PRINCIPLES OF QCA

QCA is a research strategy that aims to combine in-depth knowledge of cases with the goal of generalization (Ragin Reference Ragin1987). The key features of this approach are different from (but not necessarily opposite to) those of statistical analysis. First, QCA conceives cases as holistic entities that cannot be decomposed into single variables/properties. Secondly, QCA envisions causal processes in terms of set relations or relations of implication between phenomena. It starts from maximum complexity of conditions and outcomes and uses Boolean truth tables to identify subsets of conditions that engender particular outcomes. It follows that (a) conditions often display their effect only in combinations with others (conjunctural causation); (b) a given condition may well have different effects depending on the context (contextual effects); (c) alternative sets of conditions may produce the same outcome (equifinality); (d) the conditions for the occurrence and nonoccurrence of an outcome are generally different (asymmetrical causation) (Schneider and Wagemann Reference Schneider and Wagemann2012). This vision of causality well reflects feminist understanding of sociopolitical phenomena as inherently complex, local, and historically contingent (Spierings Reference Spierings2012).

Since QCA was first introduced in the social sciences (Ragin Reference Ragin1987), the initial framework has been extended to include different techniques. While QCA originally operated only on dichotomous sets where cases could either be a member (1) or nonmember (0) (crisp set QCA), recent developments allow for any degree of membership between 0 and 1 (fuzzy set QCA). For instance, a country with a fuzzy score of 0.8 on the set of gender equality is more gender equal than gender unequal, but it still falls short of fully realizing gender equality. Closely related to QCA is the use of fuzzy-set-ideal-type analysis (hereafter FSITA) to develop typologies. This approach is common in comparative welfare state literature where typologies have played an important role in the development of the field.

FUZZY SETS AS A TOOL FOR CONCEPT CLARIFICATION AND TYPOLOGY BUILDING

FSITA takes a deductive approach to typologies. It starts from a concept of theoretical interest, translates it into sets that then combine into a number of configurations (or ideal types), and uses fuzzy set principles to compute memberships in those configurations. This method has been used in cross-national analysis of childcare policies. An and Peng (Reference An and Peng2015) and Szelewa and Polakowski (Reference Szelewa and Polakowski2008) refer to the concept of defamilialization in their typology, while Ciccia and Verloo (Reference Ciccia and Verloo2012) and Ciccia and Bleijenbergh (Reference Ciccia and Bleijenbergh2014) use Fraser's (Reference Fraser1994) models of gender division of labor.

These concepts have long-standing traditions in the gender and citizenship literature, but their success has also generated conceptual confusion. Empirical analyses use different labels to identify similar models leading to poor systematization of existing knowledge, while defamilialization has been reworked by “mainstream” research in such a way as to dilute its gendered meaning. Given FSITA emphasis on theory-driven measurement, these studies have helped to draw out crucial dimensions that more closely reflect feminist debates about gender inequalities in the division of labor. For instance, indices of generosity are commonly used in cross-national analyses of leave policies in spite of the fact that they conflate aspects (time and money) that are known to have very different effects on gender equality—long leaves are detrimental for maternal employment regardless of levels of payment. FSITA can incorporate this criticism because it relies on set intersections: if a country offers low payments for parental leave, this cannot be compensated by offering long durations (Ciccia and Verloo Reference Ciccia and Verloo2012). Further misunderstandings concern the relationship between generosity and gender equality, which are often considered different aspects of leave policies. Ciccia and Verloo (Reference Ciccia and Verloo2012) rely on set intersections to clarify that generosity is a necessary precondition of gender equality since equally few rights for men and women do not promote gender equality. Therefore, FSITA enhances our ability to theorize about the meaning of multiple dimensions—also those that are apparently contradictory.

