PART I. Methods and Fields of Comparative Politics

Transcription

1 PART I Method and Field of Comparative Politic _Chap01.indd 11 11/29/ :05:54 AM

2 _Chap01.indd 12 11/29/ :05:55 AM

3 1 The Ditinctivene of Comparative Reearch Charle C. Ragin and Claude Rubinon INTRODUCTION Social reearch i inherently comparative (Lieberon, 1985). Reearcher compare the relative effect of variable acro cae; they compare cae directly with one another; and they compare empirical cae with counterfactual cae. But the comparative method ometime referred to a mall- N comparion contitute a ditinctive approach to undertanding ocial phenomena. Frequently, comparative method are portrayed a a bridge between qualitative, cae-oriented reearch and quantitative, variable-oriented reearch. Thi interpretation i certainly valid. By embracing apect of both qualitative and quantitative method, comparative method can circumvent ome of the limitation of both approache. But comparative reearch i not merely a bridge, for it ha many ditinctive feature and trength. We begin thi chapter by reviewing the conventional view of comparative method a imultaneouly qualitative and quantitative. The moderate number of cae employed by comparative reearcher allow them to engage in the development, teting, and reviion of theory traditionally the province of cae-oriented reearch, a well a hypothei teting and theory adjudication uually een a the province of variableoriented reearch. But the greatet trength of comparative methodology arie from it ditinctivene. Fundamentally et theoretic in nature, comparative method preuppoe particular epitemological and theoretical perpective (Ragin, 1987; 2000; Rubinon and Ragin, 2007). Although ocial reearcher conventionally conceive of ocial reality in term of tendencie and probabilitie, ocial cientific theory like comparative reearch i predominately et theoretic in nature. Frequently, however, even comparative reearcher do not recognize the et theoretic character of their work. In the econd part of thi chapter, we explicate the et theoretic nature of a number of claic comparative tudie. We identify and illutrate three type of et theoretic relationhip and dicu how they form the bai of three form of comparative analyi: decriptive, contitutive, and caual. Next, we dicu the cae-oriented nature of ocial reearch _Chap01.indd 13 11/29/ :05:55 AM

4 14 THE SAGE HANDBOOK OF COMPARATIVE POLITICS The et theoretic orientation of comparative reearch invoke a cae-oriented perpective: et are compoed of element (cae) and comparative reearch i the analyi of how cae in one et relate to cae in another et. We then dicu the iue of caual complexity. Comparative method are epecially welluited for the tudy of how combination of caual condition produce particular outcome. The tudy of neceary and ufficient condition a prominent concern among comparativit i but one apect of comparative caual analyi. We conclude our dicuion of comparative reearch by examining it formal method. Reviewing two contemporary application of comparative methodology, we illutrate the contruction and analyi of truth table. Truth table form the foundation of comparative analyi; whether explicitly or implicitly, mot comparative reearcher contruct truth table. We demontrate how Mill (1875) method of agreement and difference a well a Ragin Qualitative Comparative Analyi (QCA) the two dominant implementation of formal comparative method make ue of them. QCA build upon and extend Mill method, a we demontrate through analye of caual complexity and counterfactual cae. COMPARATIVE METHODS AS A BRIDGE In decribing the methodological landcape of the ocial cience, it i conventional to ditinguih between quantitative, variableoriented analyi and qualitative, caeoriented analyi (Ragin, 1987). There i, of coure, no inherent reaon that variableoriented analyi mut be quantitative and cae-oriented analyi, qualitative (Rubinon and Ragin, 2007). Still, there i a natural affinity, which i a conequence of the way in which the number of cae influence the reearch proce. Quantitative technique require a large number of cae the more, the better o a to meet model aumption and enhance tatitical power (Cohen, 1988). Faced with hundred or thouand of cae, however, it i impoible for reearcher to know the detail of each cae. A the cae become obcured and retreat to the background, variable advance to the fore. Large-N analyi, then, tend to focu on variable and their relationhip. The fundamental goal of variable-oriented reearch i the production of decriptive or explanatory inference (Brady, 2004). Decriptive inference are produced by generalizing from pattern found within ample (King et al., 1994). All ele being equal, the larger a ample i the greater the reearcher confidence in generalizing to a wider population. Explanatory inference are produced through hypothei teting (King et al., 1994). Hypothei teting require a well-pecified theory of the relationhip among variable, which may be confirmed or refuted by comparing the theory prediction againt evidence. Again, all ele being equal, the larger the ample, the greater the reearcher confidence that a relationhip found in a ample doe, in fact, exit in the wider population. A both benefit from a large number of obervation, the affinity between variable-oriented reearch and quantitative method i mutually reinforcing. Cae-oriented reearch and qualitative method, by contrat, are mot ueful when applied to a mall number of cae. Becaue qualitative technique leverage the reearcher in-depth knowledge of cae, every additional cae require reearcher to further divide their attention. Examination of detail highlight the ditinctivene of each cae. While impoing limit on generalization and thereby hindering hypothei teting, thi focu facilitate theory development (George and Bennett, 2005). In-depth cae knowledge make it eaier to ee which cae apect are relevant to the quetion at hand and how thee apect fit together. Thi undertanding may be ued to contruct new theory or revie exiting theory, thu generating new hypothee for future teting. Thi i not to ay that cae-oriented reearcher cannot engage in hypothei teting; _Chap01.indd 14 11/29/ :05:56 AM

