Abstract
We test two theories explaining domestic exploitation and violent abuse against women in couples. Exchange theory predicts both to increase when men outpower women; status inconsistency, when women do. As violence and exploitation can affect a couple’s resources, making the model recursive, we focus on native status. Using data from a women’s victimization survey (N = 8,000), we apply biprobit models to compare violent abuse and domestic exploitation in homogeneous and mixed couples in which one is a migrant. The results validate status inconsistency theory: native women with male migrant partners are less exploited but have the highest risks of being abused.
Introduction: Patriarchal Relations in Couples
Gender relations are power relations (Agarwal, 1997; Osmond & Thorne, 1993). They manifest in multiple ambits, also in close relations. Intimate partners in heterosexual relations “do gender” (West & Zimmerman, 1987) and exercise power over each other in everyday life (Komter, 1989) via processes of “contention, bargaining, negotiation and domination” (Tilly & Scott, 1987). These power relations are often asymmetric (Komter, 1989), benefiting men in different domains, such as in the division of housework (Berk, 1985; Delphy, 1984; Ferree, 1990; Kynaston, 1996). They occur with different degrees of intensity, women’s acquiescence, and violence (Dobash & Dobash, 1979; Martin, 1976; Stark & Flitcraft, 1996). For many scholars, these power imbalances conform to and help to self-perpetuate a patriarchal system of power that seeks to exploit women and subject them to men (Bianchi et al., 2000; Coltrane, 2000; Delphy, 1984; Kynaston, 1996).
In this article, we discuss two expressions of men’s power that are often treated separately in the scientific literature—a nonviolent form coalescing into an exploitative division of housework, and a violent form manifesting in violent abuse. Starting with the division of housework, research shows that it has become more egalitarian (Esping-Andersen, 2009). However, the changes have been incomplete (Dribe & Stanfors, 2009) and women still do at least twice as much housework as men (Coltrane, 2000). The advantages that household work asymmetries confer upon men pave the way for women’s exploitation under conditions of oppression and dependence (Brines, 1994; Delphy, 1984). As for violence in intimate relationships (intimate partner violence [IPV])—i.e., “the intentional use of physical force or power, threatened or actual” (Krug et al., 2002) against an intimate partner—research has shown that it is “both a process and an outcome of interactions between the partners” (Cheung & Choi, 2016, p. 16) and that it can take multiple forms (Macmillan & Gartner, 1999). While it can be bidirectional and be affected by a couple’s “violent context/environment” (Pantelides & Manzelli, 2005), it is more common (Heise et al., 2002) and its effects, more harmful (Muehlenhard & Kimes, 1999) and persistent (Sagot, 2005), when it is exercised against women.
In this article, we aim to explore the extent to which these two expressions of power—household exploitation and violence against women—go together, and which distributions of resources within the couple best predict their occurrence. We test two theories to explain women’s exploitation and abuse by their partners, which either emphasize the accumulation and intersectionality of women’s disadvantages, or the challenge posed by women’s empowerment on men’s traditional superiority at home. As argued in the next section, each theory conceptualizes differently the role played by each form of power exertion in facilitating the other. In exchange theory, dependence facilitates the exertion of violence; in status inconsistency theory, violence is an instrument that helps men impose unfair household burdens on their partners. For methodological reasons explained in the following sections, we focus on mixed couples formed by migrants and nationals—on power imbalances based on status differences—to test the two theories. The results of our empirical analyses give more support to status inconsistency than to exchange theory. In the conclusions, we discuss the implications for the development of a theory on the genesis of patriarchy and the role played by violence and exploitation on its reproduction.
Explanations of Patriarchal Relations
There are two main theories for explaining patriarchal relations of power in couples in modern societies. Exchange theory explains these power relations as resulting from an unequal distribution of various types of resources among the intimate partners: education, job opportunities, earning potential, time availability, income, and wealth (see different theoretical models by Agarwal, 1997; Bianchi et al., 2000; Doss, 1996; Dribe & Stanfors, 2011; Haddad et al., 1994; McDonald, 1980; Szinovacz, 1987). Exchange theory considers that men exert power over women “because they can” (Gelles, 1983, p. 157). It expects both violent and nonviolent forms of power exertion by men over their intimate female partners to unfold when women are more vulnerable, as when women’s disadvantages intersect across multiple sectors (e.g., employment and income, race) (Crenshaw, 1990). Conversely, the theory predicts that the more resources women bring to the relationship, the higher their bargaining power will be, and the less likely it would be to observe an unfair housework division and violent abuse (Harway & Hansen, 2004; Kim & Gray, 2008). In this theory, violence is the most severe—even despotic—expression of gender power (Counts et al., 1992; Dobash & Dobash, 1979; Martin, 1976; Pagelow & Pagelow, 1984), one facilitated by woman’s high level of dependence and vulnerability.
