Gender Distrust and Intimate Unions among Low-Income Hispanic and African-American Women

Abstract

We investigate levels of generalized distrust of men among low-income African American, Mexican, Puerto Rican, Dominican, and non-Hispanic white women in a three-city survey. The results reveal substantial variation. We find Hispanics' overall levels of distrust to be higher than levels for either African Americans or whites. Among Hispanics, however, Dominicans are the most distrusting group followed by Puerto Ricans; whereas Mexicans report levels of distrust that are comparable to African Americans and non-Hispanic whites. Married women are less distrusting than cohabiting women who, in turn, are less distrusting than non-cohabiting women. Nevertheless, distrust is not a significant predictor of a woman's total number of lifetime marital and cohabiting relationships; and distrust only marginally predicts a woman's desire to be in a steady relationship. We suggest that studies of trust in this population should focus more on attitudes displayed in specific encounters than on overall, generalized attitudes about gender distrust.

Introduction

The concept of “distrust” has been prominent in the recent literature on the decline of marriage among low-income women and men. Previous research describes high levels of generalized gender distrust among low-income women and the important role distrust may play in the low rates of marriage. Although these studies have been informative, we would argue that the concept of distrust is being applied too globally, as if it were equally important across racial and ethnic groups and as if it were uniformly predictive of low-income women's behaviors with regard to intimate relationships. Instead, we will claim, distrust should be seen as varying by social context and as of more limited, if still important, value in predicting the formation and dissolution of relationships. The specific social context we will explore is the variation in family patterns that were common in the country of origin (or island of origin for Puerto Ricans) of women who are Hispanic immigrants and their descendants as compared to the family patterns of African Americans and non-Hispanic whites. We will examine how this variation is associated with levels of gender distrust; and we will further study how levels of distrust may be associated with women's histories of intimate relationships.

In particular, we will contrast native-born non-Hispanic whites and African Americans with women from two regions with different patterns of cohabitation, marriage, and non-marital births, namely Mexico and the Caribbean. Further, we will consider nativity status for Mexicans (U.S.-born versus Mexican born) and Puerto Ricans (mainland- versus island-born). Our data come from a study of African-American women, non-Hispanic white women, and women from the first, second, and fourth largest Hispanic groups in the United States: Mexicans, Puerto Ricans and Dominicans, respectively (U.S. Bureau of the Census, 2004). The families resided in low-income neighborhoods in Boston, Chicago, and San Antonio.¹ The presence of both Puerto Ricans and Dominicans in sufficient numbers to allow for separate analyses is unusual in surveys. It allows us to examine the potential contrasts between a Caribbean and a Mexican context of migration; and within the Caribbean context, it allows us to examine potential differences between Puerto Rican and Dominican immigrants and their descendants. Moreover, the Dominican and Puerto Rican groups, because of their relatively high rates of poverty and single-parent families, provide an interesting comparison group to African Americans. We will compare levels of trust and then investigate whether distrust is associated with three outcomes: the number of cohabiting or marital unions a woman has experienced; whether, if she is not currently in a relationship, she desires to be in one; and her current union status. Our results will suggest that levels of distrust vary across racial and ethnic categories, as well as among Hispanic sub-groups. The results will also suggest that, for all groups in the sample, expressed levels of gender distrust are not always good predictors of relationship and union formation.

Background

According to recent studies, generalized gender distrust among low-income women appears to be quite prevalent and may exist across several domains (Carlson, McLanahan, & England, 2004; Edin, England, & Linnenberg, 2003). Edin and Kefalas (2005, pp. 126-127) write “…they don't trust them to remain faithful, to stay working, to pay their share of the bills, to not beat them or abuse their children, to stay out of trouble with the law, and to stay free of addiction to alcohol and drugs.” In a study of 40 unmarried parents, more than half report at least one instance of infidelity (by either parent) while the couple was in a relationship (Hill, 2007). Moreover, the growth of multipartnered fertility among unmarried parents (Carlson & Furstenberg, 2006) is increasingly creating a context in which fathers or mothers may have continued contact with previous partners with whom they have had a child, thereby exacerbating distrust and fears of infidelity. Domestic violence and sexual abuse are other sources of distrust commonly described in the literature. Studies suggest long-term impacts on an abused individual's ability to trust and form intimate relationships (Macmillan, 2001; McNeal & Amato, 1998). Exposure to abuse can also lead to mental health problems such as depression and to errors in judgment about trusting partners in relationships (McCloskey & Figueredo, 1995). While for some women these processes may lead to fewer intimate partnerships, other studies suggest that women with a history of childhood abuse may, as adults, enter into more relationships than women without a history of abuse (Elliott, Avery, Fishman, & Hoshiko, 2002; Polusny & Follette, 1995). For instance Cherlin, Burton, Hurt, and Purvin (2004) reported that women who had been abused sexually as children were less likely to be in sustained marital or cohabiting unions as adults but more likely to have multiple short-term unions as an adult.