5 THE DISTINCTIVENESS OF COMPARATIVE RESEARCH 15 indeed, popular example of cae-oriented reearch include crucial, mot-likely, and leat-likely cae tudie that tet whether a theory operate a predicted (Ecktein, 1992). In general, however, reearcher who want to develop new theory tend to ue qualitative, cae-oriented technique to examine mall-n, while thoe who want to tet theory tend to apply quantitative, variableoriented method to large-n (Ragin, 1994). A conequence of thi bifurcation i that ocial reearch i characterized by a large number of tudie that examine either mall-n or large-n, but relatively few tudie that examine a moderate number of cae (Ragin, 2000: 25). Comparative reearch can bridge the divide between qualitative, cae-oriented reearch and quantitative, variable-oriented reearch. Like cae-oriented method, comparative method maintain the integrity of cae; like variable-oriented method, comparative method examine pattern of relationhip among variable. Comparative method, then, may be ued for both theory development and hypothei teting. With a moderate number of cae (uually around 5 50), it become poible to examine cro-cae pattern while till attending to the detail of each cae. In comparative reearch, theory development and hypothei teting interact in two way. Firt, comparative method may be ued to develop, tet, and revie a particular theory. Second, comparative method may be ued to adjudicate between competing theorie. Developing, teting, and reviing theory Comparative method encourage a reciprocal relationhip between theory development and theory teting. In a trictly qualitative caeoriented tudy, reearcher enter the field armed only with enitizing concept, which they ue to help them contruct new theory a they try to make ene of their cae. In a purely quantitative variable-oriented tudy, reearcher begin their reearch armed with a well-pecified theoretical model and hypothee regarding how change in one variable affect change in other. In contrat to thee two extreme, comparativit typically begin their reearch with a rough idea of the concept, variable, and cae that are likely to be relevant to their reearch quetion. Becaue comparative reearcher typically tudy a moderate number of cae, it i not feaible to ue a purely exploratory approach and conduct an open-ended, in-depth examination of every cae. At the ame time, conventional hypothei teting fail due to the limited degree of freedom available. Intead, comparative reearcher eek to anwer their reearch quetion by examining the fit between concept and cae, idea and evidence. The notion of fit i key. For comparativit, a good theory i a middle-range theory that fit the evidence well (Mjoet and Clauen, 2007). Such a theory will identify which variable are relevant to the quetion at hand, explain how thee variable are related to one another, and, pecify the context under which they operate. Through invetigation of the fit between theory and data, comparativit dicover area for adjutment and improvement. Ultimately, the reulting theory mut be judged on it own merit. Doe the theory provide a compelling explanation of the oberved cae? Doe it explain unoberved or previouly anomalou cae? Doe it produce obervable implication and novel inight? I it falifiable? There i alway a trade-off between pecificity and generality, and theorie vary in their explanatory cope. Very pecific theorie may not generalize beyond the oberved cae; very general one may not add anything new to our undertanding. Ultimately, whether any particular theory i ucceful depend upon triking the proper balance between pecificity and generality for the reearch quetion at hand. The advantage of uing comparative method to develop, tet, and revie theory i that they make thee choice, conideration, and deciion explicit _Chap01.indd 15 11/29/ :05:56 AM

6 16 THE SAGE HANDBOOK OF COMPARATIVE POLITICS Adjudicating between theorie A popular ue of quantitative method i to adjudicate between competing theoretical perpective. Given two or more theorie that purport to explain the ame phenomenon, reearcher identify the variable pecified by each theory and etimate a erie of model uing the ame dataet. By examining meaure of tatitical ignificance, explained variation, and model fit, reearcher can determine which theory bet explain the outcome of interet. Comparative method may be applied to thi purpoe a well. While the proce tend to differ, the logic i the ame. A the goal of hypothei teting i the generation of explanatory inference (ee above), the contruction of the et of relevant cae i crucial. Becaue comparativit tudy fewer cae than quantitative reearcher, they typically devote much more attention to the tak of delineating the et of relevant cae and then contructing their dataet. Of particular concern are cope and poibility condition. Scope condition pecify the condition under which a theory i relevant (Cohen, 1989). Cae that do not meet a theory cope condition are conidered irrelevant, regardle of whether they exhibit the outcome of interet. Skocpol (1979), for example, limit her theory of the caue of ocial revolution to countrie without a recent hitory of colonial domination and thereby exclude the cae of Mexico and Vietnam. Poibility condition pecify the condition under which an outcome i poible (Goertz, 2005; Mahoney and Goertz, 2004). For example, there are many (in fact, infinite) cae of the non-occurrence of ocial revolution. When the outcome of interet i clearly impoible, the cae i irrelevant for the theory at hand. Irrelevant cae hould be excluded from the data et becaue their incluion doe not benefit hypothei teting and, indeed, may hinder it (Braumoeller and Goertz, 2002; Clarke, 2002; Mahoney and Goertz, 2004). Mahoney and Goertz (2004) codify the pecification of poibility condition that i, whether cae are conidered relevant or not a the poibility principle. Given an initial et of cae, the reearcher firt examine cope condition in order to exclude cae to which the theory doe not apply. The reearcher then applie the poibility principle in order to retrict the ample to relevant cae. Mahoney and Goertz (2004) operationalize the poibility principle a a rule of excluion and a rule of incluion. The excluion rule, which take priority over the incluion rule, direct the reearcher to develop a lit of robut eliminatory variable which predict the abence of the outcome. Cae are excluded a irrelevant if their value on any eliminatory independent variable predict the nonoccurrence of the outcome of interet (Mahoney and Goertz, 2004: 658). The incluion rule tate that [cae are relevant if their value on at leat one independent variable i poitively related to the outcome of interet (Mahoney and Goertz, 2004: 657). We regard thi threhold a a preliminary guideline and recommend that reearcher ue their theoretical and ubtantive knowledge when pecifying incluion threhold. The proper application of cope condition and the poibility principle to a population will produce a theoretically and ubtantively relevant ample of both poitive and negative cae. If the reulting ample i too large for comparative analyi, then reearcher hould elect repreentative cae that maximize the diverity of cae on relevant caual condition. To adjudicate among competing theorie, the comparative reearcher examine each cae to ee if the relationhip between caual condition and outcome hold a predicted by each theory. Theorie are evaluated according to how well they predict both the preence and abence of the outcome. Like conventional quantitative adjudication technique, comparative technique of adjudication can indicate opportunitie for theory refinement by expoing the way in which different theorie complement one another and might be combined _Chap01.indd 16 11/29/ :05:56 AM

