces ‘when With Sytem Relionsip are Similar. Inference ‘when Within Syam Relaonsipe Dilfer Syatomatilly. Coach. 1 rouR Formulating Theories Across Systems ‘aeons A Procedure. Comparative Explanation in the Soil Sener" A Conclaon. a” “ ” Contens pant two: Measurement Ive Measurement ia Comparative Reseach ‘The Language of Comparison. Ditet vets laferted Measurement Dret Mesurement in Comparative Researah Inferred Mesurement System Valiity fd Sytem Interference. Equant Measurement ‘Actos Sytems. Summary and Concisions cuarren so Establishing Equivalence ‘A General Coterion of Equivalence. The Place {ot Multiple Indicators. tdiatr Diatbuton Condions of Eautnhing Equivalence, Common Indestor: Ory. Common and Sytem Specie Indieatr. System Specife Instore Only ‘Summary and Concosns Implications fr Comparative Research a” na 1m 135 131 An Overview of Problems ‘The purpose ofthis book isto dentify problems encountered in one area of social scenic inquiry and 10 ofer some soktions to them. But the Identification of problems i not independent of the proposed sohtont both are bused of a set of assumptions about the accepted goals andthe posited model ofingiry. Since the sty of societies les at the cromrad ‘tseveal mndele-ofeience, thee autumptone ae of crucial importance, ‘Thecontoversies concerning mebodologieal and techrcal strategies in com- parative research often stem from dtsgreements about underlying astm tone eather than from dieences inthe evaluation ofthe “proeisty" of ‘kde stent pesve-viean agreed set of pontltes Only the Inter ype of controversies can bereaved inthe course of logical analy Controverie concerning the choice of dhe Basi postulates of sceace require explication of endelying metacienie citeria and are orally unsolvable. Thar the {uesion whether pics should be concerned with the measurement of, (rvity oe with actions of angele ie largely an haue of store scale | {ntellecvally only partially debatable, whereas the question wheter st Under an apple tee the best way to measure gravity ean be fly well decided” The question whether hitory should be concerned with every ‘moment in the life of George Washington or with the role of elites ia ‘atonal revoltons it again unresolvable, but the gucrtion of how 10 Aetemine whether George Marshalls report oa te life of Washington & a reliable source can be answered more easy. Infeed, if some basic post Utes are aeceped, the question of measuring the seapiim's superonty over the cherubim isan easier one to solve then the problem of whether tides “can” be meaesred. ‘Since diferent models of scence persist within and berween various beanches of the sidy of society, an attempt will be made to state sane 3 4 Inroduction sumptions of the set of recommendations dat will be presented. The extent to which we are aware of these assumption fs moet certainly Ine complete, bat hopefully even a patial explication will increase the i> telly ofthe problens and the recomended solutions Some Assumptions of Comparative Research ~The pivotal assumption of this analy tha social sekence research, Incoding comparative Ing. should and can lad to general statements about seit phenomena Ths assumption imple that aan o¢ soca tetavior can be explained interns of general hws esublated by ober- vation Intoduced ere as an expresion of pefrence, this astimpton ‘will nt be logically ast, Iti his Knd of assumption that accompanied the development of com- parative ingury during the last wo cena The corolry chs sun tin is that soil havioe conforms toa fmited name of recognizable Pater. One of the majo ptterts etd by soci scents sich a5 Comte, Mars, Duekheim, Webee, ad Spence i that societies undergo @ Severed proces of deslepment Contemporary crossocietal nguiis share many ofthe perspectives and sumptions of tho thinker, partsuarlyingutes that combine com> porate and developmental approsshes. Whether explicit o implicit, d= ‘elopmestal assumptions inderhe mich of the cxcrentcrosealsea fesurch Are mobility ote is developed soit higher o owe tha in tee developed societies? Ts the politcal cstere of “modern” pole systens more "patton thn tht i adona” sytem? Ie recat ‘ent t soc fle Based on the same ere in tedonel and moder socieies™ ‘The combination of comparative and developmental pecspctves is largely a consequence ofthe soil mie ef contemporary socal nee tnd, ously they need not be combined. We can vey well imagine asta of Ceylon, Poland, or the United States ha i devoid of developmental Sisunptions. We read states not about Cyl, however, but about "the ase of” Css, Poland, ot the United Stats 1 illsratins of deel tren proces aking pace in diferest paces and at dierent nes 1 is abo important to realize that any description —any desrgtive com Parave stiement—o «parca soca station «a “eases tase on “eB Misi “sees Uni i New i, TC of Cn” tment, cal Sete aa ie Formation: "The Case f Patan fn). Wi Inrodueton s some theory that treat the papular social system atleast a confining or diconfrmingiltration Een ostensibly descriptive historical studies Of particular socal sytems of instations often ssuame a whol set of several theoretical statement Nomothetic versus Idiographic Approaches ‘The orientation in the socal sinoes toward statements of theoretical ature hn been and mains a subject of controversy in sare dln, _Himorians argued