Gender Structures and Women and Politics

A women and politics approach has generally been understood to mean a recovering of women's perspectives and experiences.¹

Gender Structure and Structural Inequalities among Women

Some have argued that a focus on gender unmodified is obsolete in the context of the burgeoning theoretical critiques and empirical research documenting differences among women.³

In reviewing comparative literature on women and the welfare state, Joy Misra and Frances Akins (1998, 260) argue that a focus on gender relations “may privilege the structural constraints facing women and undermine important insights about women's agency.” Although gender as a structure does form an important part of the context for women's action, they argue, researchers ought to focus more on the agency of diverse groups of women in local contexts, and devote less attention to generalizations about the impact of gender structures. So doing, they contend, will show “that structure is a complex phenomenon that has had varying effects on women and their agency based on a variety of statuses, including class and race/ethnicity” (p. 260).

This argument assumes that generalizations about gender structures imply an assumption of a shared experience or impact of particular institutions (say, the state, or marriage) on women. Exploring difference here is equated with exploring agency, while structure is equated with constraint and homogeneity. But arguing that we need to retain gender as a structure suggests nothing about the specific effects of that structure on particular groups of women (Young 2005; see also Young 1994). Indeed, much comparative analysis of gender delineates the ways that gender norms and practices vary across groups of women and men as well as across nations, regions, generations, and cultural groups (Duncan 1995, 1996; Inglehart and Norris 2003; Lewis 1993). Moreover, the critique of gender analysis advanced by many women of color focuses on differences in structural position, not just differences in identity or agency (Collins 1990; Crenshaw 1991; Hurtado 1989).

Analyzing gender relations, or social structures more generally, means focusing on social relations between and among groups of women and groups of men, and on the way the broader social context constrains and enables individual agents (Htun 2005; Hurtado 1989; Young 2005, 2000). Focusing only on mobilization by specific race-class-gender groups of women does not help us understand how these groups of women are positioned vis-à-vis other groups of women and men by social norms, practices, and institutions around which they may or may not be mobilized (Hurtado 1989). The focus on agency does not solve the fundamental conceptual problem of whether women or subgroups of women can be compared across national boundaries, or what we might expect to gain from such a comparison.

Gender as a Category of Analysis: The Analysis of Gender Structures

Some scholars have responded to this problem by suggesting that we think of gender as a category of analysis (Beckwith, 2000; 2005; Hawkesworth, 2005). But the leverage gained from considering gender as a category of analysis stems from the importance of gender in everyday life—the importance of gender structures, symbols, and identities (Hawkesworth 2005). If gender as a social structure has no independent effects, then it is hard to see what justifies an analytic focus on gender as a category, as opposed to, say, gender-race, or gender-race-class. Indeed, this is the force of some current critiques of feminist scholarship that uses gender simpliciter as a category: Some of these scholars argue that gender as an analytic category has no meaning apart from race, class, and other axes of disadvantage (and that these other categories similarly have no autonomous effects) (Brewer 1999; Burnham 2001; Collins 1990; Ferber 1998; Harris 1990).

Iris Marion Young (2005) has argued that we ought to retain the category of gender for political analysis, that focusing on gender as a social structure abstracts from the complex experience of particular individuals and focuses on the macropolitics of social organization. Young (2005, 21) argues that “social groups defined by caste, class, race, age, ethnicity, and, of course gender name subjective identities less than axes of structural inequality. They name structural positions whose occupants are privileged or disadvantaged in relation to one another due to the adherence of actors to institutional rules and norms and the pursuit of their interests and goals within institutions.” Taking this approach has the advantage, Young argues, that “we no longer need to ascribe a single or shared gender identity to men and women” (p. 22). While attributing a shared gender identity to women is problematic, seeing “women” as sharing a structurally defined social position is not: “Thus, membership in the group called ‘women’ is the product of a loose configuration of different structural factors” (p. 21). Following Young, Mala Htun (2005) explicitly distinguishes between structure and identity, arguing that scholars ought to focus on large-scale social structures and processes.

Young (2005) is right to point out that social structures are more than identities, and that gender organizes society systematically to disadvantage women. We need a structural account of politics because we need to be able to criticize social structures, she argues. Moreover, I agree that such macrolevel analysis need not imply shared identities across gender, race, or class groups.