7 THE DISTINCTIVENESS OF COMPARATIVE RESEARCH 17 Example: Berg-Schloer and de Meur (1994) Condition of Democracy Social cientit have a longtanding interet in the rie and evolution of democracy and have developed a wide array of theoretical explanation for it emergence, peritence, and breakdown. Oberving that there have been few attempt to adjudicate among thee theorie, Berg-Schloer and de Meur (1994) tet eight of the mot prominent model predicting the ucce or failure of democratization: Dahl (1971, 1989), Hermen (1941), Linz (1978), Lipet (1981), Luebbert (1987), Moore (1966), Sani and Sartori (1983), and Vanhanen (1984). Berg-Schloer and de Meur (1994) contruct their dataet by drawing from the inter-war period in Europe, emphaizing the imilaritie among the cae: For a tudy of the chance and failure of democracy in a comparative perpective, the interwar period in Europe provide a unique etting, ince the cae to be conidered hare many common ocioeconomic and political-cultural characteritic. Their hitory i relatively well reearched and well documented. The time period under conideration i clearly demarcated by common event, the two world war which ignificantly altered both the internal and external political landcape and et them apart from earlier and later development. All cae can initially be deignated a parliamentary democracie, ome of them relatively well etablihed, other more recent, and ome exiting more in form than in ubtance. They were ubequently affected by a common external timulu, the world economic crii of the late 1920 and early Some parliamentary regime urvived, while other yielded to authoritarian rule and, in particular, facim. Prevailing aumption concerning modernization and progre, whether liberal or Marxit, were everely hattered (1994: 253). The theorie under invetigation were developed primarily through the examination of Wetern European countrie, and it i not clear that they hould alo apply to non- Wetern ocietie. In order to atify cope condition, Berg-Schloer and de Meur (1994), therefore, retrict their ample to 16 Wetern European countrie. The limited time period eek to atify poibility condition. All of the countrie in the ample entered the period a parliamentary democracie and were ubjected to the ame hitorical event. That ome emerged a democratic and other a authoritarian ugget that each country poeed the potential for democratic breakdown. Expanding the ample beyond the mall handful normally examined in comparative reearch provide a tet of the generalizability of the theorie: all of the major breakdown cae with their pecific pattern and the major urvivor, including ome of the maller countrie which often tend to be overlooked, are conidered (Berg- Schloer and de Meur, 1994: 254). At the ame time, the moderate ample ize permit Berg-Schloer and de Meur to interrogate the individual cae when eeking to make ene of reult. Indeed, to thi end, the reearcher exclude certain countrie from the ample: cae like Denmark and Norway have not been included becaue they were found to add relatively little variation over and above the condition and relevant factor for a cae like Sweden (Berg-Schloer and de Meur, 1994: 254). In contrat to conventional quantitative method, the incluion of additional cae in comparative reearch i of no benefit unle they erve to better repreent the combination of condition preent in the population: Berg-Schloer and de Meur 1994 ample i repreentative not of Wetern European countrie during the interwar period per e but, rather, of the type of countrie that exited in Wetern Europe at that time. For each theoretical model, Berg-Schloer and de Meur (1994) examine whether the countrie in their ample conform to the model prediction of democratic urvival veru breakdown. A model i confirmed when it correctly predict the urvival or breakdown of democracy. The attention to individual cae permit the reearcher to identify cae that partially upport or contradict a model hypothei a well. Comparing the uccee and failure of each theory, Berg-Schloer and de Meur find the tronget upport for, firt, tructural model that emphaize ocioeconomic condition and, econd, agency model that emphaize the _Chap01.indd 17 11/29/ :05:56 AM

