In the United States, low marriage rates and high divorce rates among the poor have led policymakers to target this group for skills-and values-based interventions. The current research evaluated the assumptions underlying these interventions; specifically, the authors examined whether low-income respondents held less traditional values toward marriage, had unrealistic standards for marriage, and had more problems managing relational problems than higher income respondents. They assessed these issues in a stratified random sample that oversampled low-income and non-White populations (N = 6,012). The results demonstrated that, relative to higher income respondents, lowincome respondents held more traditional values toward marriage, had similar romantic standards for marriage, and experienced similar skills-based relationship problems. Low-income groups had higher economic standards for marriage and experienced more problems related to economic and social issues (e.g., money, drinking/drug use) than did higher income respondents. Thus, efforts to save low-income marriages should directly confront the economic and social realities these couples face.
Key Words: family policy, low-income families, socioeconomic status, welfare reform.
Because divorce rates are higher and marriage rates are declining faster in lower income communities than in more affluent communities (Bramlett & Mosher, 2002; Pew Research Center, 2010), many politicians, advocacy groups, and researchers are concerned about the state of low-income marriages. Over the past 15 years, these concerns have fueled government efforts to strengthen marriages through federally funded marital enhancement programs targeting primarily low-income couples (Ooms, Bouchet, & Parke, 2004). Guiding the design of these programs are assumptions about why the institution of marriage appears so fragile in this population. For example, public education programs touting the benefits of marriage assume that there is something wrong with how low-income groups view marriage: that they lack traditional, family-oriented values or that their standards for relationships are unrealistic. Programs offering training in relationship skills imply that lowincome couples have more difficulty managing relationship issues than more affluent couples. Yet, despite the millions of dollars already spent on these programs, the support for the assumptions underlying them have received only minimal empirical scrutiny. Very little is known about how the most likely recipients of federal marriage promotion programs actually perceive and experience marriage, and what is known is based on limited populations (e.g., low-income single mothers) and surveys that do not specifically sample by race and income.
BACKGROUND AND GOALS OF THE STUDY
The goal of the current article is to evaluate current theories about how low-income populations view and experience marriage, with an eye toward how this evaluation might inform policy initiatives to strengthen marriage within low-income communities. Toward this end, the rest of this introduction is divided into three sections. In the first section, we review the existing literature on traditional family values, noting the limits of this literature for making statements about low-income populations. Next, we examine two alternative explanations for what is wrong with low-income marriages, noting that even if low-income populations value marriage as much as more affluent groups, they may have more unrealistic standards for marriage, or they may have more problems once they get married. We end this introduction with a third section in which we provide an overview of the current study, which was designed to evaluate each of these possibilities through analyses of survey data from a stratified random sample that oversampled low-income and non-White populations.
Do Lower Income Groups Value Marriage Less?
The assumption that low-income people do not value marriage as an institution has motivated federal programs to promote the value of marriage among the poor (Small, Harding, & Lamont, 2010). Is there any evidence that low-income people value marriage and traditional family arrangements less than those with higher incomes? For the nation as a whole, data from national opinion surveys reveal an increasing acceptance of nontraditional relationship behaviors but scant evidence that the value of marriage itself is on the decline (Axinn & Thornton, 2000; Thornton & YoungDeMarco, 2001). Indeed, marriage remains the norm for couples in the United States, and surveys have found that most single people express the desire to get married some day (Pew Research Center, 2010), yet prior analyses have not directly compared the values of lowincome people with those of people with higher incomes. On the contrary, as recent reviews have observed, the literature specifically focusing on low-income marriage is sparse by any standard (e.g., Fein, Burstein, Fein, & Lindberg, 2003; McLanahan, 2009; Seefeldt & Smock, 2004). Most research relating to families in low-income communities has focused not on marriage itself but on parenting and child outcomes or on the decision to get married (e.g, the Welfare, Children, and Families study; Fomby, Estación, & Moffitt, 2003; the Fragile Families and Child Wellbeing study; Reichman, Teitler, Garfinkel, & McLanahan, 2001). As such, the participants in this research have been limited to low-income parents (especially mothers) or unmarried cohabiting couples. The findings emerging from this research suggest that low-income mothers hold fairly traditional attitudes toward marriage - they think that marriage is a good thing and that people should get married before having children - but they are also accepting of cohabitation and divorce (Cherlin, Cross-Barnet, Burton, & Garrett-Peters, 2008). In addition, low-income, unmarried couples that have children together have positive attitudes toward marriage and expect to marry eventually (Edin, 2000; Edin & Refalas, 2005; Waller, 2001; Waller & McLanahan, 2005). Nevertheless, it remains unclear whether these findings generalize to people who have low incomes but are not parents or to individuals who are already married. Also, because comparison groups of higher income couples were not analyzed in this research, these studies did not address whether the values of low-income people differ significantly from those with higher incomes.