In contrast, the status inconsistency perspective—also know in the literature as status incompatibility theory (Kaukinen, 2004) or, simply, the relative resource theory (Atkinson et al., 2005)—highlights the importance of violence as the source of patriarchy and as an instrument for its restoration when its operation is threatened. It can be traced back to O’Brien (1971), Rodman (1972), and Gelles’s (1974) work. It focuses on how differences in income, prestige, and occupational and educational attainment favoring women in intimate relations disrupt traditional patriarchal roles within the household, leading to more violent expressions of power aimed at restoring men’s power over women (Atkinson et al., 2005; Gelles, 1974; Kaukinen, 2004; Rodríguez-Menés & Safranoff, 2012; Yllö & Bograd, 1988). In this theory, violence against women is a tool that men use when the system of traditional domination and subordination fails to operate as a consequence of women gaining more resources and being in a position of superiority compared to their partners. Status inconsistency theory illustrates perfectly Weber’s (1978) distinction between nude power (violent subjugation) and authority (willing obedience, or domination). It predicts that the former will ensue when the latter, from which it originates, breaks down.
Objectives and Analytical Strategy
The purpose of this research is to test the validity of the two explanations of patriarchy presented above. It contrasts two sets of predictions derived from each alternative theory regarding the use by men of nonviolent and violent forms of women’s subjugation under different distributions of resources within the couple.
If, like exchange theory argues, men subject women because they can, they should do it in proportion to their degree of power over them. Thus, one should expect women’s subjugation to be more frequent and intense when women have fewer personal resources and are more dependent on men, and to manifest both in nonviolent and violent forms of patriarchy—both in more exploitative distributions of housework and in stronger abuse of the female partner.
If, on the contrary, the theory of status inconsistency is correct, one should expect men to change the means of subjugating women depending on the distribution of resources within the couple. They would rely more often on nonviolent forms of domination and on exploitative allocations of housework when women have lower resources and are more dependent on men, and more often on more violent abuse when women have higher resources and challenge the traditional patriarchal order.
The top diagram in Figure 1 summarizes the two models being contrasted. The predictions derived from exchange theory about the effects of an unequal distribution of resources favoring men on the domestic distribution of work and on men’s violent behavior against their partners are displayed with dashed lines. Men’s higher relative resources are expected to lead to a more exploitative distribution of domestic work, which in turn will increase the probability of violence. It is possible that violence may ensue directly from the unequal distribution of resources favoring men but also to expect the unfair distribution of housework to play an intermediate position in the cumulative path to violence; we only omit this direct path for simplicity in the explanation.
Theoretical Models of Patriarchy.
The predictions derived from status inconsistency theory are displayed with solid lines in Figure 1. A distribution of resources favoring women is expected to lead to more violence against them, which in turn would lead to a more unfair allocation of domestic work within the couple, thus restoring the traditionally patriarchal domination of men over women. It is possible that situations of status inconsistency may directly result in more unfair burdens of household chores for women, although one should expect violence to play an instrumental role in imposing it in the absence of other resources; we only omit this direct path for simplicity in the explanation.
The top diagram in Figure 1 also shows the problems in estimating the two models, for they both might be affected by the presence of recursive effects. Exchange theory predicts that violence against women (and an exploitative division of labor, which we have omitted for simplicity) would result in women’s more subordinate position in other aspects of life, and hence, also in lower relative resources. Status inconsistency predicts that less violence (and more fair divisions of labor) would result also in fairer distributions of resources in other ambits of life.