8 18 THE SAGE HANDBOOK OF COMPARATIVE POLITICS action of individual actor. Intitutional and meo-level model are only weakly upported. Complex model incorporating nine or more independent variable were more likely to produce mixed reult, a conequence of the diverity preent among cae. Applying QCA (Ragin, 1987) to implify their reult, Berg-Schloer and de Meur (1994: 276) find the mot parimoniou olution to emphaize baic factor like democratic legitimacy and the political role of the military (a with Dahl) together with ome actor-related apect like intervention by member of the upper cla (for Linz). Perhap the mot ignificant finding, however, i that the emphai on hitorically important cae uch a Great Britain, Germany, Italy, and Spain ha, in fact, hindered undertanding of democratic peritence and breakdown: But it can alo be een that countrie like Finland, Czecholovakia, Greece, and even France are hardly in line with the expectation of any of the theorit and in everal intance provide direct counterfactual example. Thi reult point to the often rather limited perpective in theory building of ome of thee author a far [a] geographical ditribution and hitorical depth are concerned (Berg-Schloer and de Meur, 1994: 276). Bridging the world of cae-oriented and variable-oriented analyi, comparative method make it poible for Berg-Schloer and de Meur to adjudicate among competing theorie, examine relation among explanatory variable, and identify cae that contradict pecific theorie. THE DISTINCTIVENESS OF THE COMPARATIVE METHOD The greatet trength of the comparative method arie from it ditinctivene, not from it facility for bridging variableoriented and cae-oriented analyi. Social reearcher traditionally conceive of ocial reality in term of tendencie and probabilitic relationhip. That i, we generally frame our obervation in a contingent manner: Given certain condition, a given effect i likely to occur. In thi view, ocial reality i inherently random and all ocial procee embody a tochatic component that cannot, even in principle, be identified, much le modeled (Goldthorpe, 2000: chapter 7). The mot formal verion of thi approach are found in probabilitic model uch a regreion analyi that produce precie prediction of the likely effect of one variable upon another. In contrat to the tendential conception of ocial reality, comparative method ee the ocial world in term of et and et theoretic relation, epecially thoe that are conitent with argument of caual neceity or ufficiency. Thi view motivate a earch for invariant or at leat highly conitent connection between caual condition and outcome. Set theory in comparative reearch Although the tendential view of reality dominate ocial cientific dicoure, mot ocial cientific theory invoke a et theoretic notion of reality. For example, when reearcher oberve that religiou fundamentalit are politically conervative they are arguing that religiou fundamentalit form a rough ubet of the et of political conervative. Another example: Eping-Anderen (1990) propoe a et theoretic model of the nation-tate. Liberal, corporatit, and ocial democratic countrie are each a ubet of the welfare capitalit countrie; together, they contitute a complete et: The Three World of Welfare Capitalim. Capitalit countrie are not necearily welfare tate, however. Eping-Anderen ue the term welfare a a modifier, indicating that welfare tate are a ubet of capitalit countrie. And capitalit countrie are, in turn, a ubet of all countrie. Type of et theoretic connection The exitence of a et theoretic relationhip indicate that ome type of connection may link ditinct phenomena. Thi connection can take _Chap01.indd 18 11/29/ :05:57 AM

9 THE DISTINCTIVENESS OF COMPARATIVE RESEARCH 19 one of three baic form: decriptive, contitutive, or caual. Conider, for illutration, the opening obervation that motivate Weber invetigation of The Protetant Ethic and the Spirit of Capitalim: the fact that buine leader and owner of capital, a well a the higher grade of killed labour, and even more the higher technically and commercially trained peronnel of modern enterprie are overwhelmingly Protetant (1958: 35). The puzzle that motivate Weber analyi i explaining the fact that people in thee occupation contituted a rough ubet of the et of Protetant. Hi obervation i not imply that there i an aociation between two variable, but that a pecific connection between certain occupation and Protetantim i remarkably conitent (they are overwhelmingly Protetant). He goe on to query: [W]hy were the ditrict of highet economic development at the ame time particularly favourable to a revolution in the Church? (Weber, 1958: 36). Weber note, in effect, that thee highly developed ditrict contitute a ubet of thoe oppoed to the Catholic Church that again there i a remarkably conitent, almot uniform, connection. A firt preented by Weber, thee two et theoretic connection are primarily decriptive in nature becaue he ay nothing about why thee connection exit. Weber goal in The Protetant Ethic i to unravel the puzzle of the et-theoretic connection between capitalim and Protetantim. Defining the pirit of capitalim a the methodical, continuou puruit of profit, Weber begin by elucidating the relevant characteritic of modern capitalim: rational calculation, entrepreneur, credit market, the eparation of work and home, and double-entry accounting (Weber, 1958: 17 22). 1 Each of thee condition, however, ha exited before and, therefore, cannot be olely reponible for capitalit exchange. It i only with the rie of the labor market the rational capitalitic organization of (formally) free labour (Weber, 1958: 21) that thee condition take on a modern, capitalitic character: For without the rational capitalitic organization of labour, all thi, o far a it wa poible at all, would have nothing like the ame ignificance, above all for the ocial tructure and all the pecific problem of the modern Occident connected with it. Exact calculation the bai of everything ele i only poible on a bai of free labour (1958: 22). Weber model of capitalim, then, i a combination of ix eential condition. Such a model i contitutive: the element of the model are cloely intertwined and together contitute rational capitalim. For example, double-entry accounting i alo found in the Far Eat, the Near Eat, and in antiquity (Weber, 1958: 22) but contribute to the continual puruit of profit only when combined with free labor. Whenever a et-theoretic connection i interpreted a evidence that omething i an eential component, apect, or part of another, it may be viewed a contitutive. Whether a et-theoretic relation i een a evidence of a contitutive veru a caual connection i a matter of interpretation. For example, ome might interpret the fact that the advanced indutrial ocietie are uniformly democratic a evidence that economic development i caually ufficient for democratic government. Other, however, might avoid making a caual argument and interpret thi ame connection a evidence that having a democratic government i an eential part of being an advanced indutrial ociety. The key ditinction i in how the connection i undertood and interpreted. Contitutive analyi i central to what Ragin (1992) ha referred to a the proce of caing. Reearcher engage in caing whenever they attempt to delineate the conceptual boundarie of a cae or et of cae. Reearcher often engage in caing a they attempt to identify conceptually the obervation included in their analye. In The Protetant Ethic, Weber primary concern i the caing of rational capitalim. He ak, What i rational capitalim? How i it different from other economic ytem? Caing eek to anwer the quetion, If I ee it, how will I know? What are it eential element? _Chap01.indd 19 11/29/ :05:57 AM