whether ter approach should be Wographic or nome- het? anthropologists diseased whether functionally integrated systems ist be tented only in toto oe whether tris can be abstracted an com- pared seclologisis continue to disagree about the importance of the "postulate of indgpeasbity™™—whother every existing stroctue i fne- tional forthe sytem; economists argued whether universal definition of their icplne canbe constructed’ and political sietsts debated the valve of generalizations confined toa specifi se of insttatons* ‘Although the wadiona jxtaposition of the “iiographic™ and “vome- ete” seiences fas lst mich of is ee, many ofthe problems currently ‘dacuised ae tll claifabe within thse terms Socal tienen rem ott of philosophy, particule the philosophy of hory, and repereuaons {com the tational controversies an stl be felt. These controversies con een the model cf inguicy that is best applicable to social ecaty and the at of general statement i the soil sciences Aron recontrct this geld controversy ine following manner Histerial scence sno move the meve reproduction of that which has eon than satural since i the reroduetion of sate... But though tor, Hike natural sence, is reconstruction, i is a reconstruction “Hans Meer. 6 The Phonnhy of Hay Our Tone, Dob: ‘Yor 939; Lots Gotta ty Generate the Weta of May "Seite chon teen Rakiferown ad Lovie conering Ge sity fe ciney af fimo, Pata), A. R Raifledown, "Or Oe ome ‘of Puncon Ine Saal Semen” eran Anoop, 940 SM. Lipa Plier Mon Dovey, New Yok 1980, Memorial Append to Chapter Ov Para mméary se NJ. Ser, Nt ot he Methanlagy of Compare ‘Amat of ronnie ASt” Sarl Senet Information 6 196 "WiC Mac The Sih of Compare Covramen, Banbeey, New York, 1955;G. AL Almond, “Compare Fic! Stee" foun of Paitin, 19862 Hasty else "A Prete ce Compare Pol, Po nd Preset” fe Hay ‘skin and Dovid Apc ede Conpartie Pali, Pee Prom New Yor, 16 6 Inroduction cy diferent type. The ultimate im of natorl science isa systematined complen of vt... The wile abject of histony ‘a unio ver The ulimit subject of history isa unigue series of ‘Sus... Natural sien Secs Tor aw, tory for the_partelag [Nagel notes the same dtleence “Even a cusory examination of treaties in theoretical ratural scisce and of books on history reveals the prima face diference between them, that by and large the satements of the former are general ia form, and contin few i any rlerences to specie objects, places, times, whereas the Statements ofthe Inter are almost without exception singular and replete ‘wth proper names, date, and sographi sec "To ths exten, at the alleged contast Between the satiral sconces as nomochetis and history as idegrapie appeacs to he well founded."™ 1 this distinction i accepted, the form of statements in natural sciences is “whenever and wherever X occurs, X is in a certain relation to Y," wheres the form of socal scence statements specifies 1, spatietempor location and makes all subsequent propositions rative to tse parameters. Natural since statements say that "a particle to which no force is applied vil be movieg with constant velocity in a straight line” whereas social Sciences ate required to state tat “in Affic during the carly 1960's ‘cthicly sed partis entered stations of violent cone” A statement Spectyng that partles are moving int straight ine in Africa is tre, but redundant, Bechuse partiles move In the sme way everywhere and at all tunes A statement generalizing that "plial systems in which al partis are ethnically based are unstable” i judged to be "impeoper” Because Iris do not behave ina universal fthin. Statements of« general nature. fre thos mesningles because they involve conepts that are excessively fener and violate the specie features ‘of exch socal ster, The main argument agaist the posites ef nomesete statements concerns Soll ‘sly stems fom the observation that sch stnterents are never universally true; even if they appear tobe tue, they are meaningless insofar as they are Isa on general concepis that "do vielence” othe speci features af each f (= sgstem. ‘The contoversy ths concerns (1) the nature of general ‘eoretical statements in the social wcences snd (2) theatre of standard laed observations of social phenomena across euties. ‘Why are general statements concerning social reality not universally tre? Some ofthe ponuble reasons may sem frm. the easement in the observed arable,» problem also encountered inthe vara si {Bint Ree “The Lge of Moral Anaya an Meet 9 Ieroduction 7 Another eeaso is that selected ses of social variables are rarely auton ‘mons; there are alvays other factors that ear iauence any observed re- [abonshin Am indviduas polis! partisiation i aot completly dete rined by the level of edvcation; it extent i influenced by many more factors Friston sot the soe predictor of agpression, and tal pe ‘ception of husband and wife isnot the sole predictor of marital succes ‘Two interpretations ofthis incomplete determination of social phenomena possible, and they ae of eral importance for getral soa scence theory Tn one imerpretation, the incompleteness of a factorial stem (stem of sacabies) isthe only reason for indeterminatin: ial reevant factors were Anown, then the some mulvarate statement would yield 2 deterministic ‘explanation regardless of dine and space. This kind of an interpretation onnttes a