Comparative Case Study Methods in Urban Political Development

Abstract

Over the past decade there has been a concerted attempt among a growing group of authors to bring together the political science subfields of urban politics and American political development (APD). In this paper, I look at specifically how the comparative study of different cities and urban areas might contribute to this intellectual project, beginning with a brief illustrative comparison of Philadelphia and Montreal. I then place that comparison in the larger context of recent literature in postcolonialism, assemblage, and planetary urbanization, which I use to establish what I call an aggregation strategy for constructing variables—or, alternately, for denying the very existence of variables. I then suggest how my aggregation strategy could improve upon urban regime analysis, and inform new directions in studies at the intersection of urban politics and APD.

1. Introduction

My goal in this paper is to make a few conceptual contributions to the effort to meld the fields of urban politics and American political development (e.g., Lucas 2017b; Fortner 2016; Weaver 2016; Stone [1993] 2015; Ogorzalek 2018; Rast 2012, 2015, 2019; Taylor 2019; Dilworth 2009, 2016; Dilworth and Weaver 2020). Largely as a means of illustration, I begin with a brief comparison of urban political development in the US and Canada, and more specifically in Philadelphia and Montreal, in which I attempt to discern four kinds of comparisons between different time periods in different cities. I then place my comparison in the larger context of urban studies, and in particular the recent literatures on postcolonialism, assemblage, and planetary urbanization, which I use to establish what I call an aggregation strategy for constructing variables—or, alternately, for denying the very existence of variables. In the fourth part of the paper I suggest how my aggregation strategy could improve upon urban regime analysis, and in the fifth and final part I argue that both urban regime analysis and American political development (APD) suffer from a tendency to attempt to define cases in the absence of variables.

2. Comparing US and Canadian Cities

Philadelphia and Montreal were two of the string of settlements established along the eastern seaboard of North America in the seventeenth century and they followed roughly parallel trajectories, emerging as important nodal points for trade, then finance, manufacturing, and medicine, among other things; and eventually being eclipsed in size and importance by other nearby cities. There are also, of course, significant differences: Philadelphia and Montreal were products of the different colonial policies and strategies of Britain and France, respectively; the fact that Montreal was populated by mostly Catholic French colonists while Philadelphia’s most influential early settlers were British and Quaker set the two cities on somewhat different trajectories; as did the fact that each city’s growth depended on different types of trade (for instance, fur in Montreal and coal in Philadelphia) and thus different linkages to surrounding territories and populations.

There are enough similarities between Philadelphia and Montreal that we might assume they are both examples of the same developmental phenomena. The differences between them point to the causal factors that explain the dynamics of that development, in the form of the question, “if X happened in Philadelphia but not in Montreal, what happened in Philadelphia to cause X that did not happen in Montreal?” This question, expanded into a comparison of all Canadian and US cities, has largely driven a slow but steady stream of literature, from William Munro (1929) to the 2014 special issue of the International Journal of Canadian Studies on Reopening the “Myth of the North American City” Debate (Good 2014).

These comparisons have, for the most part, used time as a constraint in the sense that the things of interest that happened in one set of cities but in another either occurred or did not occur at roughly the same times. Thus, for instance, Zack Taylor (2014; 2019, chp. 3), who has provided a comparison of US and Canadian cities that is perhaps the most informed by historical institutionalist sensibilities, compares the variable institutional configurations of the two sets of cities during the decades prior to the US Civil War (when general municipal laws were adopted uniformly in Canada and unevenly in the US), the decades after the Civil War (when urban political machines developed in cities in the US but not in Canada), the Progressive era (when provinces took a greater role in urban affairs while states largely withdrew from policymaking in cities), and the ensuing Great Depression (when all of these divergent trends were reinforced).

As with the example of Taylor (2014; 2019, chp. 3), comparisons that attempt to explain the differences between all or at least most cities across two or more countries almost by definition have to look at factors exogenous to those cities, such as larger economic depressions, national cultures, industrialization, or higher-level laws. By contrast, comparative case studies of individual cities are more capable of comparing factors that could be considered endogenous to those cities, and which do not necessarily characterize other cities in the same country.

It is precisely the examination of endogenous factors, and more specifically the endogenization of “friction among multiple political orders” (Lieberman 2002, p. 701) that has been suggested as one of the driving forces in the study of APD. Yet, the downfall of the strategy of comparing factors endogenous to specific cities is that those factors are potentially unique to each city and thus incomparable. For instance, the ability to answer why Philadelphia is almost entirely unique among large North American cities for having a city-owned gas company (Philadelphia Gas Works 2020) requires an examination of factors obviously endogenous to Philadelphia. Asking why Montreal is supplied gas by a private company while Philadelphia is supplied by a city-owned company simply leads to the uninteresting answer that Montreal is not Philadelphia.