10 20 THE SAGE HANDBOOK OF COMPARATIVE POLITICS An important apect of caing i that it identifie the theoretically alient characteritic of the cae. That capitalim involve exchange for gain i not alient becaue exchange for gain exit everywhere, at all time (Weber, 1958: 17). Exchange for gain may be central to rational capitalim but, theoretically, it i not deciive. The eparation of work and home, however, ha much greater theoretical alience. Although the patial eparation of place of work from thoe of reidence exit elewhere (Weber, 1958: 22), it i only under condition of capitalim that thi eparation force worker to eek their livelihood on the labor market: The tendency everywhere ele wa for acquiitive enterprie to arie a part of a royal or manorial houehold (of the oiko), which i, a Rodbertu ha perceived, with all it uperficial imilarity, a fundamentally different, even oppoite, development (Weber, 1958: 22, emphai in original). Thu, contitutive analyi eek to identify the interconnected component of a type of cae it eential feature. It produce a litmu tet for memberhip in the et of cae under obervation. An obervation i excluded if it lack any of the eential feature or diplay any feature that i prohibited. Having outlined the characteritic of modern capitalim, Weber oberve that modern capitalim firt took root in Protetantdominated region. That i to ay, the et of region where modern capitalim aroe are a ubet of the et of Protetant countrie. To etablih a caual connection, Weber eek to link the rie of Protetantim with the rie of modern capitalim: Montequieu ay (Eprit de Loi, Book XX, chap.7) of the Englih that they had progreed the farthet of all people of the world in three important thing: in piety, in commerce, and in freedom. I it not poible that their commercial uperiority and their adaptation to free political intitution are connected in ome way with that record of piety which Montequieu acribe to them? (1958: 45) To etablih cauality, reearcher mut pecify the mechanim by which memberhip in one et i connected to memberhip in another. Weber doe thi by linking Luther conception of the calling to the pirit of capitalim. By decribing how the ene of calling combined with a worldly aceticim led Protetant to embrace capitalit production and exchange, Weber detail how memberhip in one et (people poeing the Protetant ethic) connect to memberhip in another (people poeing the pirit of capitalim). Note that Weber doe not claim that all Protetant embraced capitalim, nor doe he claim that all thoe who embraced capitalim were Protetant. That i, Weber doe not argue for the exitence of a perfect ubet relationhip at the individual level but rather an affinity between the two. The two rough ubet relation decribed at the outet of thi dicuion are evidence of thi affinity. Set theory and within-cae analyi Social reearcher ue the term cae tudy in a variety of way (Ragin and Becker, 1992). Frequently, comparative reearch i ubumed under the term (e.g., Gerring, 2007: chapter 2). However, we find it ueful to ditinguih between within-cae analyi (i.e., ingle cae tudie) and cro-cae analyi (i.e., comparative tudie) a each of thee form of analyi entail ditinct reearch method. Within-cae analyi focue on ingular cae. Many reearcher feel that becaue ingle-cae tudie lack a comparative element, they are not real reearch. We diagree. Cae tudie are particularly ueful in interpreting culturally or hitorically ignificant phenomenon (Ragin, 1994). Cae tudie of event uch a the torming of the Batille (Sewell, 1996), the funeral of Abraham Lincoln (Schwartz, 1991), and the rie of the Englih working cla (Thompon, 1966) are valuable becaue they apply ocial cience theory and concept to the tudy of the caue and conequence of important moment of hitorical tranformation or qualitative change. Moreover, cae tudie are not necearily non-comparative. There i a crucial ditinction between the unit of analyi and the unit _Chap01.indd 20 11/29/ :05:57 AM

11 THE DISTINCTIVENESS OF COMPARATIVE RESEARCH 21 of obervation (Denton, 2007). Unfortunately, ocial reearcher often ue the term cae to refer to both. A King et al. (1994) point out, a ingle cae tudy may involve many obervation. Thompon (1966), for example, ue a variety of method to examine and yntheize data from a variety of ource in order to explain the emergence of Englih working cla concioune. Today it i fahionable to refer to uch reearch a triangulated or multi-method (Brewer and Hunter, 2006; Reinharz, 1992), but cae tudy reearcher have alway leveraged a variety of data ource and analytic technique in order to confirm their finding and make ene of their cae. A cae exit in pace and acro time, implicit if not explicit comparion are inevitable. Thompon tudy cover 50 year and pan the whole of the country. He invite comparion of the Englih worker by including eparate chapter on field laborer, artian, and weaver. Likewie, hi examination of worker repone compare variou form of reitance, protet, and working-cla radicalim. By the concluion of the text, Thompon ha reviewed the cultural and tructural tranformation between 1780 and 1832 that gave rie to cla concioune among Englih worker. Thompon project may be undertood a a comparion between the working cla of 1780, a cla an ich, and the working cla of 1832, a cla für ich. Finally, within-cae analyi i an eential component of good cro-cae, comparative reearch. It i for thi reaon that ome reearcher ubume comparative reearch under the moniker of cae tudy. Good cro-cae comparion necearily involve the invetigation of individual cae on their own term. Indeed, a common method of preenting comparative reearch i exemplified by Barrington Moore, Jr. (1966) in Social Origin of Dictatorhip and Democracy. Moore pend the bulk of the text reviewing the individual cae (that i, conducting within-cae analyi) and offer ytematic comparative analyi at the concluion. It i here that Moore contruct hi three route to modernity; that i, hi three et of modernizing revolution. The et of bourgeoi revolution culminating in democracy include the cae of the French Revolution and the Englih and American civil war. The et of conervative revolution culminating in facim include the cae of Japan Meiji Retoration and Bimarck unification of Germany. Finally, the et of peaant revolution culminating in communim include the Ruian and Chinee revolution. In turn, thee three et of revolution contitute a ubet of what might be termed modernizing revolution. Hobbawm (1967) oberve that [t]he author of a comparative analyi doe not compete with the pecialit; he exploit them and may have to quetion them. (p. 821) Within-cae and comparative tudie are complementary. The et theoretic nature of comparative reearch require the conduct of within-cae analyi, a et are all about cae the element contained within et and the characterization of cae in term of et memberhip. Caual complexity in comparative reearch Conventionally, reearcher do not preent their caual argument in term of et relation but, rather, in term of how caue come together to produce outcome. Moore (1966), for example, peak of three route to the modern world, and Skocpol (1979) dicue ocial revolution a the product of tate breakdown and peaant revolt. Neither author make their et explicit, and frequently it take ome effort to dicern the et relation. However, cloe analyi of mot comparative work reveal a conjunctural undertanding of cauation. In eence, caual conjuncture involve interection of condition, which in turn can be undertood a et interection. Moore (1966), for example, identifie five condition for the development of democracy _Chap01.indd 21 11/29/ :05:57 AM