direct trarserence of the dominant natural since model (2 the stay of socal realty. I al factors lading to agaresive betavons were Towa, thn a mulisarlte statement woul! expan aggression wherever an whenever osc. This view of soil reality impli eran research Strategy tat us fat, 20 one has followed: research conduct on random samples ofthe work population, regardless of the soca systems to which Indivicinls, groups, ce subsystems belong Socal scence tased on this [ind of assumption would be a pio ahstrieal: Historia eireunstaaces in'wbich parca obeerVtions were made woul! be ignored since they ate assumed to have no significance, Tas kind of model of social sience is Tplicei the writings of some psychologists and was actually posted bby Malewski as te proper model for corstructing general theories of be- Invioe? Hf this model is accepted, cece can be no bridge between theo- retical aid historical socal scence, Theoetal statements would be form ated in terms devoid f proper names of social systems, whereas historical atements woul iclode such names. In the second interpretation, ‘Scieace statements cannot be univerally te Because the interaction various characteristics within each social system creates unique, oa ‘atying patter of determination relative to each sora system, Therefore {de Mentietion of Ge socal system in which a given phenomenon occtes isa partof ts expansion. In the extreme version of thi postion no gener Statements applicable across social systems are possible, and all social Seicce statements must be confined 1 partclar socal systems. Is a ss ‘xtreme version, probly deminnt within pois! scence today, sti tiene statements are rlave to clases of naGons or “area” that share ‘syndromes of historia, cultural, and sil characteristics, ‘Thus one can A. Molen, ‘Dea Mode Solon” (Two Moles of Socios), Sidi SexjiopeneIei96h * Itrodeetion rake general statements concerning Latin America, Southent Asi, Eastern Forope but statements of s more general rare a peep “The asumptons underving the choice of ster model ef socal selene ace of am empscalnatore: they concsre the most Wkly pattem of de- termination of social phenomena, Tn one view, is more Lely tht this ‘termination of social phenomena = suienty geneeal to warrant am a ‘ice! abstiscton from spatiotemporal parameters Inthe other view, the cterminaton is expected to be highly speife to particular social toms ‘or groupe of soil spters so thatthe generality of satemests concerning ck realty ir inkerentyHimited. One can expt that te valdy of either fof thee perpestivs Is eitive to the nature of phenomeea under invest ftion a well asthe stato of knowledge The abistrial assumptions are Probably more ali with regard to payeholgial poceses, whereas re. Miva assumptions may eur oat tobe more valid in explaining social and ‘lal bohavior At long as thes asiumptions are explet frther de- Selopment of comparative inquiry wil gradually provide information on their validity. Systems versus Variables in Comparative Research Thor view, the crux ofthe problem ein the stn ofthe proper cares social systems within genoral theory. What does it mean that observations ‘of socal reality ate tlatve to particular scil units? The real diference Tween ths ssstmption ad the one dargarding social relativism is tht particular Secil units age weated as predictors, inthe Same manner as arabes are used as predictors in geveral theories The status of social Unis, osever, is aot the mime a the ator of variable. Te ot medley clear why the ientfeation of a partcular soil wit predicts a sven social phenomenon. The relationship between edvcation and pital Pcicipation may be difeert in Latin America and Avi, and consequently the knowledge that ths relationship i obverved in a certain aca ies ‘iin in petition of pla! participation “The goal of comparative research isto substitute names, of variables for the names of social systems, such as Ghund the United States, Ati, ot ‘Asia To use fist aa example from the aatral scences, suppose that @ roup of physicists Mente the temperatre at which water tol in New York Asother sroup conducted the same research ia Denver, and the results obviously turned out to be diferent. Ar idigraphicaly oriented physicist woul naturally conchae that all we can do isto deseribe New ‘York and Dever la tens of the temperature of bilng watt But thee Physics wouk! replace the spatial parameter by a varinble: They wok! Introduction 9 sate that the tempeatere of boiling depends upon atmospheric preste and Iwould darepard other charities of New York and Denver. Whea ‘modal scholars discovered thatthe rate of heart atacks is lower in Japan than inthe United States, they too replaced the names of sccal systems by ‘variable, apuing that the incidence of such attacks depends upon the eon ‘umption of plysturate ats. Ths vole of cmmperstive studi ia the veil Sciences was formulated by Benda, Suchma, and Ossowski. Bendix ‘Social realty may be infsitely diverse. This bel leads some to con- lode hat socal reaity can be “understood” only within the context in ‘which cis observed, and car never be explalaed by general like sate nents But these who accep the model of science can fo general stat tment belee that regards of the extent of sci diversi, i nonetheless ‘an be expres in terms of general theories "Tn our view