Most factors endogenous to specific cities are likely not so unique as to defy comparisons. And while they do seem more likely to be the result of an idiosyncratic mix of factors that makes comparison across cities more difficult, they also have the potential to make those comparisons more exploratory, thus suggesting new theories rather than confirming existing ones, in the same sense as the postcolonial urban theorist Robinson (2015, p. 193) has asserted that “comparison begins with the ambition to test, and to change, theoretical propositions.”

It seems plausible to take Robinson’s basic (and hardly original—see Siggelkow 2007 and Gerring 2004) insight about the nature of using comparisons to build theories without also accepting her rejection of the traditional comparative methodology of exploring small variations among similar cities, which she refers to as a “quasi-scientific methodological formulation” (Robinson 2015, p. 194). Certainly the very subject of APD, even expanded to include Britain and Canada (Lieberman 2017), is narrow compared to the broader subject range of postcolonial urban studies, and comparative APD studies thus might as well take advantage of their greater potential for “controlling” for that variation, quasi-scientific or not.

Yet, a comparative approach to American urban political development that specifically compares endogenous frictions among multiple orders is more exploratory than the traditional search for variation among similarity and might thus build on the postcolonial promise of “building theory from many different starting points, perhaps resonating with a range of different urban outcomes, but being respectful of the limits of always located insights” (Robinson 2015, p. 194). In my following brief comparison of Philadelphia and Montreal, I attempt to suggest how that exploratory potential might work, specifically by suggesting that comparing endogenous factors allows us a greater flexibility to compare moments across time—something Orren and Skowronek (2004, pp. 8–9) suggest is a signature of APD studies—and trajectories that are moving in different directions across time.

3. Philadelphia and Montreal

I use a highly selective comparative story of political development in Montreal and Philadelphia to identify four types of comparisons: (1) endogenous phenomena in one city compared to exogenous phenomena in another city during the same time period; (2) endogenous phenomena in two cities during different time periods; (3) endogenous phenomena triggered by common exogenous factors during the same time period; and (4) similar developmental trajectories, though moving in different directions, at different time periods. My story starts in 1912 in Philadelphia with the election of Rudolph Blankenburg as mayor, in a city that was famously “corrupt and contented” (as Lincoln Steffens said in 1904) with a fairly typical, if possibly more effective, political machine that ignored civil service laws, appointed electoral candidates by caucus, engaged in flagrant electoral fraud, charged assessments on city employees, padded contracts for kickbacks, and sold off utility franchises. It was a cohesive and essentially integrated part of a larger statewide Republican machine.

3.1. Endogenous Phenomena Compared to Exogenous Phenomena during the Same Time Period

Blankenburg was a successful businessman who had been active in the city’s reform movement since its inception in the 1880s. He was a classic reformer who campaigned on bringing business principles to City Hall, and when he was elected he ended the assessment system on city employees, implemented the civil service system, and awarded contracts to the lowest responsible bidder, though many of his initiatives were blocked by the city council and his legacy, like that of many other reformers, is one of what George Washington Plunkitt famously referred to as “morning” glories—"looked lovely in the mornin’ and withered up in a short time, while the regular machines went on flourishin’ forever, like fine old oaks”. (Riordan 1905, pp. 30–31) Blankenburg was a social liberal, and avid supporter of women’s rights (his wife Lucretia was a somewhat prominent activist in this area) and a cosmopolitan—an immigrant from Germany himself, he actively encouraged immigration and naturalization in his city and made various speeches about the importance of international engagement on the part of the US (Blankenburg 1929).

Two years after Blankenburg was elected mayor of Philadelphia, Mederic Martin was elected mayor of Montreal. Like Blankenburg, the election of Martin was certainly a disruption of business as usual in his city, not least because it ended what was apparently an informal agreement that Montreal mayors would alternate between being French and English. Yet, with the relative uniqueness of their elections the similarity of Martin and Blankenburg ends. Martin was a working-class career politician who was apparently comfortable with the relatively high and visible corruption in city government, and his major campaign message appears to have been a sort of populist appeal to the French working-class combined with an open hostility to the more upper-class English, who were in any case leaving the city in droves (Sancton 1985).