Research on the causes and consequences of gender distrust, however, has not been explicitly comparative across racial and ethnic groups. And yet the potential variations in distrust could be instructive in comparing the major groups. Much of the early literature on gender distrust was written at a time when there were fewer Hispanics in the United States population and when the poverty debates, correspondingly, focused on African Americans. That was the context for studies such as Hannerz (1969) and Rainwater (1970). Even recent studies, however, have had limited information on Hispanics. Edin and Kefalas (2005) compared low-income African Americans and non-Hispanic whites to Puerto Ricans, the only Hispanics who lived in substantial numbers in the neighborhoods that they studied.

Given the varied experiences of family life and gender relations among specific Hispanic groups, levels of gender distrust may differ. In the next sections we describe patterns of family life in both countries of origin and in the United States that could be relevant to levels of distrust for Hispanics of Mexican, Dominican, and Puerto Rican descent. We focus on a combination of conditions in the sending countries including prevailing family forms, roles for men and women, and childhood socialization. In turn, these factors, to differing degrees, affect family life for Hispanics in the U.S. We argue that conditions abroad and in the U.S. may shape attitudes about gender distrust as well as ensuing relationship behaviors.

Family life in Mexico, at one end of the spectrum, is described as patriarchal while the Dominican Republic, at the opposite end of the spectrum, is characterized as matrifocal, with Puerto Rico in the middle of this continuum (Massey, Fischer, & Capoferro, 2006). The patriarchal system includes the traditional household structure in which males dominate and marriage is the norm. On the other hand, matrifocal societies are defined by the degree of female autonomy, low rates of legal marriage, high rates of nonmarital childbearing, and a large proportion of female-headed households (Safa, 2005). Dominicans are said to exemplify the Caribbean matrifocal model centered on the mother-child as the focal unit with men playing only marginal roles (Barrow, 1996; Massey et al., 2006). In the matrifocal setting of the Dominican Republic, informal unions are quite prevalent as is unwed childbearing and union disruption. Using Mexican and Dominican census data, Massey et al. (2000) showed that the ratio of marriages to informal unions and marital disruptions supports these categorizations. Based on Mexican census data from 2000, of those aged 15 and over the ratio of marriages to consensual and disrupted unions (divorced/separated) was 2.9. In contrast, 1993 Dominican census data show a ratio of 0.6 marriages to consensual and disrupted unions. As is the case in Mexico and other patriarchal societies, norms in Puerto Rico favor legal marriage; more than half of the 15 and over population is married in Puerto Rico (compared to forty six percent in Mexico and twenty one percent in the Dominican Republic). However, marital disruption and female-headed households in Puerto Rico are prevalent, suggesting more similarities with matrifocal Dominican Republic.

Several aspects of the Caribbean matrifocal model may contribute to greater gender distrust. First, as described previously, men play less of a role in family life leaving women to depend on other women –including immediate family members and extended kin networks– for economic and emotional support and often to help raise their children. Safa (2005) writes “These consanguineal ties are seen as lasting in comparison to the often transient nature of the conjugal bond. Men are important for sexuality and reproduction, but their support cannot be relied upon (p. 333).” Thus, in matrifocal settings there may be long-standing relationship behaviors and expectations between men and women. Second, multipartnered fertility appears to be common. In a case study of a Dominican transnational village (its counterpart in Boston, Massachusetts), Levit (2001)writes

….men also frequently form stable unions with a “second wife.” They have a primary household where they sleep, eat, and keep their belongings, and a secondary household that they also contribute money to, visit regularly, and feel a sense of commitment toward. These types of relationships are an integral part of community life”(p. 76).

The author concluded that the community expects young men to sow their “wild oats,” have extramarital affairs, and not need to settle down until middle age. In fact, men who did not “go out looking” had their manhood called into question.

Third, the combination of multipartnered fertility and the marginal role of men in childrearing contribute to “child shifting,” or change(s) in the child's residence. In one study based primarily on English-speaking Caribbean countries, between 15 and 30 percent of children grew up with relatives or neighbors rather than with a parent (Evans & Davies, 1997). Another study of 1,600 homes found that 50 percent of children were cared for by female relatives other than their mothers (Leo-Rhymie, 1997). This sometimes happens when a mother is in a new “visiting union” or if she leaves the country to work. The former study concludes that the shifting of residences and relationships may prevent children from creating bonds of attachment (Evans & Davies, 1997) and developing stable, secure human relationships (Leo-Rhymie, 1997). In this way, the early socialization of Caribbean children could very likely build the foundation for gender distrust as adult women. Thus, characteristics of patriarchal and matrifocal societies provide the setting for expectations about men and women and the roles and relationships they enter. It follows that these experiences affect migration streams as well as family life in the U.S. for these Hispanic groups.

