For over twenty-five years, Charles C. Ragin has developed Qualitative Comparative Analysis and related set-analytic techniques as a means of bridging qualitative and quantitative methods of research. Now, with Peer C. Fiss, Ragin uses these impressive new tools to unravel the varied conditions affecting life chances. Ragin and Fiss begin by taking up the controversy regarding the relative importance of test scores versus socioeconomic background on life chances, a debate that has raged since the 1994 publication of Richard Herrnstein and Charles Murray’s TheBell Curve. In contrast to prior work, Ragin and Fiss bring an intersectional approach to the evidence, analyzing the different ways that advantages and disadvantages combine in their impact on life chances. Moving beyond controversy and fixed policy positions, the authors propose sophisticated new methods of analysis to underscore the importance of attending to configurations of race, gender, family background, educational achievement, and related conditions when addressing social inequality in America today.
|Publisher:||University of Chicago Press|
|Product dimensions:||6.00(w) x 9.00(h) x 0.70(d)|
About the Author
Charles C. Ragin is Chancellor’s Professor of Sociology at the University of California, Irvine.
Peer C. Fiss is associate professor of management and organization at the Marshall School of Business at the University of Southern California.
Read an Excerpt
Race, Class, Test Scores, and Poverty
By Charles C. Ragin, Peer C. Fiss
The University of Chicago PressCopyright © 2017 The University of Chicago
All rights reserved.
When Inequalities Coincide
The Compound Nature of Social Inequality
Inequality is a key feature of human social organization — some would say it is the key feature. Inequality is praised by some as an engine of progress via the competition it spawns and condemned by others as a scourge, the source of countless social ills. One key aspect of inequality is that it pervades virtually all spheres of social life, from straightforward issues such as the distribution of income (who gets the high-paying jobs?) and education (who gets to stay in school longer and who gets to go to the best schools?) to more mundane things like parenting (who has more time to supervise and support their children and their education?) and focus (who has fewer anxieties about basic survival issues in their lives?).
In almost all known societies, inequalities coincide, compound, and reinforce. It is certainly not accidental that the people with the most income tend to have the best educations, live in the best neighborhoods, send their children to the best schools, take more vacations, have more time to supervise and socialize their children, have fewer demands on their attention, have greater personal security, can pay others to perform time-consuming manual tasks, and so on. These advantages are all connected, and they tend to support and reinforce each other. The most obvious example of a reinforcing link is the one between education and occupation, but there are other connections that are not so obvious, for instance between education and personal safety. Consider, for instance, children living in "rough" neighborhoods who face a daily challenge of navigating dangerous turf on their way to school. It takes a great deal of time, energy, and intellectual effort to accomplish things that would be considered completely routine and taken for granted in "good" neighborhoods. Children in rough neighborhoods would be much better off if they could pour all this "just-staying-safe" energy into their schoolwork. In short, these and other disadvantages often "cascade" (Lin and Harris, 2010: 3) in the sense that they tend to exacerbate other disadvantages.
The argument that advantages and disadvantages are cumulative in nature is, of course, not novel and has frequently been invoked as a key to understanding social inequality (e.g., Merton, 1968; Jencks and Mayer, 1990; Nolan and Whelan, 1999; DiPrete and Eirich, 2006). It is evident that inequalities reinforce each other at both ends of the continuum from poor to rich, resulting in a system of advantages and disadvantages that has profound implications for understanding social inequality (Lin and Harris, 2010). In our society, those in the middle strive to achieve the nexus of reinforcing advantages that comes with being at the top, and they desperately try to avoid the reinforcing disadvantages that push people toward the bottom (Pattillo-McCoy, 1999; Leicht and Fitzgerald, 2006). The bottom is a trap to be avoided at all costs; the top is a promised land commonly known as "having really made it." For many, however, fear of the dreaded bottom and the perilous strata of the middle that border it keeps pushing the definition of "having really made it" ever upward. While definitions shift, the fundamental truth remains that up and down the ladder of human social organization, hierarchies reinforce and reproduce each other. Of course, the fit between hierarchies is rarely perfect, and the strength of both the connections and the reinforcing mechanisms (at both the top and the bottom) varies by time and place. Nevertheless, the connections are durable (Tilly, 1999) and seemingly inevitable, as people everywhere do whatever they can to acquire, maintain, and reproduce advantage for themselves and their children.
The fact that social inequalities coincide and reinforce has direct methodological implications, as variables that characterize the positions of individuals in social hierarchies tend to be correlated, sometimes very strongly. For example, there is a well-known and documented connection between years of education and occupational prestige, which is illustrated in figure 1.1, using data from the General Social Survey. As the figure shows, there is a clear upward trend to the plot — having more education is linked to occupations with higher prestige. Perhaps more striking than the well-documented upward trend, however, is the fact that the upper left triangle — almost half the area of the plot — is virtually devoid of cases. This section of the plot is for lower-education individuals with higher-prestige occupations. As the evidence in figure 1.1 indicates, low-education individuals are virtually barred from high-prestige occupations, suggesting in turn that spending many years in school is a virtual necessary condition for obtaining a high prestige job. Notice also, however, that the lower right triangle of the plot contains a considerable number of cases — indicating that high-education individuals do not always obtain high-prestige jobs. Thus, the triangular shape of the plot indicates that the connection between education and occupational prestige is both correlational (the upward drift) and set-theoretic: people with high-prestige jobs constituting a subset of the people with high levels of education, suggesting that a high level of education is necessary but not sufficient for a high prestige job.
