The methodological legacies of Theda Skocpol’s State and Social Revolutions: Locating the three pillars of comparative historical analysis

Comparison as exploration

Skocpol recognised that in problem-driven research, questions do rarely pose themselves but instead require active exploration and reveal themselves only slowly by following Moore’s advice to read a lot and look for interesting patterns. She points out that:

the normal science mode that innovation does not matter and that all we should do is test existing theories is the conceit of a very dull, over-developed field. . . . It is hopeless to approach things this way . . . because our questions and angles of vision change with changes in our and society’s normative concerns. I think it is very important to test theories rigorously, but the sense of discovery is equally important. And in human terms, the sense of discovery is more important, otherwise we get bored with our work. (Cited in Munck and Snyder, 2008: 678–679; see also Skocpol, 2023)

Skocpol and Somers recognised that comparing plays a central role for turning these explorations into discoveries. They ask, ‘What motivates the use of comparison as opposed to focusing on a single historical trajectory? What purposes are pursued – and how – through the specific modalities of comparative history?’ (Skocpol and Somers, 1980: 176). They point out that comparison is a plural and not limited to the frequentist logic of cross-sectional variations and using comparisons to statistically control for potential confounders. In employing Mill’s methods in SSR, Skocpol partially embraces this frequentist logic but she also recognises its exploratory limitations. She and Somers draw attention to contextualised comparisons that treat cases holistically and heterogeneously and thus provide a corrective to overgeneralisations. Contextualised comparisons approach cases in a more inductive manner to describe something new or un-describe or re-describe something that has been mis-described before. Contextualised comparisons are interested just as much in differences as they are in similarities because:

the selection of unlikely comparisons . . . highlight the distinct features of a particular system. . . . It brings new insights to scholarship by highlighting unexpected parallels across cases that conventional literatures see as very different, and conversely by underscoring significant differences between cases seen as most similar. (Locke and Thelen, 1995: 338, 360; Kreuzer, 2023: 19–23)

Contextualised comparisons help to identify causal factors that existing theories overlook as well as new cases, aka casing, that cannot readily be placed within populations with known frequency distributions (Soss, 2021). Finally, Skocpol and Somers safeguard against historians’ propensity to undergeneralise by being sceptical of historians’ claim that each case is exceptional and unique. They point out that comparison of contexts is just as likely to reveal similarities that require revising exceptionalist claims, as it does point to differences that serve to identify new cases (Kreuzer, 2003). In short, contextualised comparisons are incredibly versatile.

Varieties of CHA

Skocpol and Somers (1980: 196) wrote that ‘Comparative history is not a single, homogeneous logic of macrosocial inquiry. Rather it encompasses at least three major logics with distinct purposes, characteristics, strengths, and limitations’. They labelled those logics contrast of contexts, parallel demonstration of theory, and macro-causal analysis. Table 1 summarises their key features and juxtaposes them with variance-based modes of analysis that follow a more frequentist logic. The contrast of contexts is the most exploratory and interpretivist strand that employs contextualised comparisons to identify interesting patterns across different contexts or time periods. It is more descriptive and employs concepts to guide the analysis or uses the results of the analysis to refine the concepts themselves. The parallel demonstration of theory uses historical cases to test existing theories. It consequently is very theory-driven and leaves little room for exploring patterns and developing concepts. Macro-causal analysis reflects most closely the logic of SSR and Moore’s Social Origins. It explains cross-sectional static outcomes by developing historically situated, theoretically grounded explanations that carefully elaborate the oftentimes multiple causal chains and intricate generative processes that produce the divergent outcomes (for more on this, see Drochon, 2023; Kalyvas, 2024).

OPEN IN VIEWER

Table 1.

Three strands of CHA in Skocpol and Somers (1980).

Strands of comparative historical analysis			Variance-based analysis
Contrast of contexts	Parallel demonstration of theory	Macro-causal analysis	Multivariate analysis
- Using concepts for meaningful historical interpretations - Highlight differences together with similarities	- Applying general theories to explain historical instances - Using history to test theories	- Analysing causal regularities in history - Using history to generate theories - Focus on explaining cross-sectional variations	- Same as macro-causal but with more cases

Skocpol and Somers recognised that the diverse CHA works could be lumped into three strands that differ in how much emphasis they place on exploration and description relative to confirmation and testing.