One important study that did address these issues demonstrated that women receiving Temporary Assistance to Needy Families (TANF) value marriage equally to those not on such assistance, but they do not value parenthood equally. Research conducted by Mauldon, London, Fein, Patterson, and Sommer (2005) examined data from National Survey of America's Families and the National Survey of Family Growth and compared women receiving TANF with those not receiving it. This study found that, compared with women who were not receiving TANF, those receiving it held less traditional values toward single parenthood: They were more likely to agree with the statement that ' ? single parent can bring up a child as well as a married couple," and they were less likely to agree with the statement that ' 'People who want children ought to get married."
Although these studies are evocative, they have focused on a subset of the low-income population (i.e., unmarried women, parents), did not specifically sample respondents by income and race, and lacked direct comparisons between lower and upper income populations. Understanding how the values of a broader sample of people with low incomes (i.e., men as well as women, nonparents as well as parents, and married couples as well as singles) relate to similar populations of higher income people would give scholars and policymakers a fuller picture of whether a lack of family values might play a role in the high divorce rates and low marriage rates among low-income populations. Furthermore, although single people with low incomes typically report a desire to get married, that desire may arise for a variety of reasons (e.g., because their parents want them to, to share health insurance) and may not mean that they value marriage as an institution. In fact, the attitudes of people with low incomes may be more subtle than just pro-marriage or antimarriage, revealing a desire for marriage but also an acceptance of the fact that marriage is not always possible and that other family forms (e.g., cohabitation, single parenthood) may be more attainable (Cherlin et al., 2008). In the current research we addressed these issues by surveying people sampled by income, race, and gender on their attitudes toward a variety of aspects of marriage, including its fundamental value.
Do Lower Income Groups Have Unrealistic Standards?
A separate explanation for the fragility of marriage among low-income populations is that their economic and romantic standards for relationships are too high (Cherlin, 2004; Gibson-Davis, Edin, & McLanahan, 2005; The National Marriage Project, 2010). According to this perspective, low-income people are likely to embrace a modern, soul mate model of marriage that prioritizes the ability of marriage to provide emotional fulfillment over the social functions of marriage (e.g., as an economic unit, as an environment for raising children). These romantic standards can interfere with successful relationships in low-income communities to the extent that individuals in these communities "do not believe their romantic relationships or marriages meet society's new bar for a capstone marriage" (The National Marriage Project, 2010, pp. 39-40). The key issue here is not whether the poor want to have good marriages, but whether their standards for relationships are higher than those with higher incomes. If so, then these higher standards may keep low-income people from marrying in the first place and, if they do marry, may lead to greater disappointment with their marriages relative to higher income couples.
Evidence relevant to this argument has mainly come from qualitative research among lowincome unwed mothers and cohabiting couples (e.g., Edin, 2000; Edin & Refalas, 2005; Smock, Manning, & Porter, 2005). In general, this work highlights the role of economic rather than romantic standards operating as barriers to marriage among the poor. For example, lowincome single mothers (Edin; Edin & Refalas) and cohabiting couples (Gibson-Davis et al., 2005; Smock et al.) report strong desires to get married, but they also want to make sure that they have enough money and/or that their partner has a good job before they will marry. Such findings do not support a view that low-income individuals harbor especially romantic notions about marriage, but without direct comparisons between the standards of low-income and more affluent groups, it remains unclear whether lowincome couples actually have higher economic standards for marriage than do those with higher incomes and whether these standards vary by gender and race.