There were similar reform movements in Philadelphia and Montreal that largely ended or at least went dormant at roughly the same times in both cities—in Philadelphia with end of Blankenburg’s mayoralty in 1916, and in Montreal with the election of Martin in 1914. And while in Philadelphia the mayoralty returned to the control of the regular Republican organization, Martin’s election reflected a structural shift in Montreal politics. Martin stayed mayor for ten years, and then after being defeated for one term returned to the mayoralty for a final sixth term, inaugurating a series of long-running mayoralties driven in large part not by party but by political movements that ultimately became the personal vehicles of individual mayors (Sancton 1985, pp. 30–31).

Blankenburg otherwise looks very much like a classic reform mayor, similar in many respects to Seth Low in Brooklyn, Hazen Pingree in Detroit, and Tom Johnson in Cleveland. Except for a few potential differences such as his social liberalism or the fact that he was an immigrant, he could have been a reformer in practically any city. There is thus little “endogenous” to Philadelphia that cannot be explained by the fact that Blankenburg, as a relatively generic reformer, was a function of largely structural forces not necessarily specific to any city. By contrast, Martin appears somewhat unique to Montreal by his appeals to the specifically French working class, and in his inauguration of mayors serving multiple successive terms. The only result of comparing Blankenburg and Martin is really to suggest that there were very different things going on politically in Montreal than in Philadelphia during the 1910s. Like the comparison of Montreal and its typically private gasworks to Philadelphia and its entirely unique city-owned gas works, the Blankenburg-Martin pairing suggests that comparison of an endogenous phenomenon in one city to an exogenous one in another city provides minimal analytical leverage.

3.2. Endogenous Phenomena in Two Cities during Different Time Periods

Possibly a more effective means of gaining analytical leverage is not to search for comparisons across Philadelphia and Montreal at similar points in time, but rather to look for similar endogenous phenomena across time. One of the more obvious similarities is that between Martin in Montreal in the 1910s and 1920s, and the role played by Frank Rizzo, who was Philadelphia’s police commissioner from 1967 to 1971 and mayor from 1972 to 1980. Like Martin, who altered Montreal political traditions by serving four consecutive terms as mayor, Rizzo sought a charter change to the city’s term limits that would have allowed him to run for a third consecutive term. Losing that attempt, he ran for mayor again in 1983, 1987, and 1991, when he died—a desire to serve that has been demonstrated by no other person who has ever held that office.

Both Rizzo and Martin are often identified as “populists” (Sancton 1985, p. 30; Paolantonio 1993, pp. 71, 110, 215; Lombardo 2018, p. 17), meaning that both made successful working-class appeals to get elected. In both instances those appeals were aimed at one racial group (French for Martin and Whites for Rizzo) and against another (English and Blacks, respectively). Both mayors were fairly comfortable with corruption and could thus hardly be called reformers in the traditional sense, though neither could also be said to be the product of a political machine; they rather found the core of their power in their unique abilities to build up a personal following.

Comparing Rizzo and Martin thus begs the question of what was similar between Montreal in the 1910s and 1920s, and Philadelphia in the 1960s and 1970s. At least an initial answer seems straightforward: Martin’s appeals resonated with French resentment against the English presence and control in Montreal, and possibly the increasing presence of new immigrants who were neither English nor French. Rizzo’s appeals resonated with White resentment, both to the increasing Black population in the city and to the White “limousine liberal” elites who were perceived as sacrificing the welfare of the White working class on behalf of an ill-conceived understanding of civil rights (Rieder 1987).

It is worth noting here the senses in which Martin and Rizzo were “endogenous” phenomena, since certainly the appeal of both was a reflection of larger racial, ethnic, and cultural tensions that went well beyond their respective cities’ borders, especially in the case of Rizzo, whose racially-coded message of “law and order” was a reflection of broad national trends, reflected as well in the Republicans’ Southern Strategy and the election in 1968 and 1972 of Richard Nixon as president (with whom Rizzo, nominally a Democrat, was an ally). In some sense, then, Rizzo’s election especially could be seen as an outcome of exogenous forces on the city. Yet, it is also notable that hardly any other major US city elected race-baiting populists as mayors (though certainly there were populist mayoral hopefuls—see (Rieder 1987, p. 32); and Detroit’s Albert Cobo bears at least a vague similarity—see (Sugrue 1996, pp. 83–85)) and none elected a former police commissioner. Thus, Rizzo may have been a product of exogenous forces, but something internal to Philadelphia made that city uniquely susceptible to that exogenous force; the election of Rizzo was, as the sociologist Abbott (2005) might put it, a “hinge” between the linked ecologies of the city’s politics and the broader political system. Martin may well have played a similar hinging role between Montreal and the larger politics of language (that is, French and English), while still also being a uniquely endogenous product of his city.