The important point for now is that social inequalities (based upon education, income, occupational prestige, neighborhood, home ownership, property values, school quality, race, ethnicity, and so on) tend to be strongly linked in society and thus tend to be correlated as variables at the individual level. Consequently, when social scientists study connections between different dimensions of social inequality, they are typically confronted with the problem of confounded effects (i.e., multicollinearity). Take, for example, the well-known link between race and poverty — the rate of poverty is considerably higher for blacks than for whites. How much of this difference in rates can be explained by the fact that blacks, on average, have less education than whites? How much of the difference can be explained by the fact that the average parental income (and parental education and parental property ownership and so on) is much higher for whites than blacks? Given that these explanatory factors tend to correlate with each other and also with the outcome, differences in "raw" rates can be misleading.
Accordingly, researchers have spent considerable effort teasing apart the unique or "net" effects of confounded explanatory variables. So far, the key tool for analyzing the effect of variables has been advanced multivariate regression techniques, which are commonly used to provide "corrected" estimates that take confounded inequalities into account by statistically controlling for the effects of correlated variables such as education, parental income, and so on when assessing the connection between variables like race and poverty. State-of-the-art multivariate techniques have been fine-tuned by social scientists and statisticians into powerful tools for the task of producing correct estimates of the effects of correlated inequalities that impact important life outcomes such as poverty.
Yet, given the overlapping nature of social inequalities, the Gordian knot of confounded variables is often difficult to unravel. For instance, consider recent debates on the importance of female-headed families in explaining rising income inequality. A significant body of regression-based studies have aimed to estimate the correct effect size, yet results remain inconclusive and estimates range from 11% to 41% (McLanahan and Percheski, 2008. Similarly, a large literature has examined the size of the welfare effect on family formation, yet the magnitudes of the estimated effects vary widely (e.g., Mofitt, 1998).
Generally, when social scientists seek to estimate the corrected effect of such causal factors as race or education or parental income, they focus on the variation in the outcome (e.g., poverty) that each of the competing variables explains uniquely. If the explanatory variables are strongly correlated with each other (and with the outcome), then each will have a relatively small net effect on the outcome. If, by contrast, they are only weakly correlated with each other (and, again, all are correlated with the outcome), then each will have a relatively larger net effect on the outcome.
Focusing on the unique contribution of each explanatory variable is a more promising approach when predictors are weakly correlated. In contrast, when the predictors are strongly correlated, the variation explained uniquely by each construct of interest becomes minor in comparison to the explained variation that is confounded, suggesting that it may be more promising to focus on the ways in which these explanatory factors overlap and intersect. Furthermore, if these predictors indeed compound and reinforce each other, then we might want to further explore the ways in which the predictors overlap.
Beyond Net Effects: Intersectionality
Estimating the net effects of variables that are correlated with each other is an important task, and, as previously noted, such analyses have the potential to uncover mechanisms that would otherwise be hidden to the unschooled observer. However, the core logic of the procedure — estimating each variable's separate, unique impact on an outcome — runs counter to social inequality's core nature, namely, that its various aspects and expressions are interconnected and reinforcing. In other words, the technology that social scientists have devised to untangle confounded effects (like respondent's education and parental income) are designed to neutralize, not capture, social inequality's essential underlying character. When a researcher estimates the net effect of education or race or some other variable on an important life outcome, the estimate that is derived is in fact purged of the effects of correlated inequalities. Yet an approach where education or any other variable connected to social inequality is assumed to have its own "independent" effect is not well suited to capture the coinciding nature of social inequalities that gives them their resilience and their power.
Consider an alternate approach to the analysis of social inequality, one that draws on the growing literature on "intersectionality." Emerging primarily from critical race studies in the late 1980s, the concept of intersectionality was used by Crenshaw (1989) as an analytical tool to denote "the various ways in which race and gender interact to shape the multiple dimensions of Black women's ... experiences" (Crenshaw, 1991: 1244). Specifically, intersectionality "refers to the interaction between gender, race, and other categories of difference in individual lives, social practices, institutional arrangements, and cultural ideologies and the outcomes of these interactions in terms of power" (Davis, 2008: 68). Its analytical leverage stems from the fact that it shifts the focus from the separate effects of, for example, race and gender towards their combined and synergistic effects, thus emphasizing the multidimensionality of marginalized subjects' lived experience (Nash, 2008). The critical insight intersectionality offers is that characteristics such as race, gender, class, and so on operate as mutually constructed phenomena that shape social inequality (Collins, 2015).