Do Lower Income Groups Have More Relationship Problems?
Whether or not low-income groups can be distinguished by their attitudes and standards, a final possibility discussed in support of current policies is that low-income groups have more difficulty managing their relationships than higher income groups and so would benefit from training in relationship skills that more affluent couples presumably possess already. It is clear that low-income marriages do experience worse outcomes: Rates of divorce are nearly twice as high for women who live in low-income neighborhoods as compared with those in highincome neighborhoods (Bramlett & Mosher, 2002; Raley & Bumpass, 2003), but it is unclear whether this difference in outcomes emerges from problems with relationship processes, such as poor communication, or from factors that are external to the relationship (e.g., substance abuse, job loss; see Rarney & Bradbury, 1995, for a summary of this distinction).
Previous research has shown that lowincome couples experience more financial strain (Cutrona et al., 2003) and have more problems with substance abuse (Costello, Compton, Reeler, & Angold, 2003; Cutrona et al., 2005), than do higher income couples. Furthermore, research comparing self-reported reasons for divorce across individuals at different levels of income has found that those with lower incomes were less likely than those with higher incomes to list relationship -centered problems (e.g., incompatibility, personality differences) as a cause of their divorce (Amato & Previti, 2003). These results suggest that, although low-income couples may well face different problems than more affluent couples, those problems may not be easily addressed by improved relationship skills such as communication. Strong conclusions are at this point premature, though, because some of the prior research was retrospective (i.e., respondents were interviewed after their divorce), and the length of time since the divorce varied between respondents (Amato & Previti). Thus, it is possible that the problems respondents reported after the divorce were different from the ones they actually had during the relationship or that their memories of the problems had changed over time since the divorce. In the current research we asked respondents about problems experienced in their current relationships.
Overview of the Current Study
The current study was designed to address each of these alternative theories about the challenges facing low-income marriages. Toward this goal, we solicited a stratified random sample of individuals residing in Florida plus smaller random samples from California, Texas, and New York to participate in a survey that assessed respondents' values, relationship standards, and problems experienced in their current romantic relationships. Because federal programs specifically target recipients of TANF, we also gathered a separate sample of Florida residents receiving TANF. Thus, we used a comprehensive sampling strategy to compare how people with higher and lower incomes view marriage and their experiences with relationship problems. This strategy allowed us to describe similarities and differences across income groups while controlling for potential confounds. It also allowed us to examine interactions among race, income, and gender to ensure that conclusions drawn about income groups were consistent among subgroups within income category. To our knowledge, such a comprehensive look at income, race, and gender has not been previously published.
On the basis of past research, we predicted that people with lower incomes, including those receiving TANF, would hold values toward marriage very similar to values held by those with higher incomes, but that people with lower incomes would also be more accepting of single-parent households than would people with higher incomes. Given the economic stress that low-income populations face, we expected that economic standards for a successful relationship would be more salient for low-income people than for higher income people and that this standard would be similar for low-income men and women across different races. Furthermore, for similar reasons, we expected people with low incomes to hold lower romantic standards for marriage than would those with higher incomes. To be clear, we expected that people with low incomes would want to have fulfilling, romantic relationships in marriage but that these standards would be less salient to them than they would be for those with higher incomes. Finally, we had no a priori reason to predict that lower income couples would have more problems with basic relationship processes than would higher income couples, so we did not expect that these problems would differ by income although, because previous research has demonstrated that low-income populations experience more problems with substance abuse and with financial security, we expected that lowincome couples would report more problems with these issues than would higher income couples. This pattern of findings would contradict many of the assumptions underlying federal and state programs to strengthen marriage among low-income populations.