the formuladon of general teorles is posible if and only i these theories take into account what appears tous to be @ pervisive Property of soca lity: soil phenomena are not only diver bt always ‘ectr in mutually interdependent and interacting structures, possess = spatiotemporal leation, If stable, these pattems of interaction can be treated a8 systems. Socil syste are composed of interacting cleren's, sich as indvideals, groupe, communities, instutions, or governnents ‘Whats important for comparative inquiry s that systems with which we condinarily deal, such ay sores, nations and eultres, are organized in (eam of several levels ef components and that the interactions within these stems are not limited to any particular level But eu ares these levels ‘Thus the behavior of any clement, suchas an individual depends not ely ‘pon his interaction with othr individuals but also jon is neracton with Insitutions such asthe church othe tat TE social phenomena are treated ax compifents of systems, two major implications follow. ‘The st is that de behavior of any component cf ¢ system is determined by factors ininse-to the stem and i relately ioleted from ifueces ouside of the stem. The fot tht havior aks Introduction a place within a celavely slated context may mean that a cetan proportion fof the explanation of this beavior may be founé among factors extrinsic to all sjstems—aniveral factor—and a certain proportion may te found among factors that ar intense to porticuar systems and not generalizable strom salons. Friseation may be 2 common factor explaining stant unrest i all systema bat in addon characteristics of pela systems, such as dhe structure of Columbia University, may also form a part ofthe xplanaton. Oar poston i that he characteris of purteuar systems, a be expressed as general variables, such as the presence or absence of sent participation in univenity decision-making. and as such would be applable across all systems. Te fac, whenever there isa system specie {actor that seems to be necessary for explanation, the concison should not ‘ye that systems are unique bot eater that ie necessary to ienty some several factors far not considered, This is indeed the primary function ft comparative inguey. ‘The second inplcaton of wea socal phenowens as components of systems tht specie observations mont be interpreted within the context fof specie systems, Phenomena become fats when Uey are exprested in tome language. The problem i thatthe same Inguge may not be apl= table acres all sets but may have to be ajisted to specie systems. ‘Ths isthe cena problem of comparative measurement: wo inconporate Ingo measurement Statement the context within which observations are nade We believe thatthe exiting langage of measurement i suficint to Permic contextual adjustments in measuemect Satmcnts- "Ties, thn, are the major problems in comparative inguiy: to is teoduce systemic factor into general, theoreti statements and to rea the syemic context of measurement statement The lngieal and method ‘lope! stategs for solving these problems consite the sbject of thi book, Part One deals with problems of theory and Prt Two with problem ‘of measurement re PART ONE, Theory CHAPTER ONE Comparative Research and Social Science Theory Explanation and Theory n Social Slence. Theory and Spaotemporal Pavomaters:The Postale of Subsitucabi. The Saas of Proper Names of Stems. Suramar. While» precise dfiition of comparative resarch wil not be presentod immediately, we assume In thi Book thatthe gel of social scence isto explain social phenomena. We further postulate tat the geneity and parsimony of thoodes should be gen primacy over thle accuracy. In ‘ther words socal sence theories, rater than expan PAEROMENE as fcouately as possible in terms relative to specie historical cisimstances, Should attempt to explain phenomena wherever and whenever they O=¥, “Allhough he poultoe is not new, this chapter writen withthe bell that the implications ef this preference for the conduct of comparative research wil become clear whea the assumptions undelyng this cee are Understood, We sal fist dicus the asumptons underlying the conti tion of general and patsimorious theoeiss and, second, recommend som ‘procedares of dat-pnthering and cits snasss in comparative re ‘We wil argc thatthe bridge between hitorially anchored observations and theoretical statements cen be found in comporstive sadie and that _eneral theories eannot be constroctd in the soil eiener without explit ‘ference to factors operating atthe Ivel of systems, Explanation and Theory in the Social Sciences Tn 1984 in Bennington, Vermont, small businessmen supported the radical igh more often than salaried employes wih the sme education, Te 1963 ia Poland, wonce who hd had premarital sexual experince pe ved ther marriages as more succesful Oar those who had had no pre= n is Comparative Research and Socal Science Theory ‘marital experiences. In 1988 in Finland, persons with congruent sats {corresponding evel of income, education, and occupation) voted fr the ett ports more often than persons with incongroent stats. ‘Thee deliberately selected and disparate findings of soca scientists can ‘one day find thei way into histor books and become a part ofthe human heritage in such documents a the book rlting the eating habits of Freeh