The models cannot be estimated unless the effects of the domestic division of labor and violence on the distribution of other resources are blocked. This is difficult to do with resources that can be easily attained (like education or income), but less so when considering status resources that are ascribed to the person, like immigrant/ethnic status or, less clearly—as it will be argued below—age. The bottom diagram in Figure 1 illustrates how considering relative ethnic/national resources as an exogenous variable facilitates testing the validity of the two theoretical models. Note that while the paths leading from the domestic division of labor and from violence to relative ethnic/national resources have disappeared, the double arrows connecting the division of domestic labor and violence to each other are still displayed, indicating that it may only be possible to estimate the correlation between them. This limitation will not affect the testing of the two theoretical models provided that the effects of relative ethnic/nationality resources on violence and on the division of domestic labor—which each theory predicts to follow alternative directions—are estimated net of the correlation between the two manifestations of male power. Before presenting the data and the methods for testing the model, and its results, we next briefly review the latest evidence on the differences across ethnic and national groups on the role of women in the domestic division of labor and their probability of being victimized by their partners.
Relations in Nationally or Ethnically Homogamous and Heterogamous Couples
There are relatively few studies that have analyzed the impact on gender relations of power of differences in couples’ ethnic or immigrant status (Carbone-Lopez, 2013; Frias & Angel, 2013; Hattery, 2009). Most of them have compared groups of ethnically or nationally homogeneous couples to each other so as to evaluate the level of women’s domestic exploitation and victimization, while only a few have studied power relations in mixed couples with partners from different national or ethnic groups.
Starting with studies on cross-ethnic differences between homogeneous couples, and regarding women’s roles in the domestic division of labor, research has consistently documented the presence of significant differences across ethnic groups (Bolzendahl & Gubernskaya, 2016; Orbuch & Eyster, 1997; Parrott, 2014; Shelton & John, 1993; Wight et al., 2013), with Hispanic and Asian women taking traditionally more conservative roles within couples than whites and blacks, at least in the United States. Some argue that these patterns are affected by the immigration status of the couples. Thus, supporters of family investment theory argue that immigrant women’s labor-market participation (and consequently, household dedication) are higher (lower) than native women’s upon arrival, so as to finance migrant husbands’ acquisition of the human capital they need in the host country, with the differences declining with years since migration, signaling assimilation (Baker & Benjamin, 1997). However, the evidence is inconclusive with variations among immigrant groups based on the strength of traditional family arrangements in their countries of origin (Antecol et al., 2003; Blau et al., 2003; Fleischmann & Höhne, 2013; Yick, 2001).
Research has similarly shown important variations between ethnic groups and national and immigrant groups in women’s probability of being victimized by their male partners, with minority mono-racial couples having in general higher risks of IPV than homogamous majority couples (Caetano et al., 2000; Carbone-Lopez, 2013; Kantor et al., 1994; Straus et al., 1980; but see Frias & Angels, 2013 for alternative findings). Most differences disappear once variations in socio-economic conditions and alcohol use are taken into account, prompting some researchers to argue, along the lines of family stress theory, that stressful conditions are more important than patriarchal ideologies in explaining why minority groups—the most disadvantaged and subject to stressful conditions—are at higher risk of IPV (Benson et al., 2003; Gelles & Straus, 1988; Straus et al., 1980). And yet, other studies have shown that these group differences persist even after controlling for socio-economic conditions, with minority and migrant groups having higher odds of experiencing IPV, and these being associated with patriarchal ideologies (Carbone-Lopez, 2013; Rodríguez-Menés & Safranoff, 2012; Tastsoglou & Miedema, 2005).
More important for this study are the findings about the impact of ethnic and national heterogamy on women’s probabilities of experiencing an unfair division of labor and of experiencing violent abuse from their partners. Regarding the domestic division of labor, research has found that interracial partnerships are more gender equitable than homogamous ones (Bolzendahl & Gubernskaya, 2016). However, the amount of time devoted by women to domestic chores in interracial couples depends on their minority/migrant status; they appear to do more when they belong to these groups, and less when they are part of the majority or hosting groups and women are young (Baker & Benjamin, 1997; Basu, 2017; Grossbard et al., 2014).
As for women’s odds of being victimized by their male partners, research has shown that they are higher in mixed than in homogamous couples (Brownridge, 2016; Chartier & Caetano, 2012; Fusco, 2010; Martin et al., 2013), but only, or especially, in couples where the female partner belongs to the majority group, and the male partner, to the minority (Carbone-Lopez, 2013; Frias & Angel, 2013; Hattery, 2009). A common explanation is that minority men need to reassert their masculine superiority when, as in mixed couples, they have an inferior status, in line with the status inconsistency argument exposed in the previous section.