In Montreal, longstanding tensions between the French and English were possibly only able to express themselves in the form of electoral behavior with at least the partial elimination of property requirements in the latter half of the 19th Century and again in the first half of the 20th Century (Sancton 1985, pp. 24, 32). Working-class groups had long had voting rights in the US but the political machines had always served the role of domesticating the immigrant franchise and attempting to calm tensions between immigrants and existing working-class groups—they were, as Trounstine (2008, p. 97) has put it, “masters of inclusive coalition politics”. Thus, the rise of anti-immigrant nativist movements, reflected for instance in the devastating 1844 riots in Philadelphia, largely predated the rise of the machines. Philadelphia was in any case a relatively low-immigrant city. As Steffens (1904, p. 135) noted, “Philadelphia, with 47 per cent. of its population native-born of native-born parents, is the most American of our greater cities”. The rise of White resentment against new migrants, namely African Americans, thus emerged again after the city’s machine had deteriorated to enough of an extent that it could not control the election of someone like Rizzo as mayor. By contrast, as Taylor (2014) has pointed out, the relative absence of cohesive machines in Canadian cities allowed for the election of a race-based populist such as Martin in the 1910s.

In short, Rizzo and Martin can be lifted out of the noise and muddle of their specific historical contexts to suggest a relatively stable causal relationship between the success of race-based populist appeals in citywide elections and changing racial dynamics among the electorate. Identifying such stable relationships that appear across various places and times, and which exist at the subsystemic level of what Tilly (2001) has called “mechanisms”, also helps to highlight less common relationships that are the products of historical specificity and contingency. Such stable causal relationships are of course an ineliminable aspect of any historical story, which by virtue of having to tell a story has to imply causation, even if only at the level of ostensibly timeless human motivations related to basic needs and wants. Defining timeless variables beyond those basic motivations—the appeal of race-based populism, for instance—moves an analysis increasingly away from a developmental story, and closer to one where time periods are simply warehouses of potential variables that can be removed from their historical context.

3.3. Distinct Endogenous Phenomena Triggered by Common Exogenous Factors during the Same Time Period

The comparison of Montreal and Philadelphia can be made more historical again in two ways. First, repeating a comparison similar to that between Martin and Blankenburg before, we could move the timeline up for Montreal, to see what was happening in that city during approximately the same time as Philadelphia’s Rizzo years. What we see is the marathon mayoralty of Jean Drapeau, which Sancton (1985) has classified as being part of the same era of mayors as Martin, though Drapeau certainly seems a very different character. More specifically, however, a clearer parallel could be drawn between the Montreal police strike of 1969, which was decidedly a victory for the police in terms of not being punished for striking, getting a pay raise, and other benefits (Sancton 1985, p. 110), and Rizzo’s election, which was clearly seen as a victory for Philadelphia police, and which resulted in higher raises and more independence than they would have seen under a different mayor (certainly higher than they had seen under previous mayors).

If the comparison between Rizzo’s election and Montreal’s police strike is similar to that between Blankenburg and Martin, it is also more suggestive in that, contrary to the typical story, it downplays the relative uniqueness of the Rizzo mayoralty, suggesting that, had Rizzo not been elected, possibly the police would have had alternate means to establish their authority in response to demonstrations, riots, and rising crime levels in the late 1960s and early 1970s, as they did in Montreal. Similarly, since the Montreal police were reacting to violence and demonstrations that were somewhat unique to that city, revolving around an increasingly militant French identity, they have also been seen as largely endogenous, or specific to that city. In contrast to the comparison of Blankenburg and Martin, which I argued above is a comparison between an exogenous factor in one city and an endogenous factor in another, the comparison of Rizzo’s election and the Montreal police strike is one between two ostensibly endogenous phenomena that in fact suggest similar responses to a larger, exogenous phenomenon.