In all analyses, we controlled for marital status, age, gender, state, and race. We expected that being married would affect respondents' attitudes toward and standards for marriage, and we thought that they would have different levels of relationship problems. Similarly, we expected that respondents might differ in their attitudes and experiences by age and state of residence (e.g., people in some states have more conservative values than those in other states). We controlled for gender and race for similar reasons, but we also examined whether the attitudes or experiences of people differed not only by income but also by race and gender. Examining the intersection of income, race, and gender provides a perspective on how the context of multiple group memberships combines to affect disadvantaged populations (Choo & Ferree, 2010; Cole, 2009). Therefore, we discuss significant interactions between income and gender or race that emerged from the analysis only when they substantially qualified the overall effect of income.
Individuals in this study were recruited in the summer and fall of 2003 as part of the Florida Family Formation Survey (see http://www.healthymarriageinfo.org/resour ce-detail/index.aspx?rid=2512). The design of the survey included stratified random sampling of the state of Florida, with oversamples of Blacks, Latinos, and low-income residents (i.e., household income of less than 200% of the Federal Poverty Level [FPL]). Furthermore, a random sample of Florida residents receiving TANF assistance was selected from the complete roster of TANF recipients in the state. We also collected data via random-digit dialing in three states from different regions of the country, but with populations comparable to Florida's in terms of density and diversity: California, Texas, and New York.
Following the guidelines developed by the American Association for Public Opinion Research (201 1), we calculated the response rate for the survey using the relatively conservative RR3 method. This method defines an estimation factor that allows the researcher to assume a percentage of "unknown households" to be counted as ineligible. Thus, by this method, the response rate represents the number of completed interviews (not counting partial completes) divided by the number of eligible telephone numbers selected for the sample, including an estimate of what proportion of unknown cases were likely to have been eligible. As a result of using this procedure, the estimates may be more conservative than those reported by other comparable studies of these issues. Using this method, the response rate was 22% in Florida, 19% in California, 16% in New York, and 21% in Texas. Response rates were highest in the TANF sample (28%). In contrast to the response rate, the cooperation rate is the number of completed interviews divided by the number of households successfully contacted. The cooperation rate does not control for the efficiency of the telephone sampling, and so it will be higher than the response rate. In this survey, the cooperation rates were 34% for the Florida survey and the TANF sample, 26% in California, 23% in New York, and 29% in Texas.
Low survey response rates have traditionally been a source of concern for researchers, but in the era of cell phones and Caller ID, response rates to telephone surveys have been steadily declining over the past decade (Curtin, Presser, & Singer, 2005; Holbrook, Krosnick, & Pfent, 2007). The response rates for the current study are comparable to those of other phone survey research studies that used similar sampling designs and methods for calculating response and cooperation rates (e.g., Pew Research Center, 2010) and are higher than response rates reported for other modes of survey research (e.g., web-based panels; Dennis, Chatt, Li, Motta-Stanko, & Pulliam, 2005). Nevertheless, surveys with lower response rates typically yield results similar to those of surveys with higher response rates (e.g., Curtin, Presser, & Singer, 2000; Keeter, Miller, Kohut, Groves, & Presser, 2000). Indeed, researchers have argued that there is no clear criterion for the level of response rate that will yield nonbiased data and that pressing for higher response rates can actually lead to an increase rather than a decrease in survey bias (Groves, 2006). Because there is no reason to think that nonrespondents systematically differed from respondents on family values, relationship standards, or relationship problems, it is unlikely that the data described here were affected by nonresponse bias.
A total of 6,012 people responded to the survey: Four thousand, five hundred eight were sampled from Florida, 500 were from California, 502 from New York, and 502 from Texas. Among respondents, 65.5% were female, 14.3% were Black, 60.6% were White, 18.7% were Latino/Hispanic (non-White or Black), and 52.3% were married. Furthermore, on the basis of self-reports of total household income, 29.4% of respondents were categorized as low income (household incomes below 200% FPL), 26.1% were moderate income (incomes between 200% and 400% FPL), and 34.7% were high income (incomes above 400% FPL). An additional 9.8% of respondents were enrolled in TANF. The average age of respondents was 45.82 years (SD = 16.57). Respondents were not compensated for their participation.
Participants were surveyed via telephone interviews about their experiences, beliefs, and attitudes regarding intimate relationships. The complete interview contained roughly 200 questions, although no participant received the entire set of questions, because many items were administered only to specific groups (i.e., parents, unmarried individuals). The average length of an interview was 27 minutes.