In sum, most studies analyzing differences across minority and majority groups in the exertion of power by men over women within couples—whether minority groups are defined in terms of ethnicity or of immigrant status—indicate that minority men are more likely than their non-minority counterparts to impose heavier household burdens on their partners, especially when they have patriarchal ideologies. In contrast, a minority man’s odds of violently abusing his female partner appears to depend on her minority status (higher when women are native). Thus, previous research seems to provide preliminary support for both the exchange and the status inconsistency theories; women with fewer status resources are more likely to experience an unfair household burden, but are also less likely to be victimized. However, there are some methodological problems with the studies discussed in this section that make their conclusions problematic.
Selection Effects
One problem with these studies is that they did not analyze the two forms of men’s power simultaneously and, thus, did not control for their potential reciprocal effects. To properly test the two theories proposed above for explaining women’s domestic exploitation and violent abuse we will control for the association between both forms of males’ power exertion over their intimate partners.
In addition to this endogeneity problem, many studies do not consider possible selection effects in couple formation, that is, the fact that men and women with special characteristics that make them more prone to exert or suffer these different forms of abuse, may also be more likely to form couples with out-group members. Effectively controlling for selection effects is necessary, not just to control for individuals’ personal reasons for marrying or cohabiting outwards, but also, or especially, for the reasons that partners in the out-group find sufficiently appealing to accept the partnership. Thus, the most effective controls for selection effects are members’ characteristics when assessed in relation to their partners.
Researchers studying the association between individuals’ relative resources and the probability of ethnic intermarriage have typically focused on educational differences between partners (Basu, 2017; Chiswick & Houseworth, 2008; Furtado, 2012; Jasso et al., 2000), although only a few have explored their repercussions on partners’ labor-market participation or conflicts (Baker & Benjamin, 1997; Cooper, 2015; Grossbard et al., 2014). Most have found a positive association between assortative mating and intermarriage, with partners in mixed couples having more similarly high levels of education than mono-racial couples. This has been taken to indicate that more highly educated individuals may be willing to trade their ethnicity for the benefit of finding mates with similarly high education, due to the lower size of the marriage market for educated people (Basu, 2017; Jasso et al., 2000). There are also indications that this positive association is stronger among female than among male minorities who intermarry (Chiswick & Houseworth, 2008; K. H. Choi & Tienda, 2017), and that this is associated with lower household specialization (Basu, 2017; Grossbard et al., 2014).
A problem with focusing on partners’ educational asymmetries with cross-sectional data is that what appears to reflect an assortative process of mate selection might be partly the outcome of power relations within the couple developing after it is formed. A variable that cannot be affected by such recursive forces is age. Research has shown that age often acts as a proxy for valuable resources that individuals trade with prospective partners in exchange for other valuable resources in mating markets and that these resources vary by gender (Oppenheimer, 1988). In societies with differentiated gender roles, women often trade their youth, which is attractive to men, in exchange for older men’s social status (England & McClintock, 2009). Older women conversely adjust to their deteriorating prospects in mating markets by giving up other valuable resources or accepting less-favorable partnerships (Skopek et al., 2011). The larger age difference in ethnically mixed couples observed in some studies (Basu, 2017) might be explained by such selection effects. If in these couples age asymmetries between husband and wife were positively associated with higher household specialization and violent abuse, as some studies seem to suggest for the overall population (Coker et al., 2002; White et al., 2012), the estimates on the impact of intermarriage on these outcomes would be biased.
In this research, we control for educational, employment, and age differences between partners, and explore the interaction effects between age, type of couple (homogamous vs. hetetrogamous), and migratory status, to minimize any selection effects on our estimates of the impact of women’s relative resources on their odds of experiencing an exploitative distribution of household chores and violent abuse.
Data, Variables, and Techniques of Analysis
We use data from the IV Macro-Survey of Violence on Women conducted by the Spanish Centro de Investigaciones Sociológicas in 2011. The sample is made up by 7,898 18+-year-old women residing in Spain in 2011. Only data from women in an intimate relationship at the time of the interview were analyzed so as to unmistakably link partners’ and respondents’ employment, educational, or occupational information to women’s experiences of exploitation and violent abuse at the time of the interview. This and other constraints (see details below) reduced the sample to 5,111 women.
There are two dependent variables based on women’s subjective appraisals: an indicator of violent abuse against women in intimate relationships and an indicator of an exploitative work burden in the household.
