How Relational Methods Matter*
John Sonnett and Ronald Breiger
forthcoming in Culture vol. 18
[newsletter of the American Sociological Association Section on the Sociology of Culture]
Andrew Perrin (2004) argues that we should "privilege standard techniques of quantitative sociology—generally, linear regression and its cousins—in our quest to evaluate and demonstrate culture's empirical role." He claims that the methodological innovations of Martin, Mische, Bearman, Mohr, and Breiger are detrimental in achieving this purpose, because they might "marginalize cultural analysis from the rest of sociology."
Our position is that sociologists should use appropriate methods, whether narrative, essay, cultural-studies, feminist, critical, relational, or those more conventionally discussed under the rubrics of qualitative and/or quantitative, in pursuit of ideas, hypotheses and vision that have as much breadth and excitement as their authors can generate. We are admirers of Andrew Perrin’s research. We would join him in encouraging cultural sociologists to consider using standard quantitative methods in many situations. A view less parochial than Perrin’s, an essay encouraging the use of more relational, less-standard formal methods for cultural analysis, has been argued with such sparkle and verve by John Mohr in a recent issue of this newsletter (Mohr, 2003) that we considered just letting the matter rest. But Perrin raises a number of concerns worthy of further consideration, and we’d like to join in.
It is true that regression methods are institutionalized in quantitative sociology, as Perrin points out. Therefore using these methods may help studies by cultural sociologists get past professional gatekeepers, and perhaps communicate to a wider sociological audience. Also, regression methods are best suited for empirical modeling when theoretical questions take the form of "Does X - - > Y?", as they often do in quantitative sociology. Both of these aspects make regression a useful device for communicating cultural sociology.
The form of questioning required by regression models however is not the only way of looking at how culture matters. To the extent that cultural processes are about meaning, this suggests that discursive and semiotic processes are at work. Ever since Saussure, such processes have been usefully understood on the basis of a wide variety of relational models. We would not want to discourage those sociologists of culture who embrace the relationality on which interpretations build. The use of relational methods does not require researchers to "confuse empirical reality with its schematic representation" (as Perrin charges), but instead to attempt a closer resemblance between empirical models and theoretical understandings. Moreover, if cultural sociology is to have wider influence in the discipline, it must offer something that the discipline does not already have. Demonstrating that cultural variables can affect non-cultural outcomes is one kind of contribution, but it is not the only "value-added" of cultural sociology.
A broader contribution of cultural sociology is to emphasize the interweaving of so-called cultural and non-cultural conditions. Qualitative and ethnographic work excels at this, demonstrating the rich contexts within which cultural action takes place and emphasizing combinations of factors and the embeddedness of cultural action. If causal arguments are made, these often articulate with some form of field theory--a holistic approach to causal interpretation which explicitly rejects the Newtonian logic of regression's X - - > Y. John Levi Martin’s (2003) 50-page inquiry into field theory substantially expands on these points. And the appearance of that essay as the lead article in vol. 109 of AJS suggests that the mainstream is perhaps less conservative than Perrin in its willingness to consider the value of relational approaches to cultural studies. It would be highly ironic indeed if, just when mainstream research in economic sociology (Ruef, 2000) and organizations (Lounsbury and Ventresca, 2002) and, yes, culture (Anheier et al., 1995; Giuffre, 1999) is exhibiting increased interest in relational approaches to culture, sociologists of culture themselves were to erect a principled opposition to relationism—an opposition that we believe is misguided, and based on little more than a fear of not being standard.
Relational methods attempt to capture some of the contextual richness of more qualitative approaches through their configurational and network-oriented frameworks. Martin's (2002) study of constraints on belief, Mische and Pattison’s (2000) study of the joint construction of organizational actors, protest events, and movement projects within a civic arena, and Breiger and Mohr’s (2004) analysis of discourse roles in a post-affirmative action university climate cannot easily be reduced to a Newtonian causal argument. Rather it is the configuration of relations among actors, actions, and ideas changing over time that is emphasized.