Secondly, FSITA has been used in feminist welfare state research to deal with hybrid cases showing characteristics of more than one model. Their existence is well recognized in feminist literature. Borchorst and Siim (Reference Borchorst and Siim2008) observe, for instance, that Fraser's models coexist in Scandinavian countries and are the object of contention between various societal and political actors. This insight is not reflected in research practices aiming at reducing cases to a few unambiguous types. Fuzzy scores allow, instead, greater transparency and insight on the coexistence of multiple gender models. Since ideal types are based on analytical distinctions, they need not to be mutually exclusive and may coexist because of the complexity of the social world. This is reflected in the use of partial memberships to allocate cases to configurations.

Finally, FSITA is also used to develop normative typologies. The search for policies to diminish gender inequalities is an underlying motive of feminist scholarship. In empirical analyses, benchmark cases serve to evaluate other countries' performance. For instance, Nordic countries are generally portrayed as the most gender-inclusive model of citizenship. However, these countries also show persisting gender inequalities in many areas (e.g., unpaid work, occupational segregation, glass ceilings, and the incorporation of minority ethnic and migrant women) (Borchorst and Siim Reference Borchorst and Siim2008). This insight is lost with inductive methods (e.g., relative indices, cluster analysis), which define gender equality based on the set of cases included in the analysis. Conversely, FSITA can accommodate more utopian ideas about gender equality since ideal types are theoretical constructs with no empirical validity. Indeed, no country represents Fraser's universal caregiver model (Ciccia and Bleijenbergh Reference Ciccia and Bleijenbergh2014), but this can still be used to measure cross-national differences and identify particular aspects of improvement, as well as to inform policy and normative debates.

QUALITATIVE COMPARATIVE ANALYSIS AS A TOOL FOR CONTEXTUALIZED CAUSAL EXPLANATION

Studies using QCA within an explanatory framework are more easily found in the gender and politics literature. Krook (Reference Krook2010) and Lilifeldt (Reference Lilliefeldt2012) use csQCA to investigate factors explaining differences in levels of descriptive political representation of women. These studies aim to move beyond deterministic explanation and incorporate suggestions from previous research about the influence of specific combinations of factors. Krook aims to assess the influence of combinations of institutional, cultural, and socioeconomic variables, while Lillifeldt seeks to account for the interaction of intra- and extra-party factors. QCA offered some substantial advantages in dealing with these questions. The exploratory nature of Lilifeldt's study required a technique that allowed for openness toward the empirical combinations of conditions leading to high/low female representation. Although she could have used interactions, she was constrained by the low number of cases considered. Moreover, higher order interactions are difficult to interpret in regression analysis, while they are more easily treated with QCA. Therefore, she used conjunctures to highlight causal complexity and identify diverse and context-specific pathways toward similar outcomes.Footnote ¹

Krook (Reference Krook2010) further exploits this feature of QCA to adjudicate between the contrasting findings of large-N statistical analysis and cases studies. Her intuition is that many of these contradictions derive from the diversity of factors at work in different contexts. Therefore, she draws attention to the importance of avoiding one-size-fits-all reasoning. By comparing Western and sub-Saharan countries using context-specific measures of conditions and outcomes, Krook is able to identify different causal mechanisms leading to similar outcomes in the two regions. Her “unorthodox” approach explains patterns in the data that had been previously noticed but not adequately theorized. The goal of her study is to offer a more accurate account of developments within each group of countries. Indeed, QCA produces “modest” generalizations that are valid only for the specific contexts from which the original findings are drawn, or ones that are closely similar. Yet, this feature can be used to extend the geographical focus of comparative analysis by reassessing the validity of theories, concepts, and indicators developed for the Western world to other regions.

Finally, several authors suggest that QCA could advance the empirical study of intersectionality. Hancock (Reference Hancock2013) points to fsQCA as a technique amenable to incorporate both systematic commonalities (categorical intersections) and variation (diversity within) among groups in a way that is sensitive to the historical context and the dynamic interaction within individuals, groups, and institutions. Although McCall (Reference McCall2005) does not refer explicitly to QCA, her intercategorical approach with its focus on multigroup relations and the study of multiple configurations of inequality hints in that direction. In spite of the affinity between QCA and key principles of this theory, there is no study to date applying this method to intersectionality.Footnote ²