The first dependent variable is a binary indicator of IPV that excludes diminishing and derogatory attitudes and behaviors that do not threaten to or inflict physical damage. However, we include nonphysical forms of violence that are severe enough to capture the possible underreporting of physical violence (their correlations with physical violence are very high). We build the indicator of violence from responses to three low-level items in a questionnaire where women reported the frequency (never, rarely, sometimes, or often) with which they were experiencing denigrating and abusive attitudes and behaviors from their partners using a simplified version of the revised Conflict Tactics Scales (Straus et al., 1996). The three items asked how often women were pushed or beaten by their partners when they were angry; felt afraid of them; and were threatened or insulted by them. We performed a multiple correspondence analysis (MCA) on these items to identify an underlying scalar measure of violence along which women could be placed. MCA decomposes the variance of respondents’ responses to several questions into orthogonal scalar components along which respondents are distributed following normalization constraints. One component is often enough to capture much of the total variance. Before proceeding to extract the component(s), MCA transforms the original categorical variables into scales, using an optimizing algorithm that assigns scores to categories in ways that maximize the correlations—loadings—between the transformed variables and the extracted principal component(s). The loadings help interpret the “meaning” of the components. We only extracted one component or underlying dimension. This dimension explains 74% of the variance of the responses to the three transformed variables. The loadings are high (ranging from .85 to .86), indicating that the three questions contribute highly to explaining the uncovered dimensions, because they are highly intercorrelated (Cronbach’s Alpha was also high: 0.8). We interpret this dimension, violence, as capturing physically threatening abuse against women in intimate relationships. The dimension is subject to normalization constraints by MCA, and thus has a mean of zero and a standard deviation of 1. However, the distribution is positively skewed because most men do not abuse their partners. To isolate more clearly the cases of asymmetric power relations, we dichotomize the scale uncovered by MCA, grouping all cases on the right-hand tail of the distribution in one group, and categorizing all other women in the other. Women who take the value 1 in this binary variable have scores equal to or higher than one-half a unit of standard deviation above the mean of the scalar variable uncovered by MCA and are considered to have experienced violence in any of the three forms. The remaining women take the value of 0 and are seen as not having experienced violence in any form.
The second dependent variable is a binary indicator of exploitative household burdens. We also build this indicator from three low-level survey questions asking each woman how often their male partner (a) gets angry if the chores that affect him are not done; (b) makes her feel guilty because she does not take care of him as he expects; and (c) makes her responsible for all housework. Note that the three questions capture not just an uneven distribution of housework in the couple, but also one with exploitative overtones. We performed an MCA on these items to identify an underlying scalar measure of such exploitative housework burden along which the female respondents could be placed. This dimension explains 60% of the variance of the responses to the three transformed variables. The loadings are high (from .73 to .80), indicating that the three questions strongly contribute to explaining the uncovered dimension because they are highly intercorrelated (Cronbach’s Alpha is 0.7). As with the other dependent variable, we also dichotomized the scale uncovered by MCA; women who take the value 1 in this variable have scores equal to or higher than one-half a unit of standard deviation above the mean and are considered to have experienced an exploitative housework burden in any of the three forms; all others take the value of 0.
The analyses were run with bivariate probit regression, which estimates two simultaneous equations to predict two male behaviors—violent abuse and household exploitation—that may be interdependent. Independent variables’ effects were estimated in these equations net of the correlation between the two dependent variables, which was also estimated in the analysis. We also ran sensitivity analyses with simultaneous equations regression using the undichotomized scalar variables directly generated by MCA. The results, available upon request, were very similar to those obtained with the biprobit analyses.
The two equations include various kinds of independent variables. The main ones aim to capture, with the help of two variables and their interaction, women’s status relative to their partners with respect to their nationality. The two variables are whether or not the couple is mixed and composed of one Spaniard and one foreign immigrant, and whether or not the female partner is an immigrant (regardless of whether the couple is mixed). Unfortunately, we only know the region of birth of nonnationals, and hence any nationalized or second-generation migrants are undistinguishable from other Spaniards. Also, note that we code non-Spanish European migrants as Spaniards. Both issues may downplay the differences between homogeneous national couples and the other couples. We accept this potential source of bias so that any differences between mixed and homogeneous couples may be interpreted unambiguously.