men dorag the second haf ofthe eighteenth century or some future work reporting on the daly ife in Vermont in the middle of the twentieth tury. The way in which se fading are formulated makes ws thik ‘ofthe fate historian” most social sini re more ineesed in fining ‘ut why social phenomena oocur thas where and when. But sll obsrations lth soeopaiieal rain ae anchored in ine and space, Tei in Vermont, Poland, and Finland that these oeervations are rade They ae made a 4 certain tne and in acerain ple, and if we were rote concerned with Tistorical seaity than wih thoretal generality, we would never extend the findings beyond the particular spatiotemporal parameters within which the observations wene wade. The concer with bldg general theories of human behavior constitutes eich an overriding goal cf socal scene, however, that they ae willing to sk the error of ase seaeralizations rather than give up that concern? Ths we learn that “Iratation brings ag- gression.” Cut “outgroup hostty breeds ingroup solidarity,” that “the sparty of culty lgitinate goals and means ress in anomie," and thats high level of ecapomie development is necessry fora stable demo- cratic politcal ester.” ‘The goal f scence isto exphin and prodit why certain events occur when and where they do. Why di small bosinssmes in Vermont supe McCarthy in 1984? Why was te Kowalski mariage nt succesful? Why {id Smith commit a crine? Why did Napoleon attack Russa? Science it ‘concerned with tho explanation of spesieoveat by means of statements that ae invariant rom one set of circumstances to another. But what ‘oes it ean #0 “explain” or “predis” » concrete, specie ever? Since a dscusion of “explanation” exceeds the limits ofthis work, we {erat Gamers in the Soil Scec”Paih Sioa Ble. 196, rilcn Rowe oe potem he owan wy" ‘aim itm adn he esha) nveated wit te poses oe ‘itodsoe of sce e ray wet he propos te seer ato Teton tte anl'space Yat hy tecome teste loa e eae Explanation and Theory im he Social Sciences 9 ere eet cant rat Mise ty typecast Ten otha ofl pre Sr mre Sane pe el eat aan asasrving icon oe Cama thegunasrerel te aig tie ny ad secaeernceatanira es nates M'Rougt is a worker and works ina large factory and 15 young (24 eas old). ‘The second premise consis of «conjunction of general statements describ- Ing with high likethood the behavior of skilled workers, employees of Tare factories and young persons. (No interaction is assumed) (ne out of every two workers Yotes Communist; ant employes of Jnrge rganiatons vow Commtnist more often than employes of small organizations, and young people vote Commanst more “7 than dee people. 7 Therefore, tly that ‘ai 'M, Roaget votes Communist ‘This explanation i incomplete. The probability of @ French worker, 24 smployed in large factory, and voting Communist is sil fr from 1.00. Several other factor, such a5 pice of resigence, marital stay, fathers ‘cccupation, relight, and So forth, might have 10 be consiered ifthe explanation (prediction) of M. RougetSs behavior were to approach certainty, Most explanaons In the social sclenees are incomplete in the 56.0, Hemp darcs of Sint Expention and Other snr in the Phone yo} sence ree Prom, New Vor, 198 Se araarty ele aah PE 20 Comparative Research and Socal Science Theory sense dat the probability of the explained phenomenon taking place docs ‘ot approach 1.00 (a 2r0). Since the rules of inference ate peobabil wwe canaotexpet Ost, even ifthe premises ae tue, the conclusions wil Invariably follow, As te probability of inferential rls inceases, however, the probability of predicting» property abo inceeaes—it moves away from ‘what could be expected random. “The genera statements that serve as premises nan explanation constitate theory. Unally more than one general atement is necessiry to provide 2 felatvely complete explanation. These statements mst have certain logleal properties: they must be inercomnected, and none of tei impics- ‘ions cam contrat any over implction. Some rues mest slo be avall= abl o determine whether specie event is an element of the cas eovered by the theory. In other words, Giese sutements mst be empirically lterpreable. Finsly, the set of wich general statements shook! include & formal deductive framework, sch that the inferred consequence 5 nt a Intutivelycbvious ees ofthe premises. ‘The logical structure of thers, however, ot our ceneal concer ‘The accumulation of knowledge consists of the process of gradual cop. ‘imation and/or modifeation of the veces Ut serve as the geyal premises in the explamtor) scheme. If we fe (0 understand “what p> pete in the las lection ia Denmark” why was there drop ia sucies In Southern Tas” or "what made Joe deop out of schoo” we must have allie et of general theories sulcienty confirmed to provide a reason thle certiny Ut when these general statements are applable, the expested consequences wil follow. [At any sage in the development of science it is lkely, however. that ‘more than ons theory wil explain the same class of event. Therefore ‘dona goals are powtlted that eovide erteria forthe evaluation of Uheoeaes accraey, generality, parsons, and causality Fst, we expect a theory to be accurate, t0 exphin at completely as possible ad t prodict ar much of the variation ss posible. This erterion fan be expresicd in terms ofthe aout uP yriance account