As for the American context, the literature on Hispanic family life in the U.S. has emphasized the role of culture, specifically familism. This concept, about which there has been much debate, connotes a commitment to family expressed, for example, through strong relationships with nuclear and extended family members and the presence of married-couple households. It has been the dominant cultural explanation for the high rates of marriage among Hispanics despite their poor economic standing (Landale, Oropesa, & Bradatan, 2006). As scholars have shown, however, the concept of a marriage-centered familism applies to Hispanics of Mexican and Cuban origin more than to Hispanics of Puerto Rican origin; and among Mexicans it applies more to first-generation immigrants than to the second or third generation (Oropesa & Landale, 2004). According to an analysis of Current Population Survey data from 1998, 2000, and 2002, there is substantial variation in household structure among the diverse groups that constitute the Hispanic population. This variation appears to follow patterns in the countries of origin. In particular, the percentage of single-parent households diverges sharply among Hispanics of Mexican, Puerto Rican, and Dominican origin. Eighteen percent of Mexican family households were headed by a woman, not far above the 11 percent figure for non-Hispanic whites. But 34 percent of Puerto Rican family households were headed by women, as were 42 percent of Dominican family households. Indeed, the figure for Dominicans was almost identical to the figure of 42 percent for non-Hispanic African Americans (Landale et al., 2006). Moreover, of all groups, Dominican households were most likely to include extended families: 12 percent had relatives that were not part of the nuclear family, compared to 8 percent of Mexicans, 7 percent of Puerto Ricans, 7 percent of non-Hispanic African Americans, and 3 percent of non-Hispanic whites (Landale et al., 2006). Clearly, we might expect Dominicans, and to a lesser extent Puerto Ricans, to differ from Mexicans in their attitudes to relationships with men and union formation.

The age and sex distribution among Hispanics of different origins may shed some light on their distinct family formation behaviors. Based on Current Population Surveys from 1997 through 2002, Durand, Telles, & Flashman (2006) report that among Americans of Mexican origin, the male and female populations are similar in age and nativity. These demographics reflect long-standing immigration from Mexico. In contrast, Dominicans are part of a newer wave of immigrants coming mostly after 1970. As described previously, women in the Dominican Republic are often independent actors and research suggests their autonomy may lead to a heavily female immigration stream while Mexican migration is more often tied to their male partner's experiences (Massey et al., 2006). In 2002 the sex ratio for the foreign born from Mexico was 124 men per hundred women, while the Dominican figure was 68 men per hundred women (Grieco, 2003). Puerto Rico also has a substantial female migration; but unrestricted access to the United States mainland and a longer history of migration probably makes the sex ratio for Puerto Rican migrants more balanced than for Dominican migrants. In 2006 the sex ratio for Puerto Ricans on the mainland was 95 men per hundred women, about mid-way between the ratios for Mexicans and Dominicans (U.S. Bureau of the Census, 2007). Thus, the availability of potential mates in the U.S. may affect union formation independent of attitudes about gender distrust.

In addition, although the present study is interested in the relationship between gender distrust and union formation; we also acknowledge the very important role other factors play in determining union formation. A long tradition of research on marriage points to the role of economic incentives. Becker (1973) writes that men and women marry for the gains achieved by such a union; women gain by trading in their housework and men gain by trading their incomes, hence raising utility levels for each partner. As women work, the gains to marriage decline and they have less need to marry (Oppenheimer, 1997). As such, we include measures of employment and education to capture women's economic situations. In the same vein, some research suggests that cash assistance receipt (Temporary Assistance for Needy Families, or TANF) may also lead to economic independence among single mothers therefore we include a measure of current TANF receipt (Moffitt, 1992). We also know that intimate partner violence is related to both relationship behaviors and the ability to trust. Women's mental health may affect their abilities to make decisions about relationships or to maintain them, in part due to the experience of violence. Finally, some women enter relationships for more practical needs such as needing emotional support, money, or child care. Given these associations with relationship behaviors, we will include measures of intimate partner violence, mental health status, and the need for assistance. We discuss these measures further in the following section.

Data and Methods

Study Design

The data are from a study of the well-being of children and their families in low-income neighborhoods in Boston, Chicago, and San Antonio (Winston et al., 1999). The study included a random-sample survey of 2,402 children (age 0 to 4 or age 10 to 14) and their female caregivers in low-income neighborhoods in the three cities. ² In over 90 percent of the cases, the caregiver was the child's mother. Interviews were first conducted in 1999, and the response rate was 74 percent. This paper draws upon the 80 percent of the 1999 sample of caregivers who were successfully interviewed in a 2005 survey wave that included questions on attitudes toward trust, relationships, and marriage. The survey data are weighted to correct for oversampling and to give equal weight to the experiences of families in each city.³ Of the 2,056 caregivers completing interviews in 2005, 21 were male caregivers and 32 reported their race/ethnicity as “other.” We exclude these cases from our analyses leaving us with a sample of 2,003 women. Additionally, we examine a sub-set of women who were not currently in relationships at the time of the 2005 interview. We describe this further sample restriction in our results section.