Research cycles

Skocpol and Somers (1980: 196–198) conclude their article with a barely developed but subsequently enormously influential afterthought. They suggest that the three strands, rather than competing in a methodological Olympiad for epistemological gold, are complimentary and contribute to different stages of a research cycle. The contrast of context is the most exploratory and thus makes a descriptive contribution of figuring out what is going on and developing concepts to compare cases more readily with each other. The macro-causal analysis builds on these descriptive insights by developing theories that explain patterns more systematically. Finally, the parallel demonstration of theory takes the analysis over the finishing line by testing hypotheses in a more rigorous manner.

The research cycle idea highlights three key – and subsequently enormously influential – qualities of CHA. It underscores CHA’s commitment to placing questions ahead of methods and thus giving exploration just as much attention as the research designs used to test hypotheses. Description, identifying patterns, formulating questions, and forming concepts all are crucial exploratory, pre-testing activities without which it becomes difficult to move on to explaining macro-causal phenomena. Furthermore, it underscores CHA’s disinterest in methodological turf-battles that more often are linked to struggles over resources and prestige rather than to address practical research challenges. Skocpol and Somers’ research cycle already recognised – what the current emphasis on mixing methods is rediscovering – that the world is complex and its proper analysis requires a wide range of tools that the different strands contribute to distinct stages of the research process (Kreuzer, 2023: 67–72, 138–142). They advocate a form of bricolage that is mindful about properly aligning those methodological tools with the ontological properties of the subject matter at hand – an alignment process that is more difficult to accommodate with more formalised, math-based and hence ontologically more restrictive methodologies (Hall, 2003). Finally, the research cycle underscores a Bayesian sensitivity that research is iterative and that confidence in a particular result is highly conditional on validating it against the existing foreknowledge. It vindicates Moore’s advice to read a lot, read closely, and go light on product differentiation.

Renaissance of qualitative methods

Gary King, Robert Keohane, and Sidnev Verba’s (KKV) Designing Social Inquiry’s (1994) clear exposition, synthesis of wide-ranging methodological concerns, and hands-on nature made it an instant classic. Its punch line was to increase the number of empirical observations which had an enormous impact on qualitative scholars who saw as for what it was: a thinly veiled epistemological coup de grace hurled at them. To be told that you were doing multivariate analysis all along and were unwittingly following the same logic of causal inference as frequentist scholars was a methodological wake-up call of the first order. And to be invited into the frequentist family as a lost, albeit slightly backward, cousin was a provocation too incendiary to go unanswered. It was mostly David Brady and David Collier (2004), and an entire generation of Berkeley graduate students that they mentored, who took up the gauntlet that KKV had thrown down. The Collier and Brady response is too complex and still too ongoing to readily summarise. I therefore focus on the three ways in which their rebuttal to KKV moved CHA forward.

First, David Collier drew renewed attention to conceptualisation, first highlighted by Giovanni Sartori (1970), but largely ignored by KKV. Collier pointed out that concepts do not form themselves but are constructed from historical experiences and consequently need to be periodically updated (Collier and Adcock, 1999; Collier and Mahon, 1993). He consequently validated the role eventful analysis plays in conceptualisation. Second, Collier and Brady dug a little deeper than just focusing on the technicalities of research design and showed that those techniques themselves rest on distinct ontological assumptions. KKV’s frequentist logic, for example, assumes that geography does not matter as all locations are presumed to be homogeneous (i.e. unit homogeneity) and history does not matter as all events are independent of each other and thus in effect dateless. (i.e. conditional independence) Or if they mattered, they merely impose boundary conditions on the external validity of findings (Brady and Collier, 2004). Collier and Brady thus made clear that ignoring zip codes and calendric dates backgrounds so many geographic and temporal confounders that causal inferences are more likely to ‘produce results rather than answers’ (this phrase is paraphrased from Aschwande, 2015). They validate the role macro-causal analysis plays in theorising and foregrounding such confounders. Third, Collier and Brady devoted considerable attention to highlighting alternative causal inference strategies to the frequentist orthodoxy proposed by KKV. These efforts opened the door for mainstreaming causal process tracing, Qualitative Comparative Analysis (QcA), Bayesian analysis, and CHA. But just as importantly, they and particularly Peter Hall (2003) underscored the importance of aligning methodologies with the ontologies of the research question at hand. They consequently made it possible to think more systematically about Skocpol’s mantra of placing questions before methods.