by the Independant variables the mote variance accounted for by a. heory, tbe salle the err of peedvtion. For comparative research ths eiteion im pls that the pal of socal cince tory ito exphin » given phenomena Se accurately ex posible ch socal yvem. Te we wanted t explain the incidence of divere, for example, ke wou! contract tele: dat would ‘minimize the esor of peadtion in each social eytem. We might Bnd that {none society 999 peroet ofthe variance of divorces can be explained by the edseaton of spouses, Unie religion, and the degree of rgkty. Ta another veal stom, we may again be able to account for 99.9 percent of Explanation and Theory inthe Socal Sones 2 the variance of dvores, but withthe use of diferent factors, for example, ‘mutual perceptions ofthe sponses, extent of premarital sexual experience, fd Sextal stsfacton in marrage. The to theories wil be maximally Accurate. They will provide a nearly complete explanation of divorce in ach society But they will not met other requirements Imposed on (heocies:accursey,giaerality, parsimony, and causally When the accuracy of theories is maximized, their generally and paci= mony will often below. Geneaity of «theory refers to the eae of $i ‘phenomena to which i is applable, The greater the generality of & theory, the greater the range cf phenomena that can be explana by the theory. For example, one theoretcalpropostion may state that “edicaion fs eelted wo poll pareipation” ‘This theory proves an exphnatn one typeof political activity af adividal, but only one te of ative. Duc what are We saying when we leafy edveaticn at a determinant of ‘behavior’ “What is “education”? Ts it he numberof questions concerning diferent branches of knowledge one is able wo atswer? The darstion of proteetion ofan individual by hi fly or bys ciety? Reinforcement of Ceztan behavioral patterns by appropriate rewande? Chemical changes in the composition of brain cll? aur of more general theories can be formulated ifany of these defnitios replaces the definition of education in terms of suber of school years completed. For example «more gener theory may state that individuals who lave been protected by society over an exes petod of tine are more key to participate in socal activites, ‘Snce attending school wsslly provides mich protection and since pliteal participation i a type of seal scivity, the origin kee general theory can be dadoced from the more general one. The development of natal scence consists of more general theories supeseding less general theories. AS Hempel points out “When a sintie theory is superseded by another inthe sense fe which ‘tasial mechanics and eeewodynamiss were superseded by the spe theory of relativity, Chen the succeeding theory wil generally have a wider explanatory rungs including pheremena the ester theory could act a ‘count for; and wll asa rule provide approximative explanation for the empirical laws inp by ite predeseno.™ 1 social scence, however, i is not always apparent that a Tess geneel theory canbe deduced fom « more general theory even if both are avait 2 Comparative Research and Social Science Theory thle. Both reward theory and theory of political participation may poten- tally explie why an individual vous in am eletion; bu its ot apparent thatthe explanatory role of education can be deduced from the romard ‘theory or any other psyetolopcal theory In the example discussed shove, 99.9 percent of variance of divorce is predicted in each social yer, but nota single statement can be made that ‘would be tra for both systems Sach an explanation will not only Ise scaealty but also wil not be parsimonious. The soar the sumber of factors providing for a complete explanation of 2 sven class of events, the more parsimonious the theery* Tt would be an interesting experiment 10 compare the inerpeatins given to findings derved from particular ‘countries withthe interpeetatoas of similar dat frm various counties ‘Why ere third parte never spectaculany successful in the United Sates? ‘Because ofthe bipartisan tadition. Why were thed putes never successful in Great Britain? Because ofthe stifin the compotion ofthe abo force {hat took the working-class support away from the Liberals and gave it to Labour. But when we confront the two questions simaltaneosshy—why were thind partes never pareuily sutceatul in ether couetry—the snuver wil tend tobe formulated in terms of factors common 10 both such as the electoral system. To the extent tht diferet theores—each involving a diferent st of independent varables—are used fr ferent social sjstems, the formulation ad vesting of general theories in the social Seseees snot possible ‘The erteria of generality and parsimony imply tht he same theories ‘must be evaluated in diferent sysiemie settings and that social scence theories cam sn contemation only 1 theories formulated tn terms of the ‘common factors const the point of departure for comparaive research ‘We recogni. however, that in some sitiations accuracy in a particular social sytem maybe the most important value For example, f we wae! to rediot election rents inthe United States, we probably will not be coa- ‘cerned with the same factors that wil predict election results in Great Briain, But the goal is to provide understanding as 0 why people deity ‘with poltal parties, then generality