Traditional values. We assessed the extent to which respondents held traditional values toward marriage and family using 13 items adapted from previous research (Amato, 1988). These items were designed to capture a wide range of traditional values, including attitudes toward marriage itself (e.g., ' ? happy healthy marriage is one of the most important things in life"), attitudes toward divorce (e.g., ' 'Divorce can be a reasonable solution to an unhappy marriage"), attitudes toward traditional women's and men's family roles (e.g., "The important decisions in the family should be made by the man of the house"), attitudes toward cohabitation (e.g., "It is okay for couples who are not married to live together"), and attitudes toward single parenthood (e.g., "People who have children together ought to be married"). See Table 1 for a list of all items. Respondents rated how much they agreed with each item using a 5 -point scale that ranged from strongly disagree to strongly agree. Each item was analyzed as a single-item measure.
Marriage standards. The standards that respondents held for marriage were assessed by asking participants to rate 1 1 items as to ' 'whether you think each item is . . . important for a successful marriage" (see Table 2 for scale items). Items were rated on a 3 -point scale on which 0 = ' 'not important," 1 = "somewhat important," and 2 = "very important." Each item was analyzed as a single-item measure.
Relationship problems. Respondents who were currently in a relationship were asked to rate how much each of 10 topics was a "source of difficulty" in their current romantic relationship (see Table 3 for scale items). Responses were made on a 4-point scale that ranged from 1 (nota problem at all) to 4 (very serious problem) Each item was analyzed as a single-item measure.
Of the 6,012 respondents, 694 (1 1 .5%) had missing data on income, 400 (6.7%) had missing data on age, 26 (0.4%) had missing data on marital status, and 33 (0.6%) had missing data on race, for a total of 1,153 (19.2%) respondents with missing data. We imputed missing data using the MI procedure in SAS 9.2. Ten data sets were generated, and the imputation was informed by the variables in the regression analyses plus auxiliary variables that were highly correlated with the primary variables of interest (e.g., the respondent's level of education and employment status).
We analyzed the results using hierarchical regression in SAS 9.2, and we consolidated estimates from the 10 data sets using the MIANALYZE procedure. All categorical variables were dummy coded, and all continuous variables were centered prior to analysis. We assessed the average main effect of income group across imputed data sets using the SAS MIANALYZE procedure. Significant main effects of income were followed up with tests of simple effects between income categories. In a separate analysis, we also examined all interactions among gender, race, and income using hierarchical regression; significant interactions are reported here only when they qualify the main effects of income, but the complete analyses are available on request. Data were weighted using the product of expansion weights and a poststratification adjustment to ensure that the results were representative of each state surveyed. All statistical tests were two-tailed. It is important to note that an analysis of the raw data without imputed values yielded the same conclusions discussed below.
We first examined whether people with lower incomes had less traditional values than people with higher incomes. As shown in Table 1, with some notable exceptions, low-income respondents had similar or more traditional values than high-income respondents on most value items, and they were similar to moderateincome respondents on most items. No differences between income groups emerged for their ascribed value of marriage as an institution. All respondents agreed that ' ? happy, healthy marriage is one of the most important things in life" and that "People who have children together ought to be married." Most respondents agreed that ' 'Children do better when their parents are married." Although high-income respondents agreed less with this statement than did moderate-income respondents, neither group differed from lower income respondents.
Relative to high-income respondents, those with low incomes were significantly more traditional in their views toward divorce: They were less likely to agree that divorce was a "reasonable solution to an unhappy marriage," they agreed that parents who no longer love one another should stay married for the sake of the children, and they were more likely to think that divorce reflects badly on a couple. One exception is that TANF recipients were more likely to agree with the statement that ' 'When parents are arguing a lot, it is better for the children if they divorce" than were moderate- and lowincome respondents. High-income respondents were similar to low-income respondents on this question.