While the original understanding of intersectionality was primarily oriented around notions of identity, the concept has since been broadened, and a number of debates have ensued (see, e.g., Davis, 2008; Walby et al., 2012). For our purposes, we want to retain the focus on intersectionality as an analytical tool for understanding the combinatorial nature of disadvantages while also extending it to apply to advantages. Our approach is most closely aligned with that of Hancock (2007a, 2007b, 2013), who sees intersectionality less as an area of substantive specialization but rather as a normative and empirical research paradigm that enables researchers to conceive of better research designs and data collection efforts based on attention to causal complexity (Hancock, 2007a). More concretely, intersectionality in practice means paying attention to analytic — though not necessarily statistical — interaction: "a transformative interactivity of effects" (Choo and Marx Ferree, 2010: 131).
The intersectionality approach carries several implications for policy analysis. First, its goal is to make the researcher attentive to the various and differing ways inequalities are experienced by social groups (e.g., Hankivsky and Cormier, 2011). Beyond this and perhaps even more importantly, intersectional policy analysis has to potential to improve the diagnosis of social inequality and to promulgate new approaches to alleviate its consequences. In focusing on the intersectional nature of inequality we agree with Hancock (2007b) who argues that "emphasizing the interaction between these factors will illuminate a comprehensive picture, providing the best chance for an effective diagnosis and ultimately an effective prescription."
Intersectionality in Practice
The combinatorial and interactive emphases of the intersectionality perspective carry significant methodological implications. First, the approach is fundamentally comparative and relational (McCall, 2005; Hancock, 2013; Collins, 2015), making it compatible with the comparative set-analytic methods we use here. Second, the approach requires a methodology that is interaction seeking (Choo and Marx Ferree, 2010); that is, it assumes combinations of conditions as the default analytical starting point. As Choo and Marx Ferree note, much of current quantitative methodology is not attentive to such concerns but instead emphasizes parsimonious models that work against "seeing or seeking complexity. Ingrained habits of reductionism, above and beyond any concern with data management issues, often drive conceptual analysis rather than the reverse" (2010: 146–147). In contrast, an approach that places intersectionality at its center has to move away from a focus on the net effects of "independent" variables understood in isolation from each other, and towards a relational approach that focuses on the different ways inequalities intersect.
From a conventional quantitative viewpoint, one could argue that intersecting inequalities are captured in a correlation matrix, which shows the bivariate correlations for all the variables included in a multivariate analysis. However, the issue is not the degree to which one variable is correlated with another across cases, but how multiple aspects of inequality are jointly configured, how they intersect, at the case level. What are the most common configurations? How strongly do different aspects of social inequality clump together at the top? How strongly do they coincide at the bottom? Knowing the different ways that inequality is configured is much more analytically incisive than knowing the degree to which different aspects of inequality correlate, with researchers looking at pairs of attributes, one pair at a time. A correlation matrix has a very limited ability to show, for example, whether or not advantages or disadvantages clump together, and if so, in what specific combinations. Nor does a correlation matrix easily show whether inequality is configured differently for whites and blacks or for males and females. The alternate way of thinking about inequality we suggest here — in terms of combinations of advantages versus disadvantages — shifts the focus from variables to kinds of cases (Ragin and Becker, 1992; Byrne and Ragin, 2009). Here, different configurations of inequality can be seen as constituting different kinds or types of cases. Instead of focusing on the unique effect of just one attribute, we see cases as configurations of intersecting attributes.
As a simple illustration of this idea of examining the different ways inequality is configured, consider the hypothetical data shown in table 1.1. The data illustrate different paths towards avoiding poverty. There are two causal conditions, both dichotomies — respondent has high test scores (yes/no) and respondent has middle-to-high income parents (yes/no). The outcome is also dichotomous — respondent successfully avoids poverty (yes/no). With two dichotomous causal conditions, there are four combinations of values, as shown in the table. In this hypothetical data set, however, there is a further issue — namely, there are no cases that combine the presence of high test scores with low-income parents. Such situations, we would suggest, are common, though certainly not as extreme as presented in table 1.1. In most nonexperimental research, we do not observe all logically possible combinations of conditions, a phenomenon known as limited diversity (Ragin, 1987).
Excerpted from Intersectional Inequality by Charles C. Ragin, Peer C. Fiss. Copyright © 2017 The University of Chicago. Excerpted by permission of The University of Chicago Press.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Table of Contents
1 When Inequalities Coincide 6
2 Policy Context: Test Scores and Life Chances 20
3 Explaining Poverty: The Key Causal Conditions 33
4 From Variables to Fuzzy Sets 61
5 Test Scores, Parental Income, and Poverty 80
6 Coinciding Advantages versus Coinciding Disadvantages 101
7 Intersectional Analysis of Causal Conditions Linked to Avoiding Poverty 121
8 Conclusion: The Black-White Cap and the Path Forward for Policy Research 146