and parsimony wil be more important than system specific ecuracy. If the role o theory ist provide immediate Bema ones no nt wo of re or 8 a Sein Kehna Rent Stow, Plat natal Phony, Rone, igen aa Landon 96. Explanation and Theory inthe Soci Sciences 2 _gulelnes for socal practic, then accuracy In 2 speife socal systema may be the most important value® ‘The fourth citron imposed on theory concerns causality. We can think cf cauaity ina twofold pespectve, Cusaiy is a property of «sytem fof variables, A system of variables Is sid to be cau to the extent that (1) the dependent vate ie not “overdetrmined’—ro two variables within the spter expan the same part f the variation ofthe dependent ‘Phenomenon apd (2) the system of variables i slated —the explanttry Itern"does ot change when new variables are added The extent 0 Which «thoes caus at is, ie extent to which general premises are Invariant, inreases a the number of fete incorporated into the theory inereset Jn terms of comparative research, the postulate of easly iat pls that factors operating at diferent levels of analsis—groups, como Dit, region, atlons, te—should be incorporated into theories and tht their ineaction with the factors operating within cach of thee systems should be examined This pariular model of theory ab general. parsimonious, and causal set of statements eased thoughout the ret Of tht book, We are nt arguing that this porticulae model ensues the oa or eve the best redel of theory, but that if this model accepted, then crose-systemic studies rust become an inepral part of theor-bulding and theor-testing lave sto cron No geen hey wil ld pete peti, Si cite neo ply ee wig the poem an ag ‘come fuente mong neh setae Tllotng Sint arc ou "Caaf (rans and Toesabley in My G- Hand and. Koopman ee. Sader nly Tp service The astomption of uncrees ees mal sated bor not {Seed "te seems dat Hempel come dont to sepeting nepal tt of option in potting ti erro of inal pes” fp, 9. ) Tis ‘eon imple tht chit ephatins sald be apt! tematey and ten Pa Comparaive Research and Social Science Theory Theory and Spatiotemporal Parameters: The Postulate of Substituiability Tate ltcoduction we defined some general eves underlying te alleged ‘contradiction betwen historically based observations and absityforma- Intedstntements A question often discowed by stadent of society is ‘whether historically anchored observations soul be treated as specie to particular social sytems or whether general theories, fre of spatchemporal Parameter, ean be developed and sted, As jaded arin, the ie to Toager appears in this extreme form. The problem no isto defize the coa- ditions ander which general theories can be developed ard the proederst that are appropiate fr the developmect and testing of general theories, “The etreme version ofthe cctv srgamat allows ms Way of Bridging storied and thoretial statements, Once % proposition is sated in historical terms, esing such proper names as Ghana, Hitkr, or British ‘workers during tb 1950s, could be incorporated aly iat a theory coa- Sting of more geneal historical propositions containing such nares a ‘Arie, German leaders, cr the Brit popalation dung the 1980s If how ter, & propeition i ted ia terms foe of proper same, i cat be i> corporated ool into theories consisting exclusively of such propositions For example, the proposition that “economic criss give rise to charismatic. leadership" cas be generalized ito statement tht "all res give Fie to charismatic lendereip, “his disieton between historical and theoretial generalization clrfx the ateretive medes of Ceory construction Any se of observations ca be generalized in one of two Ways, depeadiag upon whether historical of thooretial generality is sought ut ths dinstion is only anabticl Actually spatiotemporal propositions (e. measoremeat statement) cam be geneeaed theoretically, and general statements can be speciied ise torieally. On the one hand, the statement that “Hier was » charismatic Jeadeeho came to power in Germany asa rel ofa ers can be geaerae rod ito a theoreti proposition relating cies apd charisma leadership. (On the other hand, the natement that “criss give ris to ehaciamati lead. er may have Been observed tobe true only in Africa. Thus the observed ‘storisal staan is net unigue, but nite isthe genera theoretical state rent univers, Histol statements are imply theoretic "They sb time under the proper nares ofthe socal systema a broad range of factors that might be used in theoretical explastion. But theocetial statements will generally iclode historical componeat. As Joag as Africa difers from the ober parts of the Worl, tora analy i no loger posible, and therefore the name of social sjtem will aye to be wsed in explana tion, Thor “snigue” factors can nether be the only ones nor can they be Theory and Spatiotemporal Parameters 2 totally discarded in theoretical analyse THEy ae redefined, rather, as the "sideom of theoredcal explanation, ‘The bridge between Witorieal observations amd general theory 9 thé subuition of variables for proper names of social ayer i the ccurs of \Leomparaive research. The theretial importance of this atemert is best understood in temas ot Hempel’ requirement that clases of evens referred. ton heoretial lawlike")staements be essetilly generalizable. Hemel aegud “Surly hike sentence