Lower income respondents also held more traditional values for women's and men's family roles within relationships than did highincome respondents. They were more likely to think that the man of the house should make important decisions, and they were more likely to think that it is "better for a family if the man earns a living and the woman takes care of the home" than were higher income respondents, although the results for traditional values regarding cohabitation were mixed. Lower income respondents were less likely than those with high incomes to agree that "It is okay for couples who are not married to live together," but they were more likely than high- and moderate-income respondents to think that mothers who are living alone ' 'can bring up their children as well as married couples." This latter result mirrors the findings of Mauldon et al. 's (2005) research with mothers receiving TANF. Still, lower income respondents held more traditional values toward premarital sex than did those with high incomes. When asked whether they agreed that "When a couple is committed to each other, it makes no difference whether they are married or just living together,' ' it was the moderate-income respondents who held the most traditional values: They disagreed with this idea more than any of the other income groups, who did not differ from each other. Thus, relative to high-income respondents, those with low incomes held traditional views of the role of marriage in their lives, in society, and as a fundamental basis of parenting.
Several significant interactions emerged from the analysis of family values (i.e., 5 of the 39 possible interactions between gender or race and income were statistically significant with a Bonferroni-corrected alpha of p < .0013; details are available on request), but none of the interactions revealed income effects that contradicted the pattern of results just described. For example, a significant Race x Income interaction emerged for the value "Couples should wait to have sex until they are married," F (6, 5993) = 5.10, p < .001. Examination of the simple effect of income for each racial group revealed that this effect was significant for White (p < .001) and Latino (p < .001) respondents, but not for Black respondents (p = .61). Nonetheless, the pattern of means for Black respondents mirrored the overall pattern (i.e., high-income Black respondents rated this value lower than respondents in the low- and moderate-income groups). Although this difference is interesting, it does not qualify the general conclusion that low- and moderateincome respondents held more traditional views for this value than did high-income respondents.
With our next set of analyses, we examined whether people with lower incomes had higher standards for marriage than those with higher incomes. As illustrated in Table 2, except where employment is concerned, lowincome respondents actually had similar or lower standards for marriage than did those with higher incomes. Controlling for marital status, age, gender, state, and race, low-income respondents and/or TANF recipients tended to perceive emotional support and intimacy in marriage as less important than did higher income respondents; specifically, lower income respondents rated ' 'having the same values and beliefs," "having good sex," "supporting each other through difficult times," and "being able to communicate effectively" as less important to successful marriage than did higher income respondents.
Ratings of standards concerning income resulted in a different pattern of responses. Lowincome respondents and/or TANF recipients held higher standards than higher income respondents for "husband having a steady job" and "wife having a steady job." Lower income respondents also rated members of a couple "being of the same race or ethnic group" as more important than did higher income respondents. Thus, as shown in previous, qualitative research (e.g., Edin, 2000; Edin & Kefalas, 2005), low-income respondents reported higher standards for the economic aspects of marriage.
Finally, all respondents held equal standards for "spending time together," "having savings you can draw from," "understanding each other's hopes and dreams," and "having a family that supports you,' ' regardless of income. It is important to note that, where differences did emerge between income groups, the effect sizes of these differences were very small (i.e., R2s < .004). The one exception was for "husband having a steady job." The effect size for income differences on that standard was a respectable R2 = .011. As with our analysis of values, significant interactions emerged from the analysis of marriage standards (i.e., 2 of the 33 possible interactions between gender or race and income were statistically significant with a Bonferroni-corrected alpha of p < .0015), but again, neither of the interactions revealed income effects that contradicted the pattern of results just described.
With our last set of analyses we examined whether low-income couples experienced different problems in their relationships than did higher income respondents. We were especially interested in problems that might result from a lack of relationship skills, because those skills have been targeted by marital intervention programs for the poor. This set of analyses included only those respondents who were in a romantic relationship at the time of the survey (n - 4,258). As shown in Table 3, many of the problems that low-income couples faced were similar to those faced by higher income couples. No significant differences emerged between income groups for ratings of problems related to the following: being a parent, communication, sex, household chores, or each other's parents. Low-income respondents reported more difficulty "finding time to spend together" than did TANF recipients, but no other significant differences emerged.