Key Variables

The 2005 survey included questions asked of women about generalized distrust in men and about whether or not they desired to be in a relationship. Using these data, we create two scales: a distrust scale and a relationship scale. The distrust scale, with an alpha reliability of .64, comes from the responses to two items: (1) “I find it difficult to trust most men;” and (2) “I believe that most men cannot be sexually faithful to one woman.”⁴ We use the distrust scale to measure overall levels of gender distrust and present descriptive statistics in our results sections. This scale is our key independent variable in our multivariate analyses. Women who said they were not currently in a relationship were asked a set of items about their desire to be in a steady relationship. Their responses to the following five items formed a relationship scale with an alpha reliability of .82: (1) “I'd like to be in a steady relationship now;” (2) “I feel I am ready for a steady relationship;” (3) “I need some distance from romantic relationships now;” (4) “I'm looking for someone to have a steady relationship with;” and (5) “I'm too busy right now to think about having a steady relationship.” The scales were coded so that higher scores indicated greater distrust and a greater desire to be in a relationship, respectively. Desire to be in a relationship is our first dependent variable.

We use retrospective relationship histories collected in the 2005 survey to create the total number of lifetime unions. (If a woman had started a relationship by cohabiting and then had married her partner, this was counted as one union.) The total number of unions is a second dependent variable. We also use information from the 2005 survey on current union status, as follows: We created four mutually exclusive categories: not in a relationship; in a relationship (but not living together); cohabiting; and married. These categories serve as our final dependent variable in our multivariate analyses.

Independent Variables

In addition to the distrust scale we described previously, our multivariate analyses includes the Brief Symptom Inventory (BSI), a widely-used, 18-item measure of psychiatric symptomology that combines indicators of depression, anxiety, and somatization. We include a transformed version of the total score (natural logarithm of the total score plus 1). The BSI scale has an alpha reliability of 0.94.

Based on previous clinical studies and other research using Three-City Study data, we include a measure of intimate partner violence. We use a dummy variable measuring whether or not the respondent had ever experienced abuse by a partner. She was asked “Has anyone you have been in a romantic relationship with ever.…” 1) slapped, kicked, bit, or punched; 2) beaten; 3) choked or burned; or 4) used a weapon or threatened to use a weapon on her.⁵ A woman who reports experiencing any of these four items is considered having experienced intimate partner violence.

Our race/ethnicity categories are Non-Hispanic white, Dominican, foreign-born Mexican, U.S.-born Mexican, island-born Puerto Rican, mainland-born Puerto Rican, Non-Hispanic African American (which we will sometimes refer to as “African American”), and a residual category, “Other Hispanic.” (The latter category comprises a diverse group of Central Americans and South Americans.) Since the Dominicans in this sample are almost entirely foreign born (99 percent) we will refer to this group as simply “Dominican.” Based on research suggesting generational differences in familism among immigrants of Mexican descent in particular, we separate Mexicans into first generation (Mexican-born) and second or higher order generation (U.S.-born). In the case of Puerto Ricans, we consider island-born and mainland-born Puerto Ricans separately in light of the matrifocal Caribbean influence on the island. The Hispanic sub-groups clustered in specific cities: most of the Mexican Americans resided in San Antonio or Chicago, most of the Puerto Ricans were in Boston or Chicago, and most of the Dominicans in Boston. We therefore also control for city of residence (Boston is the reference category in the multivariate analyses).

We include a 4-item Assistance Scale which represents whether the respondent said she needs assistance in several domains, with higher scores reflecting greater need. She was asked whether she has enough people she can count on, too few people, or no one she can count on in terms of 1) needing someone to listen to her problems when she's feeling low; 2) needing someone to take care of her child/children when she isn't around; 3) needing help with small favors; and 4) needing someone to loan her money in an emergency. The scale's alpha reliability is 0.82.

To measure the respondents' general health we include a dummy variable reflecting whether the respondent reported her general health to be fair or poor (as opposed to excellent, very good, and good). To capture economic factors that may affect relationship behaviors, we include current employment status and current TANF receipt. Finally, we include the traditional background variables of age in years and education (measured by the presence of a high school diploma at the time of the 2005 interview).

Methods

We use OLS regression to predict both the score on the relationship scale and the total number of lifetime cohabiting and marital unions. In other analyses not presented here we analyzed the total number of unions using Poisson regression as well as models using a top-coded version of the number of unions. All methods yielded similar results therefore we report the OLS results. For the final outcome, current union status during the 2005 interview, we use multinomial logistic regression to predict the likelihood of being in a union (a non-cohabiting relationship, a cohabiting relationship, or a marriage) relative to the reference category (no relationship).

Missing Data

As described, these analyses are based on the pool of caregivers re-interviewed in 2005 (the third wave of data collection), or 80 percent of the original sample. The sample of women lost to attrition does not differ significantly in age, race and Hispanic ethnicity, educational attainment, welfare usage, or marital status from the sample of the caregivers in our current analyses. The 2005 sample does have significantly fewer respondents from Boston, which is the city with the lowest response rate. We did not encounter any substantial missing data issues on particular variables. Our item non-response is no more than 5 percent for any variable. Nevertheless, we use multiple imputation (Acock, 2005) to replace missing data by creating 5 complete imputed data sets using Stata's “ice” routine (Royston, 2005). Our results come from the joint analyses of these 5 data sets. By conducting sensitivity analyses, we find our substantive interpretations remain the same whether we use list-wise deletion (versus multiple imputation) and whether we use 5 or more imputed data sets.