3.4. Similar Developmental Trajectories Moving in Different Directions at Different Time Periods

Finally, instead of moving Montreal ahead in time, we might maintain the cross-temporal comparison between the two cities but look in those different periods for developmental trajectories or paths rather than discrete variables. It is, for instance, striking that, while Martin ended the tradition of alternating French and English mayors, it was soon after the end of Rizzo’s mayoralty that Philadelphia’s mayors began alternating between being White and Black—admittedly a rough trend interrupted by two successive Black mayors who collectively served for 16 years. Still, after Rizzo the mayoralty alternated from White (Bill Green), to Black (Wilson Goode), to White (Ed Rendell), to Black (John Street), to Black again (Michael Nutter), to White (Jim Kenney).

Montreal’s alternation between French and English mayors was apparently an explicit if informal agreement, while Philadelphia’s alternation between White and Black mayors is apparently haphazard and possibly simply the outcome of random chance. Yet, the alternation between White and Black mayors also reflects not only the racial profile of Philadelphia (41 percent White and 43 percent Black in 2010), but also the relative political power of both racial groups in the city, just as the alternation between French and English mayors in Montreal reflected not just an agreement but also the relative power of both groups.

The fact that Montreal’s agreement over alternating mayors has long ago ended while Philadelphia’s apparent default alternation began only a few decades ago and appears to be continuing suggests the relatively obvious conclusion that in the Canadian city the English lost political control to the French long ago, while there is at least in some respects a sort of racial pluralism in Philadelphia. Yet, alternating mayors suggests more than just the relative power of racial groups; it suggests as well an institutionalization of expectations about representative figureheads and the symbolic imagery of the city. The pattern in Montreal begs the question of whether alternating Black and White mayors may in fact become an explicit agreement and what might cause that agreement to fall apart.

4. Questioning the Comparability of Cities

To be sure, my brief comparisons of Montreal and Philadelphia are sweeping, oversimplistic, and can be challenged in any number of ways, which is in fact part of the point. The assumption behind the strategy of comparing the differences in two or more roughly similar cities is that, as examples of the same developmental phenomena, the cities are for all intents and purposes indistinguishable from one another, thereby “controlling” for variation. Yet, the differences between even such similar cities as Philadelphia and Montreal are so numerous, and apparent similarities can almost always be challenged, such that any attempt to trace the causes of those differences can easily lead back to the conclusion, reflective more generally of skepticism in the social sciences (Tilly 2001), that the idea of the two cities as being similar in any but the most superficial ways is untenable, and that there is thus no general phenomenon called urban development. There are instead simply two different settlements, in two different locations, serving different functions.

Storper and Scott’s (2016) critique of three influential schools of thought in urban studies—postcolonialism, assemblage, and planetary urbanization—hinges precisely on the claims, at least of the postcolonial and assemblage schools, of incomparability across cities, which Robinson (2015, p. 188) has referred to as “the locatedness of all theoretical endeavor”. Storper and Scott by contrast refer to the claim of incomparability across cities as a “self-defeating preference for intellectual parochialism at the expense of more searching theoretical generalization” (1122). As for planetary urbanization, which largely denies the importance of individual cities and claims that urbanism is an all-encompassing, “planetary” phenomenon, Storper and Scott simply note their bafflement at claims denying the fact that there are visibly distinctly urban places in the world, “characterized by agglomeration involving the gravitational pull of people, economic activities and other relata into interlocking, high-density, nodal blocks of land use” (1116).

Rather than Storper and Scott’s dismissal of claims regarding the incomparability of cities or the ubiquity of urbanism, here I want to extend those claims to their logical conclusion so that they serve as the baseline against which we can evaluate theoretical generalizations. Taken to their logical conclusions, planetary urbanization suggests a radical holism, while the postcolonial and assemblage schools suggest a radical incommensurability and incomparability.

The most extreme versions of holism and incomparability are of course not particularly useful as analytical tools, as the former would fail to make any distinctions between the essential and inessential elements needed for an explanation, while the latter would claim that different elements are so unique that they cannot explain anything besides themselves. Yet, these extreme versions can be relaxed by acknowledging that they are different versions of specifically urban theories that seek to explain distinctly urban phenomena. For instance, the postcolonial and assemblage schools seek to define incomparable phenomena that are specifically urban, which for practical purposes means phenomena associated with, and occurring within, cities. The planetary school denies the significance of specific cities but acknowledges that planetary urbanism resulted from a developmental path in which cities were at least at one point distinctly and uniquely urban.