Configurational logic is the key to these kinds of approaches, and metric scaling and lattices are only some of the ways this can be approached. Qualitative Comparative Analysis (Ragin, 2000) for example, is a formal and systematic method that incorporates the logic of case-based qualitative work. In this logic, empirical cases are described by the complex configurations of conditions describing them, so that the linear logic of X - - >Y is replaced with the configurational logic of X*Z*Q - - > Y. This approach can help illuminate cultural topics, for example by showing how relations between musical genres and symbolic boundaries do not follow linear patterns (Sonnett 2004).
Our objection to privileging regression methods, therefore, is explicitly not that they are just "shallow numbers" accompanied by an "antiseptic rhetoric," although they are indeed sometimes just that (a point that statistician David Freedman illuminates for sociological methodologists; Freedman, 1991). Rather, the problem with privileging this framework is that it sometimes uses numbers, and theory, in ways that disconnect it from the bulk of "cultural sociology." Assuming the independence of observations, as one must do in regression, explicitly does away with context, reducing it to the assumption of ceteris paribus. Perrin is right that hierarchical linear models can provide important insight into and analysis of cross-level cultural and social processes, and we applaud the use of HLM along with other multivariate techniques where such techniques are appropriate.
Sometimes, nonetheless, it is useful to conceptualize culture as “so entirely situated and context dependent” (as Perrin writes early in his essay) that techniques such as correspondence analysis (think of Bourdieu’s “fields”) become very useful. Moreover, we see the boundaries between “standard” methods and relational or fuzzy—we are tempted to say “funky”—techniques to be very slippery, and worthy of analysis. Who would have thought that the quantitative techniques of field theorist Pierre Bourdieu and those of rational-choice theorist James Coleman are so similar and intimately intertwined? This is precisely a point argued in Breiger (2000), along with an insistence that quantitative approaches themselves need to be understood as cultural manifestations.
In conclusion, it is not our intent to discredit or to throw over regression approaches to cultural sociology, but only to challenge the claim that they should be privileged over alternative approaches. Using regression methods may sometimes help cultural sociologists to provide superb analyses, and perhaps also to publish their work in places where more "non-cultural" sociologists can see it, although we suspect that this latter argument is becoming increasingly less valid. Privileging regression methods would be detrimental to the longer-term development of cultural sociology, because adhering to a hierarchy of privileged methods established a priori requires ignoring some of the more fundamental contributions that methodological frameworks, broadly construed, can make to understanding the complex interweavings of culture and social action.
*Thanks to John Mohr for helpful comments on this essay.
Anheier, Helmut K., Jurgen Gerhards, and Frank P. Romo. 1995. “Forms of Capital and Social Structure in Cultural Fields: Examining Bourdieu’s Social Topography.” American Journal of Sociology 100: 859-903.
Breiger, Ronald L. 2000. “A Tool Kit for Practice Theory.” Poetics 27: 91-115.
Breiger, Ronald L., and John W. Mohr. 2004. “Institutional Logics from the Aggregation of Organizational Networks: Operational Procedures for the Analysis of Counted Data.” Computational & Mathematical Organization Theory 10: 17-43.
Freedman, David. 1991. “Statistical
Models and Shoe Leather.” Sociological Methodology 1991,
ed. P.V. Marsden.
Giuffre, Katherine. 1999. “Sandpiles
Lounsbury, Michael, and Marc J. Ventresca. 2002. “Social Structure and Organizations Revisited.” Research in the Sociology of Organizations 19: 3-36.
Martin, John Levi. 2002. “Power, Authority, and the Constraint of Belief Systems.” American Journal of Sociology 107: 861-904.
Martin, John Levi. 2003. “What Is Field Theory?” American Journal of Sociology 109: 1-49.
Mohr, John. 2003. “The Cultural Turn in American Sociology—A Report from the Field.” Culture 17 (Spring 2003).
Mische, Ann, and Philippa Pattison. 2000. “Composing a Civic Arena: Publics, Projects, and Social Settings.” Poetics 27: 163-194.
Perrin, Andrew J. 2004. “Who’s Afraid of General Linear Regression?” Culture 18 (Spring 2004).
Ragin, Charles C.
2000. Fuzzy-Set Social Science.
Ruef, Martin. 2000. “The Emergence of Organizational Forms: A Community Ecology Approach.” American Journal of Sociology 106: 658-714.
Sonnett, John. 2004. “Musical Boundaries: Intersections of Form and Content.” Poetics 32: 247-264.