Results

Descriptive Statistics

Table 1 presents the weighted descriptive statistics. Based on the sample of 2,003 women, 41 percent are Non-Hispanic African American, and 6 percent are Non-Hispanic white. The largest Hispanic group in the sample is Mexican (33 percent of entire sample). More than three-quarters of the Mexicans are U.S.-born. There are also substantial numbers of Puerto Rican and Dominican women, 9 and 5 percent respectively. All other Hispanics are grouped into “Other Hispanic.” The mean age of the sample is 39 years and slightly more than half have attained a high school diploma or higher. (See Table 1 for descriptive statistics on all other independent variables)

Table 1. Descriptive Statistics (Weighted).

Demographics (n=2,003)
Race/Ethnicity
Dominican	5%
Mexican Foreign Born	9%
Mexican US Born	24%
Puerto Rican Island Born	5%
Puerto Rican Mainland Born	4%
Other Hispanic	6%
Non-Hispanic African American	41%
Non-Hispanic White	6%
Age	39
High School Diploma	51%
BSI	1.30
Fair or Poor Health	32%
Assistance Scale	1.61
Intimate Partner Violence	38%
Employed	54%
Welfare	11%

Using data from the 2005 survey, we also examine relationship status; total number of lifetime unions; general levels of gender distrust; and desire to be in a relationship. Slightly less than a third of the sample is married and approximately 9 percent are in cohabiting relationships. A full 38 percent of women report not being in any type of relationship at all (Table 2a). This figure is driven by the large percentage (nearly half) of African-American women who report not currently being in a romantic relationship. On the other hand, the married group is driven by the high percentage of non-Hispanic white and Hispanic women who report being currently married (slightly more than 40 percent each). The high rate of marriage among the Hispanic sample masks sharp differences among Hispanic sub-groups (Table 2b). The foreign-born Mexican women have the highest rate of marriage; more than 70 percent report being currently married. Marriage figures for Dominican and Puerto Rican women in the sample are less than half of the overall Mexican marriage rate. Many of the Dominican and island-born Puerto Rican women report not actually being in any sort of relationship, 58 percent and 51 percent respectively, at the time of the 2005 interview.

Table 2.

Table 2a. Relationship Status by Race (Weighted)
	% Overall Sample	% Non-Hispanic White	% Non-Hispanic African American	% Hispanic
Current Relationship Status (n=2,003)
Not in a Relationship	38	33	48	31
In a Relationship (Not Cohabitin	21	19	28	16
In a Cohabiting Relationship	9	6	10	10
Married	31	42	14	43
Total	100	100	100	100

Table 2b. Relationship Status by Hispanic Ethnicity (Weighted)
	% Dominican	% Mexican Foreign Born	% Mexican US Born	% Puerto Rican Island Born	% Puerto Rican Mainland Born	% Other Hispanic
Current Relationship Status (n=2,003)
Not in a Relationship	58	12	29	51	28	30
In a Relationship (Not Coh)	18	10	18	18	30	12
Cohabiting Relationship	1	5	10	11	17	15
Married	22	73	44	21	25	43
Total	100	100	100	100	100	100

The overall sample of women has an average of 1.3 lifetime cohabiting and marital unions (results not shown here). Whites and Hispanics have significantly more relationships than African Americans. In analyses including major Hispanic groups we find that, in general, foreign- (or island-) born Hispanics tend to have had the fewest relationships, along with African Americans. This difference appears to be driven by more marriages among the Mexican foreign born and more time spent without partners among Dominicans, foreign-born Puerto Ricans, and African Americans.

Based on the distrust scale, we examine our first research question about the prevalence of generalized gender distrust in a sample of low-income women (Table 3). The mean score among the sample, 2.71 on a scale from 1 through 4, suggests that, on average, low-income women possess somewhat high levels of generalized gender distrust. We find Hispanics as a group have the highest level of distrust among the sample of low-income women (significantly different from either African Americans or non-Hispanic whites). Considering Hispanic sub-groups separately, however, we find that Dominicans are by far the most distrusting group followed by Puerto Ricans. These two groups along with the “other Hispanic” sub-group drive the difference between Hispanics as a whole and non-Hispanics. In contrast, there is no significant difference among Mexicans, African Americans, and non-Hispanic whites.⁶

Table 3. Distrust by Race and Hispanic Ethnicity (Weighted).