But this observation does not obviate thinking about how to theorize the interaction of different axes of structural inequality (Wright 1997). Indeed, theorists of intersectionality insist that we cannot understand the ways that women are disadvantaged as women nor the ways that people of color are oppressed unless we examine the ways these structures interact. Specifically, they claim that certain aspects of social inequality, certain social problems and injustices, will not be visible as long as we focus on gender, race and class separately (Collins 1990; Crenshaw 1991; Harris 1990; Hurtado 1989; Roth 2004). It is not often recognized that structural analysis is required by the idea of intersectionality: It is the intersection of social structures, not identities, to which the concept refers. We cannot conceptualize “interstices” unless we have a concept of the structures that intersect to create these points of interaction.

Feminist scholars of color have argued convincingly that an account that focuses only on gender will not be able to provide a full reckoning of the ways that women are disadvantaged: In some ways, women of color are disadvantaged as women of color; poor women are disadvantaged as poor women. But these marks of the female condition are nevertheless race or class specific—they are not shared by all women, and may not even be visible unless we focus on specific race-class-gender groups. Moreover, these group-specific experiences reveal aspects of gender structure that are important for understanding the overall social context (Collins 1990, 1998; Crenshaw 1991; Harris 1990; Hooks 1981, 2000; Hurtado 1989). If gender as a principle of social organization cannot be fully understood without an examination of the interaction between social structures, and if women are structurally disadvantaged as women in class- and race-specific ways, then a structural approach to gender analysis requires some account of this structural interaction. The problem remains, then, of how to conceptualize and analyze the interaction between these different structures.

The Structure of Intersectionality: Two Approaches

Scholars of intersectionality point to the limits of “monism” (or a focus on one structure); argue that social structures of race, class, and gender mutually modify one another; and push for scholarship on women “at the interstices” as a way of understanding how these social structures interact (Burnham 2001; Collins 1990, 1998; Crenshaw 1991; Harris 1990; Roth 2004). But a variety of possible relations between axes of domination are consistent with these ideas. In other words, we could theoretically specify intersectional relationships among gender, race, and class structures in a number of different ways.

For example, as noted, one group of scholars seems to understand the idea of intersectionality as implying that systems of gender, race, and class have no autonomous effects (e.g., Brewer 1999; Ferber 1998; Harris 1990). In other words, we really have one social structure called gender-race-class-ability-ethnicity-sexuality, and people occupy one social position as defined by these categories. On this view, it would be nonsensical to suggest that capitalism sometimes reinforces and sometimes undermines gender or race hierarchies (Lipton 1988), that race is a more salient division than class in the United States while the reverse is true in Europe (Wacquant 1995), or that gender is more important than class in explaining some features of women's work (Hartmann 1994; Wright 1997). Making such claims requires the existence of identifiably separate dynamics for each of these axes. Precluding the possibility of autonomous effects assumes that systems of race, class, and gender always work together seamlessly as a single system, and never have any significant independent effects. This idea that all effects of gender-race-class systems are intersectional effects—that there are no autonomous effects of these axes—I will call, for purpose of discussion, the intersectionality-only model of social structural interaction.

There are other ways of thinking about how systems of gender, race, and class interact that are consistent with the core of the concept of intersectionality. For example, we might think of gender, race, and class as having some independent effects and some intersectional effects. Or we might think of gender and race as being mutually reinforcing, while class undermines these systems. Or we might think of all three systems as being mutually reinforcing but analytically separable, and also having some intersectional effects.

Let me try to illustrate by means of an example. One might think of social structures as light shining through multiple layers of colored transparencies onto a patchwork quilt: The color and play of the light shining through depends on the constitution of each layer, but there is no light that shines through just one layer. And the effect of the light will vary, depending on the patch of quilt it hits. So long as the transparencies map perfectly onto one another, a description of the light shining through just one layer of transparency (say, red, green, or purple) would not capture how the light actually falls on any part of the quilt. And the light that shines through will be one color or consistency, although it will fall on different patches differently. The effects are not patches of green beside patches of red beside patches of purple. The effect is just brown shadows: The transparencies combine to fall on every part of the quilt together. Each slide always modifies the effect of the others, and none has an independent effect. Looking at light shining through just the red slide, or just green, or just purple, will not show us how they will combine. Nor does the light from one slide affect some parts of the quilt and not others. The same color of light falls on all patches. This is the intersectionality-only version of how gender-race-and class interact.

Alternatively, the colored slides could be overlapping, but not map perfectly onto each other. This would suggest that some areas would be just green, just red, or just purple, while other areas would be brown (as light filtered through all three slides). In order to capture the play of light over the quilt, we would want to describe the areas of green, red, and purple, as well as the areas of brown. Indeed, it might even help us to notice the green and red and purple areas, even if most of the quilt is covered in brown-colored shadows, because it might help us to understand that the light that falls on the brown areas is filtered through three slides, not one single slide. In other words, each social structure could have both autonomous and combined effects. Finding that some combined effects (areas patterned brown) cannot be described solely by looking at one element of its composition (say, red) does not preclude the possibility that other areas are just red or green or purple. So finding that gender, race, and class sometimes combine to create effects that are unique to specific gender-race-class groups does not mean that every effect of social structures is unique to such groups. Finding intersectionality in some effects does not necessarily imply intersectionality in all effects.