12 22 THE SAGE HANDBOOK OF COMPARATIVE POLITICS 1. The development of a balance to avoid too trong a crown or too independent a landed aritocracy. 2. A turn toward an appropriate form of commercial agriculture. 3. The weakening of the landed aritocracy. 4. The prevention of an aritocratic-bourgeoi coalition againt the peaant and worker. 5. A revolutionary break with the pat. (pp ). Each of thee item involve a proce of tranformation: development, turn, weakening, prevention, break. When thee procee are found together, they contitute a bourgeoi revolution which i one form of modernizing revolution. However, Moore unit of obervation i not the modernizing revolution that i hi unit of analyi but, rather, individual countrie. Countrie that experienced the combination of thee procee, uch a England, France, and the United State, are found in the et of countrie that experienced bourgeoi revolution. Countrie that did not experience each of thee procee, uch a India which lacked a revolutionary break with the pat, are not found in the et of countrie that experienced bourgeoi revolution. Moreover, India i not found among the et of countrie that experienced modernizing revolution of any form (bourgeoi, conervative, or peaant), which according to Moore explain it long-term tagnation (Moore, 1966: chapter 6). Skocpol ubtitle A Comparative Analyi of France, Ruia, and China immediately reveal countrie a her unit of obervation. Searching for neceary and ufficient condition of ocial revolution, Skocpol uncover important ubet relation. Firt, countrie that experience ocial revolution are a ubet of countrie that experience tate breakdown or peaant revolt. Second, countrie that experience tate breakdown combined with peaant revolt are a ubet of countrie that experience ocial revolution. The firt relationhip in which the outcome i a ubet of the caue identifie a relationhip conitent with neceity; the econd relationhip in which the caue i a ubet of the outcome identifie one that i conitent with ufficiency. The tudy of neceity and ufficiency i a longtanding interet of comparative reearcher (Goertz and Starr, 2003). A caue i neceary when the et of cae exhibiting the caue (e.g., tate breakdown) include the entire et of cae exhibiting the outcome (e.g., ocial revolution), The preence of tate breakdown i a neceary condition for the occurrence of ocial revolution. A caue i ufficient when the et of cae exhibiting the outcome (e.g., ocial revolution) include the entire et of cae exhibiting the caue or, more commonly, a combination of caue (e.g., the combination of tate breakdown and peaant revolt). The combined preence of tate breakdown and peaant revolt i a ufficient condition for the occurrence of ocial revolution. Complex condition of neceity and ufficiency can combine to form what Ragin (1987) call multiple conjunctural cauation. Multiple conjunctural cauation exit when a ingle caual condition i neither neceary nor ufficient to produce the outcome on it own but, rather, only in combination with other caual condition. Re-examining Gamon (1990) ocial protet data, for example, Ragin (1989) find that challenging group (ocial movement organization) ecure new advantage for their contituent if they atify any one of the following combination of caual condition. 1. No bureaucratic organization, middle or mixed trata contituent, no help from outider, and acceptance by antagonit. 2. Middle or mixed trata contituent, non-diplacement goal, and help from outider. 3. Bureaucratic organization, lower trata contituent, non-diplacement goal, and acceptance by antagonit. 4. Non-diplacement goal, help from outider, and acceptance by antagonit (pp ). While any one of thee combination of condition will reult in the challenging group ecuring new advantage, there i no individual caue common to all combination. Mackie (1974) refer to uch caue a _Chap01.indd 22 11/29/ :05:57 AM