	Mean Score
Overall Sample (n=2,003)	2.71
Hispanica	2.77
Dominicanb	3.37
Mexican FOBc	2.52
Mexican USBd	2.63
PR FOBe	3.06
PR USBf	2.91
Other Hispanicsg	2.92
Non-Hispanic African Americanh	2.64
Non-Hispanic Whitei	2.57

^aHispanic significantly different from African American and White

^bDominican significantly different from Mexican FOB, Mexican USB, PR FOB, PR USB, Other Hispanic, African American, and White

^cMexican FOB significantly different from Dominican, PR FOB, PR USB, and Other Hispanic, and African American

^dMexican USB significantly different from Dominican, PR FOB, PR USB, and Other Hispanic

^ePR FOB significantly different from Dominican, Mexican FOB, Mexican USB, PR USB, Other Hispanic, African American, and White

^fPR USB significantly different from Dominican, Mexican FOB, Mexican USB, Puerto Rican FOB, African American, and White

^gOther Hispanic significantly different from Dominican, Mexican FOB, Mexican USB, PR FOB, African American, and White

^hAfrican American significantly different from Hispanic, Dominican, Mexican FOB, PR FOB, PR USB, and Other Hispanic

ⁱWhite significantly different from Hispanic, Dominican, PR FOB, PR USB, and Other Hispanic

Finally, among women who report that they are not currently in a romantic relationship, we find an average score of 2.17 on the relationship scale (results not shown here). Based on this scale of 1 through 4, with higher scores indicating more desire to be in a relationship, overall, women appear only moderately interested in wanting to be in relationships. The scores seem similar across non-Hispanic whites, African Americans, and Hispanics, although African-American women score significantly higher than Hispanic women as a group.

Multivariate Analyses

Table 4 presents the results of an OLS regression analysis in which the dependent variable is a woman's score on the relationship scale. (The case base is all women who were not currently in a relationship.) We see that generalized distrust is only a marginally significant (p < .10) predictor of whether women desire to be in a relationship. A one unit increase in the distrust scale is associated with less desire to be in a relationship but only marginally so. In contrast, a woman's score on the BSI is a highly significant predictor (p <.001), and the sign of the coefficient shows that women with more psychiatric symptoms had higher scores on the relationship scale – that is, they had a greater desire to be in a relationship. In other words, among women not in relationships, distrust was only marginally related to women's reports of wanting to be in a steady relationship while those showing indications of poorer mental health were significantly more likely to say they wanted to be in a relationship. Women in fair or poor health (versus good or excellent health) had a marginally lower desire to be in a relationship (p<0.10), but in other regression specifications, health sometimes lost even this marginal level of significance.

Table 4. Regression of Relationship Scale Score on Selected Indicators.

	Coefficient
Distrust Scale Score	-0.108* (1.95)
Chicago	0.125 (-1.24)
San Antonio	0.064 (-0.52)
Brief Symptom Inventory Score	0.162**** (4.81)
Intimate Partner Violence	0.04 (-0.55)
Assistance Scale	-0.027 (-0.4)
Fair or Poor General Health	-0.131* (1.67)
Dominican	0.343** (2.40)
Mexican Foreign Born	-0.062 (-0.53)
Mexican US Born	-0.255* (1.83)
Puerto Rican Island Born	-0.043 (-0.31)
Puerto Rican Mainland Born	0.216 (-1.02)
Other Hispanics	0.031 (-0.19)
Non-Hispanic White	-0.091 (-0.63)
Age	-0.006* (1.83)
High School Diploma	-0.022 (-0.27)
Employed	0.238*** (2.87)
Receiving TANF	0.138 (-1.25)
Constant	2.413**** (9.21)
R-squared=0.15
n=871

Notes: t statistics in parentheses

^*p<.10,

^**p<.05,

^***p<.01,

^****p<.001

Being Dominican is strongly associated with a higher desire to be in a relationship among women who are not currently in one. On the other hand, being a U.S. Born Mexican is significantly associated with less desire to be in a relationship. We also find a strong association between employment status and desire to be in a relationship; employed women show more interest in relationships. We can only speculate that more economically independent women may feel they have more to offer a partner or may feel they are in a good position to attract a mate. As for the persistent relationship between Dominican women and their desire to be in a relationship, we offer the possibility that the aforementioned gender imbalance among Dominican foreign-born, 68 men per hundred women (Grieco, 2003), may explain the greater demand for a relationship. We will comment later on the distinctive Dominican pattern of high distrust and relatively high desire to be in a relationship.

In Table 5 we present the results of a regression analysis in which the dependent variable is the total number of reported lifetime marital and cohabiting relationships. Distrust does not predict how many lifetime unions a woman has experienced. In this case, women who had experienced intimate partner violence as adults reported a significantly (p<.001) larger number of relationships. With African American as our reference group, we find that mainland-born Puerto Rican women report having more relationships. Older women reported a significantly greater number of partnerships.

Table 5. Regression of Number of Unions on Selected Indicators.