An analytical task suggested by the incomparability of the postcolonial and assemblage schools is to disaggregate cities into their component elements to the point that those elements are not identifiably urban. At some level of aggregation, social phenomena possess emergent qualities that are uniquely urban, such as a neighborhood where a high number of the residents share a common ethnic identity that they did not share in their homeland, or a police department large enough to have numerous ranks. As those phenomena are disaggregated into their component elements—a single immigrant or police officer, for instance—they become no longer distinguishably urban. It is hardly a novel exercise to try to determine the correct level of generalization, abstraction, or aggregation at which to study social phenomenon, yet it is one which in the study of urban politics tends to simply consist of single paragraphs in studies of specific cities, noting how those cities may or may not be like other cities.

An analytical task suggested by the holism of the planetary school is to reaggregate those discrete elements on the basis of those characteristics most specifically connected to the developmental path that led to the ostensible urbanization of everything. In the process of reaggregation, that social phenomena which becomes clearly distinct to a specific city suggests that level at which claims of both comparability and incomparability can be made.

For instance, a British Quaker may just as easily have lived in Montreal as a French Catholic lived in Philadelphia, and there is nothing particularly “urban” about either identity. At a higher level of aggregation, however, large and dense professional networks of French Catholics are specific to Montreal just as similar networks of Quakers of British origin were unique to Philadelphia. The establishment of dense professional networks is certainly also a step along the path of planetary urbanization, and as those networks expanded and lost much of their religious and cultural characteristics they became general and comparable phenomena. Yet at some earlier point, the distinct natures of those networks in Philadelphia and Montreal also suggests the incomparability of the two cities with one another and thus the extent to which urbanization is in fact not a uniform planetary phenomenon, at least at certain points in time.

A more typical form of dis- and reaggregation is of course to break the world apart into variables—abstractions of some key characteristic that varies among the cases in a sample—and then to analyze the relationships among variables across cases. The key difference here is that I am using the postcolonial, assemblage, and planetary schools to suggest a strategy for defining not variables that exist across cases, but rather elements that could be used to make the argument about the times and spaces in which cases are either comparable or else incommensurably unique.

In the next section of this paper I take my synthesis and analytical operationalization of the postcolonial, assemblage, and planetary approaches and suggest how it might improve upon urban regime analysis. In the section after that I examine recent attempts to combine the study of urban politics with that of APD, and how my aggregation strategy might improve upon, or at least supplement, that as well.

5. Aggregation and Urban Regime Analysis

Since Mossberger and Stoker’s (2001) review and critique of the urban regime literature (at least as it existed up through the 1990s) is still one of the most rigorous, comprehensive, and well-known analyses of what is still a dominant conceptual model in American urbanist political science, and since it focuses to a great extent on comparative urban analysis (at least between US and Western European cities), their article serves as a useful starting point in my attempt to demonstrate the utility of the aggregation strategy that I sketched above.

More specifically, Mossberger and Stoker (2001) did two things that are relevant for my purposes here. First, in recognition of the relative broadness of the notion of an urban regime and thus the risk that it could become a largely meaningless category, they provided four “core properties”, that had to be present in a given city for a regime to exist: (1) partnerships between governmental and nongovernmental “sources”; (2) “the need to bring together fragmented resources for the power to accomplish tasks”; (3) “identifiable policy agendas” related to the interests of the regime’s members; and (4) “a longstanding pattern of cooperation”. Yet, the core properties are themselves riddled with vague and metaphorical terms (for instance, “partners” and “fragmented resources”), which makes it difficult to evaluate the extent to which they might serve to distinguish urban regimes from other social and political configurations.

A focus on the smallest units within cities, and how specifically those units might aggregate to the point of defining a city as either comparable or incomparably unique, is a means of specifying more exactly the core components of a regime, and whether the very concept of a regime is a valid unit of comparison across cities. My proposed aggregation strategy would identify the smaller elements that define such things as partnerships or fragmented resources and ask whether those elements are distinguished by being unique to a specific city. If those elements appear unique to their respective cities prior to the point of being aggregated into groupings that include the core components of a regime, that would serve as the basis for claiming that the cities in question could not be compared on the basis of regimes.

For instance, if one discrete element of partnerships in a city is ex oficio seats for city officials on nongovernmental boards (on chambers of commerce, for instance), and if such ex oficio seats existed in one city but not in another, then those seats could not serve as examples of some general category called a regime that would serve as the basis for comparing the two cities. Instead, the comparison would end at the point of asking why ex oficio seats existed in one city but not in the other—a comparison similar to that between an exogenous phenomenon in one city and an endogenous phenomenon in another.