13 THE DISTINCTIVENESS OF COMPARATIVE RESEARCH 23 INUS caue becaue each i an Inufficient (not ufficient by itelf) but Neceary component of a caual combination that i, itelf, Unneceary (due to multiple path) but Sufficient to bring about the outcome. METHODS OF COMPARATIVE ANALYSIS Comparative analyi can be formalized through the contruction of truth table, which how the connection between different combination of condition and an outcome (Ragin, 1987; 2000). Although conventional dataet may be ued in their contruction, the column of a truth table do not repreent variable, per e, nor do the row repreent cae. Rather, the column of a truth table repreent et and the row repreent relationhip among et, pecifically, all logically poible interection among the relevant et. Thee interection may be undertood a repreenting logically poible combination of caual condition. Comparative reearch i the tudy of the connection between combination of condition and outcome. Whether implicitly or explicitly, comparative reearcher contruct truth table when they examine how caual condition relate to pecific outcome. To illutrate the contruction of a truth table, we review Brown and Bowell (1995) analyi of how plit labor market affected interracial trikebreaking and olidarity during the 1919 teel trike. After conducting cae tudie of 16 northern citie, 2 Brown and Bowell (1995) ue comparative method (pecifically, crip-et QCA) to identify three caual condition that explain four form of interracial trikebreaking. 3 To implify our dicuion, we review jut one of their outcome: the preence of black trikebreaking in the face of white worker olidarity. Brown and Bowell (1995) begin their analyi by contructing a ample of citie that participated in the trike: Our unit of analyi i the teel-producing communitie where relevant variation in race relation occurred. Citie in the analyi have the following characteritic in common: the population wa over 25,000, the teel indutry wa an important employer, black worker were a ignificant part of the teel workforce, and the recruitment effort of the national committee reached each location during the organizing drive. We elected cae from Foter (1920) account of the organizing drive and include only thoe location that actually participated in the trike. The final group of 16 cae range in ize from ingle-indutry town to major multi-indutry citie. (1995: ) By conducting a cae tudy of each city in the ample, Brown and Bowell produce the data et hown in Table Recent black migration i an indicator of whether the city ha recently experienced a large increae in it population of young, ingle, black male. A city i conidered to have a weak teelworker union ( Weak union ) if the union had previouly uffered a failed trike againt US Steel (the primary target of the 1919 teel trike). Political repreion exit when local government align themelve with capital rather than labor. Thee three caual condition were not the only condition that Brown and Bowell (1995) examined. Their initial truth table included only two condition, Recent black migration and Weak Union. However, thi configuration created a contradiction (Ragin, 1987: ): five citie Buffalo, Chicago, Gary, Johntown, and Cleveland all experienced recent black migration and a hitory of trong union; however, the firt four citie exhibited black trikebreaking while Cleveland did not. Contradiction indicate an inadequately pecified model (Ragin, 1987) becaue identical condition hould lead to imilar outcome. To reolve contradiction, comparative reearcher reexamine both their theory and their cae. It may be that an error wa made in the proce of caing and that the contradictory cae() doe not properly belong to the population under invetigation. More commonly, the reearcher uncover an additional caual condition that explain the contradictory reult. In the analyi at hand, Brown and Bowell (1995) conidered three variable that they _Chap01.indd 23 11/29/ :05:57 AM

14 24 THE SAGE HANDBOOK OF COMPARATIVE POLITICS Table 1.1 Caual condition and preence of Black trikebreaking (from Brown and Bowell, 1995, Table 5) City Recent Black migration Weak union Political repreion Bethlehem No Ye Ye No Buffalo Ye No Ye Ye Chicago Ye No Ye Ye Cleveland Ye No No No Decatur No No Ye No Eat Chicago Ye Ye Ye Ye Gary Ye No Ye Ye Johntown Ye No Ye Ye Joliet No Ye Ye No McKeeport No Ye Ye No Milwaukee No Ye Ye No New Catle No Ye Ye No Pittburgh Ye Ye Ye Ye Reading No Ye Ye No Wheeling No No No No Youngtown Ye Ye Ye Ye Black trikebreaking upected might affect the connection between black migration and trong union, on the one hand, and black trikebreaking, on the other: city ize, whether the city wa dominated by US Steel or Bethlehem Steel, and local government repreion. The addition of the third condition, local government repreion, yielded a truth table free of contradiction and alo wa confirmed a caually relevant in their cae tudie. A truth table conit of one row per logically poible combination of condition. A truth table, then, ha 2 k row, where k equal the number of caual condition. A the number of caual condition increae, and the ize of the truth table grow exponentially, analyi become increaingly complex. Software package uch a fqca (Ragin et al., 2006a) can help to manage thi complexity; neverthele, mot practitioner examine between five and ten caual condition. To contruct the truth table, the reearcher refer to the previouly pecified dataet and record which combination of caual condition are aociated with the preence of the outcome and which are aociated with it abence. 5 Brown and Bowell (1995) final truth table i preented in Table 1.2. Brown and Bowell (1995) find that all but one of the poible combination i repreented by at leat one city; of thoe, two combination are aociated with the emergence of black trikebreaking. The caual combination that i not repreented in the ample i referred to a a remainder (Ragin et al., 2006b) and i identified by marking the outcome with a dah ( ). Remainder are common becaue ocial phenomena are Table 1.2 Truth table for the preence of Black trikebreaking, derived from Brown and Bowell (1995: 1505, Table 5) M U R Y Cae Eat Chicago, Pittburgh, Youngtown Buffalo, Chicago, Gary, Johntown Cleveland Bethlehem, Joliet, McKeeport, New Catle, Reading Milwaukee Decatur Wheeling Note: M = recent black migration, U = weak union, R = political repreion, Y = black trikebreaking _Chap01.indd 24 11/29/ :05:57 AM