The second group of independent variables seeks to capture other resource unbalances between couples in employment, education, and age. Employment asymmetries are captured with an interaction effect between two variables recording respondent’s and her husband’s employment status. Educational asymmetries are captured with a simple variable reflecting if the respondent has the same educational level as, or lower or higher than, her partner. Age differences between partners are expressed in a variable with three categories expressing if the respondent is in the same 10-year age interval as her husband or an interval below or above his.
The final group of independent variables consists of controls expressing the respondent’s socio-demographic characteristics [education, employment, age (recoded into mid-class interval values), marital status, and religious affiliation] and family conditions (head of family’s occupation, cohabitation of other young and adult individuals with the partners, and rural or urban characteristic of household’s habitat).
Results
We start this section by describing the four different types of couples whose power relations we are interested in exploring. Table 1 presents descriptive statistics for all variables used in the analyses, distinguishing them by the four types of couples.
Descriptive Statistics.
Over 90% of female respondents are in couples formed by two Spaniards. Only 4.5% are in homogamous immigrant couples. The rest are mixed couples, with almost twice as many couples in which she is an immigrant as mixed couples in which she is a Spaniard. In line with the previous research mentioned above, the data show clear processes of positive assortative mating for more educated migrants in both types of mixed couples. However, the data suggest that what drives the over-representation of higher-educated migrants in mixed couples is not that they trade their ethnicity so as to find more educated mates, for in mixed couples migrants have higher probabilities of being more educated than their partners. Instead, it appears that higher education is an instrument for migrants to reach the national mate market. Females use this instrument more often than males (they tend to be more educated than their national mates in larger proportions). They also use their youth to attract older nationals in much larger proportions than migrant men attract older women. Both aspects may explain the larger numbers of mixed couples formed by female migrants.
There are other important differences between the two types of mixed couples. The families formed by migrant females and native males are more traditional than the more unconventional mixed families in which he is the migrant. The former marry in larger proportions, have more children, and live in lower proportions in urban dwellings than the latter. Furthermore, the female migrant works in lower occupations while the husband is more likely to work, and she has less liberal and more religious values than in families formed by migrant men and native women.
These differences among mixed couples could explain why they also differ in the two dependent variables. The percentage of respondents reporting household exploitation (as defined here) is about twice as large in mixed couples with a female migrant as in mixed couples with a female national. In contrast, the percentage reporting violent abuse is more than four times larger in the latter than in the former. This is a first indication that, as predicted by status inconsistency theory, imbalances in national status favoring women could trigger violent responses from the male partner to restore his patriarchal power. This interpretation is reinforced by the fact that the proportions of abused women are 20% higher, and those of exploited women, three times lower, than the corresponding proportions in homogamous migrant couples. An interpretation that the intersectionality of migrant women’s disadvantages as migrants and as women may promote exploitation at home is undermined by the lower prevalence of exploitation and violent abuse in mixed couples with female migrants compared with homogamous migrant couples. Rather, it points to the role played by male migrants’ patriarchal ideologies in accounting for women’s household exploitation and violent abuse, although power imbalances favoring women also play a role, especially regarding abuse.
As we saw above, mixed couples in which females are native are also more likely to have other power imbalances in employment and age. In these couples, situations in which she works and he does not and in which she is older than her partner can be more easily observed. The employment imbalance could be either a consequence or a cause of violent abuse; the age difference may point to the selection of less attractive women into more problematic marriages. We next report the results of more refined analyses aimed at controlling for these endogenous and selection effects.
Table 2 shows biprobit estimates of women’s likelihood of suffering IPV and nonviolent forms of household exploitation. The rho at the bottom of the table shows that, as expected, the two dependent variables are highly, positively, and significantly correlated—the more exploited a woman is in the household, the higher her odds of being violently victimized.
Predictors of Violent Abuse and Domestic Exploitation. Biprobit Estimates.
Note. N = 5,111.
significant at the 0.05 level; **significant at the 0.01 level; ***significant at the 0.001 level.