Let me also illustrate this point using a formalization common in quantitative analysis. Sometimes we show additive effects of particular factors using a common formula for regression analysis, where some effect of interest (Y) is produced by a combination of factors (x₁, x₂, x₃…), coefficients that determine the size of the effect of each variable (b₁, b₂, b₃), a constant (c), and some error term (e):

Say that the effect (Y) was the degree of freedom or autonomy granted to citizens in a given society. Here, the term x₁ could represent the effects of gender and x₂ the effects of race. This would be the way to model these effects as being separable from each other and combining in additive ways: gender plus race. We might think of this as a sort of “double burden” or “double jeopardy” conceptualization of the interaction among gender, race, and class: Each dimension of disadvantage creates some distinct advantages and disadvantages that combine by adding onto one another.

Sometimes factors combine in mutually reinforcing ways, so that they magnify each other's effects. This mutually reinforcing relationship is often modeled as a multiplicative one (also called interaction effects) using the formula for interactions between two variables, x₁ and x₂ (say, gender and race):

Here, the mutually reinforcing effect can be captured as a function of the original variables. We might call this the model of gender, race, and class as separable but mutually reinforcing.

Note that modeling multiplicative effects does not rule out additive effects: It is possible for social phenomena to have both sorts of effects (Wright 1997). If there are no additive effects, then b₁x₁ + b₂x₂ will be equal to zero, leaving only the interaction term, constant and error. But we will still be able to derive the combined effects from analyzing the original factors, variables, or structures that combine. So these effects are, in principle, derivable from the independent analysis of each factor or structure (say, class, gender, or race).

Sometimes quantitative researchers seem to assume that intersectional effects are the same as multiplicative effects (the convention of calling such effects interaction effects probably contributes to this confusion). But it is important to note that intersectional effects are by definition effects that cannot be derived as any function of gender, race, and class considered independently (Crenshaw 1991; Harris 1990). No mathematical manipulation of the effects of gender and race will completely capture the way they combine: Intersectional effects are qualitatively different from independent or additive effects. So the idea of intersectionality suggests that there is a third type of effect, one that is not a function of x₁ and x₂, say, x₃.⁵

To return to the earlier example, by way of illustration, if Y represents the degree of autonomy a society grants its citizens, b₁x₁ represents the effect on autonomy stemming from gender inequality, b₂x₂ represents the effect on autonomy stemming from race inequality, b₃x_1*x₂ represents the effect on autonomy stemming from the mutually reinforcing nature of race and gender inequality, and b₄x₃ represents the effect on autonomy stemming from a specific, qualitatively different reduction in autonomy resulting from a particular combination of gender-race, and particular to a particular gender-race group.

Note that here I specify this relationship as one incorporating all three types of effects: additive (b₁x₁ + b₂x₂₎, multiplicative (b₃x_1*x₂) and intersectional (b₄x₃). On the intersectionality-only hypothesis, though, the first parts of the equation (additive and multiplicative effects) would drop out (be equal to zero), leaving only the intersectional effects (x₄): The only effects are intersectional effects.

It is possible, though, that gender, race, and class interact in such a way that there are all three types of effects: additive, multiplicative, and intersectional. It is also possible that the relationship among these different structures varies over space and time. In some times or places, systems of race and class may undermine each other, while in other places they reinforce each other. Insisting that the only version of gender, race, and class is one that sees all effects of social structures as intersectional under all circumstances, it seems to me, wrongly limits the possible configurations of social structures consistent with the observation of some intersectional effects.

I propose allowing the possibility that there are additive and multiplicative as well as intersectional effects of gender, race, and class: Let us allow that the transparencies might not map perfectly onto one another, so that the play of light includes some green, red, and purple patches. Let us call this the intersectionality-plus version of the interaction of these social structures.