Second, and compounding the apparently irreducible vagueness of the notion of an urban regime, is a lack of any explanation of the purpose for identifying regimes other than basic categorization. One risk of using a concept for its own sake, especially one that is so ubiquitous within a literature, is that the concept can then easily become reified. For instance, Mossberger and Stoker’s comment that “actual regimes might combine elements of the pure types” (p. 826)—as if regimes are not analytical tools but things that really exist in cities—certainly reflects a larger reifying tendency (reflected as well in other ubiquitous terms as well, such as “elite” or “growth machine”).

By focusing on finer-grained elements within cities, an aggregation strategy defines regimes, and any other proposed category such as a growth machine or an elite, as always contingent and contested categories. Indeed, the entire purpose of the exercise is to question the very notion of categorization in comparative urban analysis.

6. Urban Politics, APD, and Defining Cases in the Absence of Variables

Mossberger and Stoker (2001) also stated it as simply an axiom that “regime research requires case study methodology” (814). With the benefit of two decades of hindsight, this unexplained and obviously incorrect claim seems a remnant of the relatively tiresome divide between “quantitative” and “qualitative” studies that was possibly more salient in 2001 (i.e., Perestroika 2000). Yet, it also reflects a deeper dilemma within both the study of urban politics and APD, which is an attempt to define cases in the absence of variables.

There is of course no sharp divide between case study research and the “large N” research that requires many cases. Study of a single “case” implies other cases to which it can be compared, and that comparison depends on similar phenomena in each case, which when compared across cases become variables. With more cases in an analysis the focus shifts from the cases to the variables, if for no other reason than simply because there is less space to discuss each case in detail.

If there is no clear distinction between variable- or case-based research, to understand Mossberger and Stoker’s claim about the necessary connection between case-based analysis and regime theory, it is instructive to make reference to the planetary, postcolonial, and assemblage schools, which all in one way or another simply deny the existence of both cases and variables. The planetary school focuses on a variable that is so ubiquitous and encompassing that it denies the possibility of being contained within cases. Indeed, the case study framework, which would typically use cities as cases, is derided by the planetary school as “methodological localism” (Brenner 2009) and “methodological cityism” (Angelo and Wachsmuth 2015). Of course a variable that is not contained in cases cannot then vary in value between cases, and it ceases to be a variable. Meanwhile, the postcolonial and assemblage schools deny the existence of variables through the claim that the phenomena that define a case are only understandable by their interactions with one another inside the case, and thus incomparable to seemingly similar phenomena in other cases. This claim of course denies the very possibility of case-based research, since cases imply comparison on the basis of variables.

The problem with Mossberger and Stoker’s claim is a reflection of a larger dilemma of “midrange” theorizing, especially as formulated in urban politics by Stone, and in American political development (APD), in which ostensibly comparable cases are constructed in the absence of variables. Demonstrating a similar issue with midrange theorizing in urban politics and APD is important given Stone’s claim that the study of urban politics might be improved by a greater reliance on APD frameworks, and in particular a conceptual shift from “regimes” to “political orders” (Stone [1993] 2015).

Stone has made three main claims about midrange theorizing. First, city political systems are too complex to be explained by reducing them to their component variables; the analytical focus should instead be “the conjunction of factors rather than attempting to isolate a key variable” (Stone [1993] 2015, p. 109; see also Stone 1989, p. 164). Second, city political systems are distinct from one another but look similar enough that they can be placed into a few categories (Stone [1993] 2015, pp. 104–5). Third, the fact that city political systems do fall into multiple categories suggests the extent to which cities respond in specific ways to larger structural forces, and it is those differences that makes city politics significant (see Stone 1998, pp. 250–51; 1989, p. 227, both quoted in Stone [1993] 2015, pp. 107, 109).

If it is important to place the political systems of different cities in different categories in order to overcome claims that structural forces largely determine city politics, then some criteria need to be established by which to distinguish between different city political systems. Yet, identifying specific criteria risks reducing city political systems to discrete variables and away from case-based research. We are then left with distinct categories for city political systems that cannot be distinguished from one another because the ostensible holism of those systems precludes us from defining their distinguishing characteristics.