Very few variables, apart from those measuring respondents’ immigrant and homogamy status, have a significant impact on these two dependent variables. The log-odds of being violently abused and exploited in the household increase as the respondent is less educated, almost linearly. They decrease when she has paid work and increase when minors live in the household (significantly only when predicting an exploitative household workload). This appears to give support to exchange theory that makes both exploitation and abuse a function of men’s and women’s resources, increasing them when women’s resources are low. However, other interpretations are possible, along the lines of stress theory (Benson et al., 2003; Gelles & Straus, 1988; Straus et al., 1980), which predicts higher violence in stressful conditions, or along the lines of cultural theories that make women’s victimization and exploitation a function of cultures of violence and traditional ideologies that are more prevalent among the least educated (Hakim, 2002; Kurz, 1989). Another problem with exchange theory is that it predicts higher abuse and exploitation when women’s resources are lower than men’s. The results in Table 2 suggest that it is the other way around, giving support to status inconsistency theory’s claim that abuse and exploitation increase when women’s resources exceed those of men, as when women are more educated or when they work and their partners do not (see the positive coefficients for these variables in Table 2). And yet, however unlikely, it is still possible that women’s work and education may themselves be a function of abuse—as when women seek an exit to their victimization—or of exploitation—as when women multiply gender displays at home so as to gain access to external resources (Monna & Gauthier, 2008). These recursive effects are even less likely to occur when dealing with resources that, like ethnic or native status, are ascribed rather than attained, especially when the victim or the exploited is the advantaged member in the couple.
The results in Table 2 show that women in heterogamous couples formed by migrants and natives are at significantly higher relative risks of being victimized, and at (nonsignificantly) lower relative risks of being exploited at home. 1 However, the interaction effect shows that the higher odds of victimization apply only to native women in mixed couples, while migrant women in mixed couples have lower odds of being victimized. Similarly, the results show that immigrant women have higher risks of experiencing both forms of abuse relative to native women, but that this applies only to an abusive household burden, for immigrant women actually have lower risks of being victimized if they are in a relationship with a native. Figure 2 shows these results graphically and expressed in terms of probabilities.
Predicted Marginal Probabilities of Being Violently and Nonviolently Abused, by Type of Couple and Female Respondent’s Nationality Status.
In Figure 2, the marginal probabilities of (a) being victimized and (b) being exploited are shown for native and immigrant women in homogamous and heterogamous couples. Migrant women have higher probabilities of experiencing an exploitative household burden compared to native women, regardless of whether they are in ethnically homogamous or heterogamous couple, providing little support both to family investment theory (Baker & Benjamin, 1997), which predicts lower exploitation of migrant women in homoganous couples due to credit constraints pushing women out of the household, and to exchange theory, which predicts higher exploitation in heterogamous couples due to migrant women’s double disadvantage in relation to their partners. The left side of Figure 2 shows that migrant women also have higher probabilities of being victimized, but only in homogamous couples. Finally, native women in intimate relationships with male migrants have the highest risks of being violently abused, giving support to status inconsistency theory, which sees violence as an instrument to restore men’s power when it is challenged.
There is still a possibility that native women’s higher probability of victimization in heterogamous couples may be a function of selection into specific types of couples of women with characteristics that make them more prone to be abused. Figure 3 distinguishes a native and a migrant woman’s predicted probabilities of being violently abused or exploited at home in homogamous and heterogamous couples by whether or not she is older than her partner. These probabilities come from a model, available upon request, similar to the one reported in Table 2, which additionally includes the three-way interactions between type of couple, female respondents’ migrant status, and a simplified version of the age difference variable, with only two categories—she is or is not older than him. The predicted probabilities are marginal: The probabilities of being violently abused are averaged across exploited and nonexploited women; and those of being exploited are averaged across women who are or are not violently abused.
Probabilities of Violent and Nonviolent Abuse Under Different Specifications.
Figure 3 shows that a native woman’s higher probability of being violently abused in mixed couples does not change when she is older than her migrant partner, contrary to what one would expect if less attractive women accepted abusive relationships in exchange for forming partnerships with younger males who are less accessible in native markets. Migrant women’s probabilities of being victimized in mixed couples are similarly low for women who are and are not older than their partners. Nor do the probabilities of being exploited at home change for native and migrant women in mixed couples according to whether they are older than their partners. We conclude that selection effects do not seem to play an important role in explaining the impact that native status has on the probabilities of being violently and nonviolently abused in mixed couples.
Summary and Discussion
The goal of this article was to test two theories—exchange theory and status inconsistency theory—in an effort to explain the two main forms in which men’s power is exerted over women in couples—violent abuse and domestic exploitation (an unfair and oppressive division of household chores). Exchange theory explains abuse as a result of a balance of power tilted in men’s favor, and hence it expects abuse to increase in frequency and intensity the more dependent women are and the fewer resources they have relative to men. In exchange theory, an exploitative division of labor would make the first step toward violence. In contrast, status inconsistency theory expects men to exert violence against their intimate partners when women’s resources exceed men’s, as an instrument to restore men’s patriarchal power and domination at home.