15 THE DISTINCTIVENESS OF COMPARATIVE RESEARCH 25 characterized by limited diverity (Ragin, 1987: ; 2000). That i, many of the logically poible combination of caual condition imply do not exit in reality. (We addre the iue of limited diverity, epecially it relation to counterfactual analyi, in greater detail below). Mill method of agreement and difference The claic technique of comparative methodology are thoe propoed by Mill (1875). The implet of all comparative logic, the method of agreement i alo the mot likely to lead to faulty concluion. The logic behind the method of agreement hold that any given outcome will, inevitably, have a ingle caue. By examining a ample of cae exhibiting the ame outcome, a reearcher eek to identify the ingle caue that all cae have in common. Ragin (1987: 36) extend thi logic to encompa combination of condition, arguing that the ucceful identification of a hared combination of caually relevant condition would alo atify the method of agreement. Applying the method of agreement to Brown and Bowell (1995) truth table (Table 1.2), reveal that all intance of black trikebreaking occurred in citie characterized by recent black migration and political repreion. According to Mill method of agreement, then, the reearch will conclude that the combined preence of recent black migration and political repreion caued black trikebreaking. The proce by which the caue produce the outcome remain to be explained. In the abence of a well-pecified theory, uch explanation are generally bet developed through the application of within-cae method uch a proce tracing. Mill indirect method of difference what Ragin (1987: 39) refer to a a double application of the method of agreement leverage negative cae (that i, cae that do not exhibit the outcome) in order to trengthen the reearcher concluion. The reearcher firt applie the method of agreement to the poitive cae exhibiting the outcome. Upon identifying a caual condition (or caual combination) aociated with the outcome, the reearcher examine the negative cae to ee if the abence of the outcome i aociated with the abence of the caue. Referring again to Brown and Bowell (1995) truth table (Table 1.2), the method of difference reconfirm the initial concluion that black trikebreaking i a product of recent black migration and political repreion. Of the five configuration that exhibit the abence of trikebreaking, none exhibit the preence of both recent black migration and political repreion. A the indirect method of difference comprie two application of the method of agreement, the ame two limitation apply to both method. The firt the threat of an omitted variable i common to all comparative reearch method and, indeed, all ocial reearch method. Reearcher may draw the wrong concluion when important caue are omitted. If the identified caual condition i, itelf, caued by an antecedent condition, the reearcher explanation will be incomplete. Spuriou relationhip raie a imilar type of iue. If both the identified caual condition and the outcome are caued by an omitted third condition, the reearcher may identify the wrong condition a cauing the outcome. In thi latter ituation, one hope that the error will be uncovered a the reearcher examine hi or her cae and when attempting to elucidate how the condition caue the outcome. The econd and more ignificant limitation of the method of agreement i that it i incapable of addreing multiple conjunctural cauation. A Mill (1875), himelf, emphaized, when an outcome ha more than one caue, the method of agreement will not be able to identify any of them: That ome one antecedent i the caue of a given effect becaue all other antecedent have been found capable of being eliminated, i a jut inference only if the effect can have but one caue. If it admit of everal, nothing i more natural than that each of thee hould eparately admit of being eliminated. (1875: 474 5) _Chap01.indd 25 11/29/ :05:57 AM

16 26 THE SAGE HANDBOOK OF COMPARATIVE POLITICS Qualitative Comparative Analyi Qualitative Comparative Analyi (QCA, ee Ragin, 1987; 2000) build upon Mill method in two fundamental way. Mot ignificantly, QCA permit the analyi of multiple conjunctural cauation, addreing the greatet limitation of Mill method. Moreover, by employing counterfactual analyi, QCA permit a more nuanced analyi of the relationhip between caual condition and the preence and abence of outcome. Rooted in et theory, QCA ue et algebra alo known a Boolean algebra to analyze caual configuration. Boolean algebra provide operation for analyzing and manipulating et. The mot ueful of thee operation for comparative reearch are dicued in Ragin (2000: ) and include et union (logical or, repreented a addition) and et interection (logical and, repreented a multiplication). Truth table configuration are repreented a Boolean equation in which an uppercae letter repreent the preence of a condition while a lowercae letter repreent it abence. In Boolean notation, Row 1 of Table 1.2 i repreented a: M U R which indicate that The preence of black migration in combination with a weak union and political repreion i ufficient for black trikebreaking. Jut a with linear algebra, a hortened notation may be ued for multiplicative term, and the above equation alo may be repreented a: MuR Row 5, which i read a The abence of recent black migration in combination with a weak union and political repreion i ufficient for the abence of black trikebreaking, i repreented a: mur Y Y y Truth table reduction The analyi of multiple conjunctural cauation involve a proce known a truth table reduction or Boolean minimization. A reduced or implified truth table reult in a Boolean equation that expree the variou caual combination that are aociated with the preence of the outcome. Ragin (1987) detail the complete minimization proce; here, we provide only an overview. 6 The baic proce of Boolean minimization i derived from the propoition that invariant, or cloe to invariant, connection exit between explanatory condition and outcome. Each row of the truth table repreent a logically poible combination of caual condition. For each combination that i aociated with the preence of the outcome, the truth table reduction proce eek to eliminate logically redundant term. In Table 1.2, the firt and third configuration are repreented by the equation: MUR MuR When a pair of equation i identical except for a ingle caual condition, the ditinguihing term may be conidered irrelevant and can be eliminated. The minimization of thee two equation, therefore, produce: MR which i the ame reult a wa obtained through the application of Mill method of agreement: the preence of recent black migration and political repreion reulted in black trikebreaking. QCA method are a uperet of Mill method. A more complex example i provided by Stokke (2007) analyi of haming a an attempt to ecure compliance with international fihing agreement. Shaming involve an attempt to change a party behavior by expoing their violation to other (in thi cae, the international community). Having Y Y Y _Chap01.indd 26 11/29/ :05:57 AM