	Coefficient
Distrust Scale Score	-0.011 (-0.27)
Chicago	0.134* (1.65)
San Antonio	0.017 (-0.23)
Brief Symptom Inventory Score	0.048 (-1.62)
Intimate Partner Violence	0.27**** (4.39)
Assistance Scale	0.029 (-0.54)
Fair or Poor General Health	0.048 (-0.74)
Dominican	0.053 (-0.5)
Mexican Foreign Born	0.037 (-0.45)
Mexican US Born	0.083 (-0.99)
Puerto Rican Island Born	-0.028 (-0.3)
Puerto Rican Mainland Born	0.394*** (2.74)
Other Hispanics	0.148 (-1.21)
Non-Hispanic White	0.051 (-0.53)
Age	0.01*** (3.46)
High School Diploma	0.078 (-1.26)
Employed	0.045 (-0.7)
Receiving TANF	-0.039 (-0.36)
Constant	0.544** (2.54)
R-squared=0.06
n=2003	0.06

Notes: t statistics in parentheses

^*p<.10,

^**p<.05,

^***p<.01,

^****p<.001

The results thus far suggest that while distrust is not a strong predictor of desire to be in a relationship nor the total of lifetime unions, mental health is a strong predictor of the former outcome and a history of intimate partner violence is a stronger predictor of the latter. And it is striking that poorer mental health and the experience of intimate partner violence predict greater involvement. The association of number of relationships with intimate partner violence is consistent with a similar result in Cherlin et al. (2004); and it is consistent with clinical studies suggesting that one of the common sequelae of experiencing intimate partner violence is a subsequent tendency to more easily enter into relationships, many of which also become abusive (van der Kolk, 1989).

While generalized distrust does not strongly predict the number of lifetime unions or desire to be in a relationship, we do find it to be associated with current union status (see first Row of Table 6). Increases in the distrust scale score are associated with lower likelihoods of being in a relationship of any kind (non-cohabiting, cohabiting, or marital) relative to being single. In accord with the assumption that marriage is a more committed relationship than cohabitation and cohabitation in turn is more so than a non-cohabiting relationship, we see that trust plays the biggest role in the most committed relationship and decreases linearly. A unit increase in the distrust scale reduces the likelihood of being in a non-cohabiting relationship by 22 percent ([1-0.78]× 100), the likelihood of cohabitation by 41 percent and the likelihood of being married by 53 percent. These results support other recent survey research on trust and union formation (Carlson et al., 2004). Whites and all Hispanic groups except island- born Puerto Ricans have an increased likelihood of being married compared to African Americans (the reference category). And Foreign Born Mexicans, in particular, are almost 17 times more likely to be currently married than African Americans. Some of the Hispanic groups are also more likely to be cohabiting: U.S.-born Mexicans, mainland-born Puerto Ricans, and Other Hispanics are all more likely to cohabit than remain single relative to African-Americans. Dominicans on the other hand are less likely to cohabit.⁷

Table 6. Multinomial Logistic Regression of Union Status on Selected Indicators.

	1. In Relationship	2. Cohabiting	3. Married
Distrust Scale Score	0.781* (-1.94)	0.587*** (2.96)	0.472**** (5.04)
Chicago	0.665 (-1.16)	0.655 (-0.86)	1.225 -0.48
San Antonio	1.03 (-0.13)	1.101 (-0.26)	1.046 -0.15
Brief Symptom Inventory Score	1.129 (-1.3)	1.148 (-0.99)	0.988 -0.12
Intimate Partner Violence	1.325 (-1.42)	1.128 (-0.41)	0.734 -1.4
Assistance Scale	0.685** (2.09)	1.082 (-0.28)	1.354 -1.56
Fair or Poor General Health	0.802 (-1.03)	2.715 (3.15)***	0.848 -0.68
Dominican	0.764 (-0.67)	0.198 (2.40)**	2.054 (1.66)*
Mexican Foreign Born	1.807 (-1.15)	2.193 -1.24	16.673 (7.01)***
Mexican US Born	1.391 (-0.89)	2.275 (1.74)*	4.111 (3.60)***
Puerto Rican Island Born	0.731 (-0.9)	1.341 -0.51	1.826 -1.35
Puerto Rican Mainland Born	1.614 (-1.16)	2.748 (1.86)*	3.353 (2.53)**
Other Hispanics	1.005 (-0.01)	4.346 (2.70)***	6.242 (4.19)***
Non-Hispanic White	1.145 (-0.3)	1.162 -0.2	5.052 (3.46)***
Age	0.946 (5.79)***	0.917 (4.98)***	0.98 (1.96)**
High School Diploma	1.174 (-0.74)	1.124 -0.38	1.112 -0.48
Employed	1.004 (-0.02)	1.253 -0.71	1.055 -0.23
Receiving TANF	0.798 (-0.81)	0.416 (1.96)*	0.303 (3.15)***
n=2003

Notes: Reference category is “Not in a Relationship”