must not be locally limite to afte number of instanecs: it must not be ogiallyequiaeat to fie conjunction of Singur sentence, of, riety, must be of ésenallygeneralid form” r Focexample, a satement tht “all Mercans are taller th all Americans” isnot a lawlike sentence that can be used forthe explanation of the beight of Mexiéans or Americans. This sentence is lgicaly equivalent to eens jimetion of satemests giving all symmetrical relations tetween individual ‘Mexicans and Americans and it cannot sustala counterfsctual and sub junctve conditional statements such aif Mr X. whois an American were | Mexicas, he would hive been tale.” The sentence, “Persons living a ‘warmer climates are invariably tlle haa person living la code limates however, isnot a conjunction of any finite numberof satementsconcerting individuals; can, atleast logically, be exended ad infiitim. In ote words, le statements are posible iv the socialsciences if and only tf Spaioteiporal parameters are treated as resides o} variables potently ‘conan 1 the explanation ‘Te postulate of substitutability conceros the ctologial sats of such concep a8 “group” “organization,” “culture” “nation,” and “potieal systen"—the ontological status of systems encked within some specie, spatial and temporal parameters. Before preceeding to the dicuson ff this postulate and its implication, we will aalyae more cksely the notion” stocclly losted socal systems," or “spatiotemporal parameters.” Historical socal system” or “all patitempora part motes obviously defines the maximal levee of generality for any statement. ‘Nothing can be more general tan always and everywhere This concept defines the entre popultion of conditions within which observations of social phenomena can be sae, and any particulars of observations is 3 sample, random or not, of ths population, We can concee ofthe set of tintorcaleircumstances ofthe most general mature tht contains all sora social systems or spaotemporal parameters This set becomes denumerable I additional assamprioss are mide, wnat ont 7 26 Comparative Research and Socil Science Theory (One such assumption is that tis et contains “stems relatively folated by some fastor”™ For example, human telags allegedly shared! some cha setertes aot shared BY anlaas. Some historical fact, wheter it was the tlevation tothe biped oc the acquisition of the supersgo, defined the " atively Bolted spsten?™=—homan being. This (pe of tolted system of hmans is often asumed by psybologlts without regard for other isolating factors, Systems ean be oat onthe bases of ll ind of historical events that determined any one of tei cormmon characteris. Classifications of civllatlons provide an example of systems isolated with regard to come ‘sic eatural nfvence, suchas "Judeo-Christian," "Sanskrit" and “Inca.” ‘Any demimeration in tems of relatively iolted systems is, obviously, hypothetial one. Crosse research must demonstrate that diferences within those systems are indeed smaller has the dleences among thm ‘Anather way denimeraing the tof all spatlotemporal parameters con sigs ‘of figing some eutfl point inthe past, seh a8 ting of counties thot existed after World War TL This set fist defined temporally ard then denumeated in terms of eountis or “rations.” Ta che light of th Sscossion of “eately olated systems,” itis cleat at there are many ways of denumeratiag the set of all historical cirumstances. The ab Termatives ange from denimeraing i as a et consting of one element, ‘he animal to denumeatin io terns of any obervable subsystem. TE ve lint the temporal parameters and confine the spa dimension (0 ations we would be she to emumerte thse systems that constitute the ‘universe in erosnatonsl research, The problem of "uniqueness versus enealiy” in cremations stadies conceras, therefore, the entlogcal etn of soch proper names as Mexico, Ghana, Australia, or Yugoslavia The Status of Proper Names of Systems In Germany and Sweden, betterpuid workers are likely co be clas consclous, whereas in Bria, the United Stats, and Alstal, theless well= paid workers ae clas concious” How can we interpret ths finding? In one inerpretton this fing 2 "tora! generation.” sie it species the spatiotemporal parameters and encloses the statement about Telationshipe win those parameters. In aroberiterpetatio, this td tea "general proposition” ating tht the relationship between income of ‘workers and fhe extent of ther clas-conscooenes depends Upon Some he pain of ley ited stam wa nada by .Osowat “Peo Cepceine of Hire! Generations” The Pah Soon Dalai 9,168 “eure Beal end SN Lipset "The Fei of Poise Ssblon” in L.A cost ede Pall Soil, Hager an Row, New York 96,3 The Stats of Proper Names of Stems n the ‘other factors rot yet considered. The nature of the cours might noe provide «cht as to what these factors might Be Consider the example of the elatonsip between motivation w earn a foreign nguage and grades among the sideas of varios deparimeats at the sme university. In some departments such a relationship can be ‘observed; In others ic cannot. Should te latrpretation of ths finding be {hat “atthe University of Warsaw during the 1960s, here isa telationship tetwecn motivation apd grades inthe English department, but notin the French department” or thatthe retatonship between motivation and grades

You might also like