Thematic broadening

CHA’s problem-driven nature is closely linked to how its four classic themes – rise of capitalism, democracy, state formation, war/conflict – have continued to reshape modernity since the 19th century. The civil rights struggles, women’s movement, de-colonialisation, and globalisation expanded these themes from the 1960s onwards. This thematic broadening reinforced the methodological heterodoxy that Skocpol first identified. It also reminded scholars that this heterodoxy provides the broad set of analytical tools necessary to properly align methods with the new questions that history continues to raise. Put differently, this temporal broadening shielded CHA from the professional pressures that equate orthodoxy with rigour and consequently place strictures on what question can be asked. Figure 1 retraces the lineage of the four classic themes that connect the 19th-century founders of CHA with their contemporary heirs as well as the themes newly studied since the 1960s. The red dots plot on this broader CHA landscape Skocpol’s works: SSR (1979) Mother and Soldiers (1995), Diminished Democracy (2013) and Tea Party and the Remaking of Republican Conservatism (Williamson et al., 2011). The dotted red arrows identify the additional literatures that Skocpol’s books engage. They give a sense of both the breadth of Skocpol’s original topics and the more modest scope of her recent works.

OPEN IN VIEWER

Figure 1.

CHA genealogy and Skocpol’s contributions.

These themes are too diverse to be readily summarised beyond a basic synopsis. The reader interested in more details can draw on a number of excellent intellectual histories (for more references, see Kreuzer, 2023: 24).

The rise of markets and the resulting commodification of ever newer domains of social life is arguably the most central theme in CHA as it is studied by economic sociologists, economic historians, labour historians, historical institutionalists, and political economists. These scholars explore the origins of capitalism, its variation across countries, or its historical evolution over longer time frames. Next to capitalism, the advent of modern liberal democracies and the modern-day alternatives of fascism, totalitarianism, and types of autocracy are the next most significant CHA topic. The literature on democratisation and regime change explores how modern forms of collective decision-making replaced older, far more personalistic and less rule-based types of governance. They also explain cross-national variations in those regime types and how each of them qualitatively changed over time.

The modern nation state also has long fascinated CHA because its structures are so intimately intertwined with the advent of capitalism and democracy. CHA scholars analyse the modern nation state’s growing bureaucratic capacity, efforts to establish a national identity, interactions with civil society, and, starting in the 20th century, expansions of social programmes. These topics inspired a large literature on state formation, nationalism, political sociology, and the welfare state. War or conflict, or how to prevent them, is the final key CHA theme. Military historians and international relations scholars study the ebbs and flows of war and peace between states while scholars of revolutions, protest, or civil wars analyse domestic conflict. Scholars analyse how technology, military strategy, geopolitics, balance of power, and international norms affect the outbreak or containment of war or produce revolutions and civil unrest.

In the 1980s, a younger generation of scholars added two additional themes: the transformation of racial, gender and sexual identities as well as the analysis of transnational phenomena related to the diffusion of ideas, migration, and proliferation of transnational actors and organisations. These themes first entered the political arena during the 1960s when new cultural movements drew attention to women, LGBTQ, racial, ethnic, and colonial issues. These cultural issues did not fit comfortably in the more material, and structural framework of the first and second generation of scholars. They also were analytically difficult to integrate in the prevailing conceptualisation of modernity as being constituted by a set of coherent categories like class, secularism, rationality, or bureaucratic domination. Such conceptualisations left little room for non-material and more fluid categories like gender, ethnicity, or more informal social practices (Adams et al., 2005: 13). The new literatures on gender and race identities, transnationalism, and post-colonialism thus have begun to fill in more fully the cultural and ideational gaps in the older CHA works. And most likely, the COVID-19 pandemic – a macro-historical event if there ever was one – will introduce epidemiological and public health questions to CHA. Clearly, CHA does not just study historical objects in motion, but is itself a historical discipline in motion.