Results presented are Relative Risk Ratios

t statistics in parentheses

^*p<.10,

^**p<.05,

^***p<.01,

^****p<.001

Conclusion

Our results have several implications for theorizing about the role of gender distrust in union formation among the low-income population. First, we found gender distrust to be highest among two Hispanic groups of Caribbean origin, Dominicans and Puerto Ricans. In contrast, we found little difference in distrust among non-Hispanic whites, non-Hispanic African Americans, and Americans of Mexican origin. Although most previous research on distrust in the low-income population has focused on African Americans, they did not appear to be more distrusting than non-Hispanic whites or U.S.-born Mexican Americans. Moreover, we would caution against considering all low-income Hispanics as a monolithic group in their level of gender distrust, just as previous studies have shown Hispanics not to be monolithic in their orientation toward marriage. Rather, low-income Hispanics' patterns of distrust in this survey, like their patterns of marriage, seem to reflect the social contexts of Mexico, on the one hand, and the Caribbean, on the other hand. Comparing foreign- (or island-) born Hispanics with U.S. (or mainland) born Hispanics, we find the following: Among Mexicans, those born in the U.S. are less likely to be married than the foreign born, whereas among Puerto Ricans, the mainland-born are more likely to be married than the island born. These differences may reflect the dominant role of marriage in Mexico versus the higher prevalence of non-marital unions in Puerto Rico.

The responses of Dominicans, a group rarely considered separately from other Hispanics, were the most distinctive. Dominicans reported the highest levels of gender distrust; but, among those not currently in relationships, Dominicans express great desire to be in relationships. We would suggest that this distinctiveness may reflect aspects of historical family patterns in the Dominican Republic and of the migration stream to the United States. Indeed, the relative recency of the Dominican migration stream (99 percent of the Dominicans women in this sample were first generation immigrants) may mean that intimate unions are influenced by the cultural and structural patterns of their place of origin. Studies of the social structure of the Dominican Republic suggest a woman-centered style of kinship with men playing a relatively smaller role. In tandem with multipartnered fertility and children's residence shifting, family patterns in Dominican society may begin to explain gender distrust among Dominican women in the U.S. As for migration characteristics, we noted earlier that Dominican migration has been more heavily female than is the case for Mexicans and, to a lesser extent, Puerto Ricans. Consequently, there is a greater gender imbalance – a shortage of men. The Dominican women in the sample were less likely to be in relationships than were other women. This imbalance in supply and demand may explain the seemingly paradoxical result that Dominican women, though highly mistrusting on average, also express great desire to be in relationships. An alternative explanation offered by Safa (2005) is that some matrifocal societies may still be influenced by the ideal of a nuclear family and the stigma attached to matrifocal households. She argues that while aware of the fragile nature of conjugal bonds, the young may still hope to find supportive partners.

In addition, we found that while women in all groups tended to express high levels of distrust of men, a woman's distrust was not a strong predictor of her desire to be in a steady relationship nor of the number of marital and cohabiting relationships she has had. These findings are consistent with ethnographic evidence from the Three-City study (Burton, Cherlin, Winn, Estacion, & Holder-Taylor, 2007) which suggests that women with high levels of generalized distrust of men may nevertheless deploy specific, limited forms of trust in starting intimate partnerships. These situationally-specific forms of interpersonal trust may even be used to facilitate a series of short-term relationships. Our finding that poorer mental health and a history of intimate partner violence are strong predictors of involvement in intimate partners is consistent with the notion of situationally-specific forms of trust co-existing with generalized distrust. Our findings suggest that low-income women's search for intimate partnerships may be influenced, in part, by anxiety- or depression-driven needs or by the effects of past trauma on the ability to make sound judgments about whether to trust potential partners. If the clinical literature is any guide, relationships formed on this basis are unlikely to be stable and satisfactory long-term.

Nevertheless, we did find that levels of generalized distrust are correlated with women's current relationship status. Women who were married at the time of the 2005 interview expressed lower levels of distrust than cohabiting women who were, in turn, less distrusting than non-cohabiting women. This pattern is consistent with other survey research that finds generalized gender distrust to be strongly associated with a woman's current union status, specifically co-residential unions. We caution that it is difficult to determine whether levels of distrust actually predict current union status or whether a woman's current union status influences her reported level of distrust. In any case, this is the main sense in which women's generalized distrust seems to be associated with their patterns of intimate unions.

Overall, our findings suggest that, contrary to what is sometimes assumed in the literature on marriage among disadvantaged populations, generalized feelings of gender distrust may not be a central influence on low-income women's and men's behavior with respect to intimate relationships. To be sure, women tend to express high levels of generalized distrust of men. But in the survey we analyzed, levels of distrust are not higher among African Americans, the group that has seen the greatest decline in marriage. A woman's distrust of men in general may be a poor indicator of how she behaves toward a particular partner. The disconnect between people's generalized and particular evaluations is not without precedent. Within the field of social psychology, research has described generalized trust as a separate entity from the interpersonal trust characteristic of relationships with partners (Couch, Adams, & Jones, 1996; Rempel, Homes, & Zanna, 1985). Based on the social psychological literature and the current study, we would suggest that scholars studying intimate partnerships among low-income women and men may be better advised to focus on the attitudes surrounding particular encounters rather than on a global, or generalized sense of distrust.