While in urban politics midrange theory is used to distinguish across cities, in APD it is meant to distinguish phenomena across time by looking for “friction” among stable “institutional and ideological arrangements that structur[e] political life in a given place at a given time” (Lieberman 2002, p. 702). The broad and vague category of “political order” includes “multiple orderings of authority whose coordination with one another cannot be assumed and whose outward reach and impingements, including to one another, are inherently problematic” (Orren and Skowronek 2004, p. 113). Thus, the task in APD is to uncover disorder in the systems of order so as to explain change. Lieberman (2002) acknowledges that “the challenge of identifying and measuring friction among orders is a serious one…different analysts can find order and disorder in the same material”. His solution is that analysts must engage in a “careful historical reconstruction of the relevant elements of the political setting of the moment under consideration…”

If the APD approach to explaining change over time requires searching for subtle interactions between stability and instability, or order and disorder, all of which are included in the category of “intercurrence”, these interactions can most likely only be uncovered through deep historical research—or, in other words, the same kind of case-based analysis that Mossberger and Stoker claimed was necessary for studying urban regimes. And if anything, the notion of intercurrent political orders is even more inhospitable to identifying discrete variables than urban regime analysis—perhaps a reflection of the fact that APD scholars tend to be more empirically focused and less concerned with at least explicit theorization than urbanists. Combining APD and urban politics would thus seem to compound the dilemma of attempting to define cases in the absence of variables.

One potential path out of midrange theory’s dilemma of case studies without variables has been provided by Lucas (2017a, 2017b) and his data-rich study of changing governance patterns (defined as the extent to which a given policy area is under the authority of a municipal council, provided publicly or privately, by special- or general-purpose government, and locally or regionally) across five service areas (health, police, schools, transit, water) in six Canadian cities (Calgary, Edmonton, Hamilton, Toronto, Vancouver, Victoria) over a century and a half.

In Lucas’s study (2017b, see esp. Figures 1 and 2), model governance patterns vary either across service areas within cities or across cities within service areas. Notably, his focus on discrete service areas is a disaggregation strategy that avoids the necessity of categorizing his six cities by their overall political systems, such as in regime theory. The complexity of the governing patterns he uncovers across the service areas in each city (or perhaps the lack of any patterns at all) suggests that no system of categorization would be particularly helpful.

What Lucas does uncover through an optimal matching technique is that sequences of organizational change are more similar within service areas than across service areas within cities. On that basis he suggests “that those of us who are interested in the long-term dynamics of urban governance must reorient ourselves away from historical studies of single cities and toward analyses of the development of particular policy domains across city and even provincial boundaries” (Lucas 2017a, p. 81).

Lucas’s study is one of the most rigorous attempts yet to apply the admittedly vague theoretical tools of APD to urban political development, and his work provides a direction forward for further studies. Yet, the specific direction is not necessarily that which Lucas suggests. If the primary conceptual vehicle intended to move APD forward is intercurrence, and if intercurrence involves the search for interactions among dissimilar and discordant elements, the subject of study within urban political development should ostensibly be precisely the lack of similar patterns across functional areas within cities.

Similar to the aggregation strategy I have sketched out above, Lucas breaks down urban governance patterns into smaller components and then seeks a method for reaggregating those elements. His method is to search for periods of stability in patterns of governance. Alternatively, he could have looked at dissimilarities as his initial elements—such as, for instance, those moments in time when different service areas within a single city were governed under systems most dissimilar to one another, which by definition would preclude aggregation by service area across cities.

In other words, Lucas’s careful and rigorous analysis that starts from a framework that seeks to theorize disorder brings that framework to the point of logical confusion: should cities be compared on the basis of the extent to which they demonstrate similar endogenous patterns and orders, or on the basis of the extent to which those patterns are in fact dissimilar? Can dissimilarity serve as the basis of comparison across cities, or does it define those cities as unique and thus incomparable? An aggregation strategy such as I have attempted to sketch above could point in either direction, but its benefit is that it would force a choice one way or the other.

7. Conclusions

The comparability of political units and events across time and space, which seems to lie at the crux of any potential connection between urban politics and APD, is confounded by a level of theorizing that attempts to compare cases on the basis of undefined criteria. These criteria are left undefined because of a commitment to case study research that ultimately unravels into a nonreductionism reminiscent of the postcolonial and assemblage schools in urban studies. Making nonreductionism explicit through the aggregation strategy sketched above establishes a baseline for distinguishing between institutional and analytic categories; it allows for more latitude in claiming that ostensibly different types of institutions are comparable, and thus also that different patterns across both time and institutions are comparable. This suggests a different methodological practice, of identifying institutions of maximum dissimilarity, and creating more varied comparisons, so as to generate more ideas. Authors like Lucas and his careful attempts identifying governance patterns across time both within and across cities point the way, sometimes inadvertently, to methodological and conceptual innovation.