Temporal awakening

Arguably, the most significant departure from SSR is CHA’s increasing attention to time. This belated attention – even though the adjective ‘historical’ in its label had long alluded to its centrality – reflects the complexity involved in thinking about time in a systematic manner. The historian David Hackett-Fischer (1970: 132) points out ‘a sense of time is not a simple thing. Time is something which people must painfully learn to think clearly about – something, indeed, which they must be taught. And there are many obstacles in the path of understanding it. It is difficult enough in this day and age for a person to imagine that there really was a past and that there will be a future’. His insight has since been corroborated by cognitive psychologists’ discussion about presentism bias, hindsight bias and our propensity to mis-remember things from the past. Time also is incredibly multidimensional involving biological, physical, political, cultural, historical and psychological elements that are difficult to keep analytically apart (Kern, 1983; Levine, 2008).

These obstacles notwithstanding, CHA has paid closer attention to time and developed a clearer temporal vocabulary that differentiates between two analytically distinct but theoretically intertwined notions of time: historical time and physical time (for more citations, see Kreuzer, 2023: 36–63). Let me illustrate this distinction by contrasting the treatments of time in the works of the late 19th-century engineer, and precursor to modern management consultants Winslow Taylor as well as the 20th-century labour historian E.P. Thompson. Comparing their works is illustrative because both study the role time plays in the workplace, but in the process employ historical and physical notions of time.

Taylor has no interest in history and is very much focused on the here and now. He pioneered his famous time motion studies that broke work down into distinct measurable units and discrete sequences. He focuses on the strictly physical dimension of time on the clock to discover – as the title of his famous book indicates – ‘The Principles of Scientific Management’ (1911). He studies how the sequencing of tasks affects the tempo with which they can be accomplished, and whether timing them in the morning or afternoon would increase efficiency. He consequently focused on the context-independent qualities of time (i.e. duration, tempo, timing, and sequencing) that have a clock-like, measurable, and material quality. It is this quality that the attribute ‘physical’ tries to convey (Kreuzer, 2023: 54–66).

E. P. Thompson’s (1967) ‘Time, Work Discipline, and Industrial Capitalism’, by contrast, operates on a longer and hence historical timescale. He is interested in how the growing observance of clock-time in the late 19th century transformed the organisation of the workplace by giving employers a new tool to impose an unprecedented degree of discipline on their employees. Thompson sees physical time not as the context-free, ahistorical foundation for a new scientific theory of work, but as a historically situated innovation that itself helped to transform the power relations between employers and employees at the workplace. In focusing on this transformation, he employs a historical notion of time that sees the past as connected to the present as well as being qualitatively different. Historical time analyzes how the past is not just prior to the present but also qualitatively different from it. It is highly contextualised and focuses on events and their dates. It uses the dates to pay close attention to the chronological order of events and splits and lumps them into periods demarcating time intervals judged to be qualitatively different.

CHA focuses on physical and historical time because they are highly complementary and inescapable elements when studying large-scale, dynamic phenomena that extend over long time periods and involve sociopolitical changes. They are necessary to analyse objects that are in motion through time. The motions of material objects can be analysed just with physical time, but the motions of social objects also require historical time because such objects, unlike the material ones, become transformed as they move through time (for more references, see Kreuzer, 2023: 36–53). Historical time captures the transformative effects of time on such social objects that change through time. Physical time still is useful because it captures the rhythms of those qualitative changes by paying attention to the tempo, duration, sequencing, and timing at which the past unfolds.

The clearer differentiation between these two notions of time marks the biggest departure from Skocpol’s understanding of CHA. She made references to something she calls world historical time (Skocpol, 1979: 39, 173, 288) which refers to the historical boundedness of her findings in SSR. The book also paid attention to elements of physical time. She points out that Russia’s state power consolidated faster than France’s and all three of her cases were late developers and thus faced additional international pressures (Skocpol, 1979; see also Falleti and Mahoney, 2015: 220). Those uses notwithstanding, she did not identify time as a central element of CHA. This led Bill Sewell to characterise her use of time as experimental, an underhanded way of saying that she ignored physical and particularly historical time (Sewell, 1996). A characterisation to which she rightly objected on the ground that Sewell (Skocpol 1994: 330–35) praised her for her use of eventful analysis.