We argued that the two theories could not be validly tested on cross-sectional data without simultaneously estimating the two forms of power exertion, due to their intercorrelations, and without blocking the reciprocal effects that an exploitative division of domestic work and violent abuse have on men’s and women’s relative resources. We argued that both could be effectively achieved by simultaneously estimating two equations—one for each form of abuse—with contemporaneous cross-equation error correlation, and by focusing on the distribution of ascribed resources which, like ethnicity or nationality, are less affected by power relations within couples.
Using data from a large victimization survey conducted in Spain in 2011 with a representative sample of adult women residing in the country, we investigated if women’s probability of being domestically exploited and violently abused in mixed couples formed by natives and migrants varied depending on whether she was the migrant (doubly disadvantaged) or the native (asymmetrically advantaged). The results confirmed that native women have a higher probability of being violently abused in mixed-couple relationships, but that, like migrant women, have a lower probability of being domestically exploited in heterogamous than in homogamous intimate relationships. On one hand, the results point to the presence of socio-cultural conditions favoring both forms of abuse in migrant and less educated families. These conditions are unrelated to the relative distribution of resources within couples and could point to the prevalence of cultures of violence and the presence of more stressful conditions in some specific contexts where these families are embedded. On the other hand, the results confirm the role played by a distribution of resources within the couple that favors women (whether in the form of a higher native status or higher education than her partner) in fostering violent abuse. The results fully support status inconsistency theory and refute the predictions of exchange theory.
Under specific assumptions, status inconsistency theory could provide the basis for a general theory of patriarchy. In this theory, violence acts as a key instrument that some males use to correct a disadvantageous balance of power when women outpower them. Assuming that the strong correlation documented here between violent abuse and domestic exploitation captured a functional relationship in which, at least partly, the former caused the latter, violence would lead to unfair distributions of housework, more opportunities for males to obtain better resources outside the household, and more uneven balances of power within couples favoring men. At this point, there would less need for males to resort to violence to subjugate their female partners; this is implied by the negative correlation between a relative distribution of resources favoring men and violence depicted in Figure 1 and documented in Table 2. Domination would substitute for the exertion of nude power, especially if complemented with patriarchal ideologies justifying women’s subjugation. Without these ideologies, the system would be unstable, for less violence might also foster less exploitation and break women’s dependence, thus sparking a new cycle of violent outbreaks to restore the “traditional” balance of relations.
More research is needed to test the assumptions on which this general theory rests. The cross-sectional nature of our data set precluded us from disentangling the direction of the effects between an exploitative division of domestic labor and violence. This could be done more effectively with panel data where couples’ relationships can be tracked over time. Yet, this limitation does not affect the study’s main contribution, which has been to demonstrate that violence against women ensues in a significant degree from power imbalances favoring women as a way for some men to reassert their superior status. This finding has important practical implications, for it points toward the possibility that IPV may rise along with women’s empowerment in society (Choi et al., 2014). It is consistent with other findings that have puzzled researchers (European Union Agency for Fundamental Rights [FRA], 2014), like the higher rates of IPV reported by women in the more gender-egalitarian countries of Northern Europe than in traditionally more patriarchal nations. Several reasons have been suggested for this puzzle—from over-reporting and different methods of data collection to the higher prevalence of alcohol abuse in the Nordic countries (FRA, 2014)—without fully succeeding in accounting for these cross-national differences (Gracia et al., 2019). Future research should investigate if this paradox can be solved by considering the higher prevalence of status inconsistent relationships in more egalitarian countries where hypogamous couples—those in which women have higher statuses than men—are more common. One lesson that we might extract from this and other research on status inconsistencies within couples is that the path to gender equality is likely to be plagued with difficulties and violent reactions to restore men’s patriarchal power, and that, to maximize its success, women’s empowerment may have to be combined with changes in cultural values which prevent men from experiencing it as a threat to their masculinity.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research has been supported by the “Fundación Recercaixa (PR02117—RECERCAIXA 2016ACUP00107) as part of the project, “An Integral Study of Intimate Partners’ Violence Against Women: Individual and Relational Factors that Increase the Risks of Perpetrating and Experiencing IPV.”