The intersectionality-plus account of the interaction of social structures has a major advantage over the intersectionality-only version when it comes to comparative analysis: It admits the possibility that the ways that social structures affect one another vary over space and time. Some axes might be more salient or politicized in some contexts than in others. For example, most of the writing about intersectionality derives from the work of women of color in the United States. Are gender, race, and class similarly entwined in other national contexts? The intersectionality-plus model of social structural interaction is consistent with the idea that different social structures might have different types of effects in different contexts. Observing such variation helps us to identify the distinctive features (and perhaps the causes) of particular national constellations of social structure, perhaps linking such structures to particular historic trajectories. This makes the intersectionality-plus approach particularly useful for comparative political analysis, and it makes comparative political analysis critical for understanding gender (and race, and class) politics.

Some Final Objections

Some will say that this argument misses the point about intersectionality. The idea of intersectionality is not compatible with the existence of autonomous effects of gender or race because we know that every single body is raced and gendered and classed: No one escapes these social structures, and since no bodies are free of these markings, it makes no sense to speak of separate “gender” or “race” effects. But saying that each person's life is marked by gender, race, and class does not imply that each and every condition he and she experiences is equally a product of gender, race, or class. Moreover, even if individuals cannot themselves tease apart the aspects of their experiences that are a product of their gender, race, or class, that are based on reflecting on their own lives, it does not follow that these aspects of social organization cannot be separated by analysis at a macrosocietal level. Indeed, one would not expect every aspect of a single social structure, much less the complex interactions among social structures, to be fully visible in any individual's experience. Are wage gaps more determined by race, gender, or occupation? Are they equally determined by these factors? This is not a question that can be answered by asking people to parse out parts of their personal experiences: It requires macrolevel social research. As Young argues, “the oppression of women and people who transgress heterosexual norms occurs through systemic processes and social structures which need description that uses different concepts from those appropriate for describing subjects and their experience” (2005, 13).

On the other hand, some readers might question whether any scholars actually intend to preclude the possibility of autonomous effects, even the analytic separability of gender as a category. Is the object of critique here a straw woman? I want to emphasize that although some theorists do recommend an intersectionality-only approach, my argument is not that such an approach is required or follows from the concept of intersectionality itself. Rather, I am trying to point out that as it is currently conceptualized, the concept of intersectionality is indeterminate as to the interrelationship of social structures. This indeterminacy points to the need for further theorization and specification of these relationships in reference to particular contexts and questions, not the rejection of work on intersectionality or on the interrelationship of gender, race, and class. It points to the need for further theoretical and empirical investigation of the structure of intersectionality, of the interaction of structures of gender, race, class, and other axes of inequality.

Conclusion: A Comparative Politics of Gender

A women and politics approach sometimes seems as though it merely “adds women” to existing analyses, incorporating sex as a variable in otherwise unreconstructed studies. A politics of gender, in contrast, aims to confront gender “on its own terms,” examining the dynamics of gender politics to see what elements or dimensions of politics, what important questions, may have been obscured by the male, heterosexist bias in the discipline. This has long been the strategy of feminist political scientists, but these days it is sometimes seen as being in opposition to work that attends to differences among women. In this essay, I have argued that far from precluding attention to difference, an intersectional approach requires a structural approach to gender analysis. Moreover, affirming the importance of intersectional dimensions of gender, race, and class need not imply that these axes of inequality have no independent effects. Indeed, the interaction of these different structures likely varies across groups and nations and over time.

The probable variation in the ways that gender, race, and class are defined and the way they interact makes comparative analysis especially valuable for understanding gender politics. Are class and race as closely intertwined in the social-democratic welfare states as they are in the United States? Are gender and class as closely related? Are some social structures more salient or fundamental in some national contexts? Such questions will help us develop theoretical accounts of how gender, race, class, and other axes of inequality interact in different contexts. The answers should help us see new solutions and strategies for resistance to (or transformation of) these structures as well.

A comparative politics of gender, then, investigates the autonomous effects of gender structures, as well as the interaction of gender structures with axes of gender, race, and class. The investigation of the interaction of gender, race, and class could include additive, multiplicative, and intersectional effects. We can best detect intersectional effects by comparing how social structures interact to create particular injustices and problems for particular intersectionally disadvantaged or privileged groups of women, for example, immigrant or refugee women, women working in sweatshops, or female celebrities. We could compare such structures across many countries, or use country-specific studies to illuminate the specific ways that structures interact to shape that particular social position. We can best explore additive effects by examining each axis of inequality (gender, race, class, sexuality, etc.) separately. We can best explore multiplicative effects by examining how the independent effects revealed in an autonomous analysis of each structure combine to mitigate or exacerbate one another's effects. Thus, a combination of analytic strategies will be required in order to paint a full picture of the politics of gender. But comparative analysis is key to illuminating the range of variation in structures of gender, race, class, and other axes of domination, the ways in which these structures interact, and the wide array of strategies for resistance and reform.