As the data in Table 3 reveal, significant differences between income groups emerged for problems with money, drinking or drug use, being faithful, and friends. Controlling for marital status, age, gender, state, and race, lower income groups experienced more of these problems than did higher income respondents. Thus, there is no evidence that lower income couples experienced more problems that might reflect a lack of relationship skills. There was, however, some evidence that low-income marriages face particular problems with money, substance abuse, infidelity, and friends - problems that are not targeted by most of the current federal marital education programs.
As with the prior set of analyses, several significant interactions emerged from the analysis of relationship problems (i.e., 6 of the 30 possible interactions between gender or race and income were statistically significant using a Bonferronicorrected alpha of p < .0017). For the most part, these interactions do not change the basic conclusions drawn from the main effects of income just described; that is, most low-income groups reported problems similar to those of higher income groups on relationship process problems and more problems with economic issues. The exceptions to this trend are that lowincome Black men and low-income Latino men reported more problems with drinking/drug use and being faithful than did any other Race x Gender x Income Subgroup combination (both three-way interaction ps < .0017; AR2s were .006 and .005, respectively).
Declining marriage rates, increased divorce, and the prevalence of one-parent families among lower income populations have been taken as evidence for a problem with the culture of marriage among the poor (Popenoe, 2001), yet to date the evidence to support these characterizations of low-income populations has been weak or absent. Through direct comparisons of survey data from representative samples of respondents from a wide range of income levels, the current study demonstrated that, whatever the reasons for the increased vulnerability of low-income marriages, lowincome populations do not value marriage less than those with higher incomes. Previous research has confirmed that low-income couples want to get married (Cherlin et al., 2008; Edin, 2000; Edin & Refalas, 2005; Waller, 2001; Waller & McLanahan, 2005), but the current research revealed that respondents from all income groups equally value the role of marriage in their lives, in society, and as a fundamental basis for parenting. Furthermore, the current study found that low-income populations do not endorse higher romantic standards for marriage; neither do they report more problems with relationship -centered behaviors, relative to higher income couples. Thus, the current research strongly suggests that the culture of marriage is just as strong among low-income populations as it is among those with higher incomes.
Why might respondents with lower incomes hold more traditional values than other groups, in light of widespread beliefs to the contrary? One possibility is that this group, more than other segments of society, has more exposure to the consequences of family dissolution because it is more prevalent among people with low incomes than it is among those with higher incomes (Bramlett & Mosher, 2002). In other words, as a group they are more likely to have struggled with finding a marriage partner or with divorce (or have seen their friends and family members struggle with these issues). Seeing one's own or a family member's children deal with the negative effects of divorce may support the view that parents should stay married for the sake of their children, strengthening positive attitudes toward marriage and traditional family life rather than weakening them (Cahn & Carbone, 2010). The current findings, although cross-sectional, thus highlight the possibility that endorsing traditional family values may be a response to their experiences with marriage, rather than a cause.
As suggested by previous research (Cherlin et al., 2008), though, the values of low-income respondents were more nuanced than simply "pro-marriage." Low-income respondents had mixed attitudes toward cohabitation, and it is possible that this reflected a divergence between their experiences and their ideals. Low-income participants clearly thought that two-parent households were important for a child's wellbeing, but they also were more likely to agree that "Most mothers living alone can bring up their children as well as married couples" than were higher income groups. In addition, relative to moderate-income respondents, TANF recipients were more likely to agree that "When parents are arguing, it is better for the children if they divorce." Although at first glance these two sets of values may seem to conflict, it is possible that low-income people idealize marriage but recognize that the marriages they desire are not always possible and so accept other family forms as inevitable. Thus, for judgments of the appropriateness of some behaviors, practicality may take precedence over idealism, and lowincome populations may have more experience with effective single-parent households than do higher income populations.
Of course, it also is possible that lowincome respondents were simply echoing current cultural norms that value marriage while also accepting alternative lifestyles (Axinn & Thornton, 2000) and that their reported values did not reflect their actual feelings toward marriage and family life. We think that this alternative explanation is unlikely for two reasons. First, if low-income respondents were just reporting what they believed were socially desirable attitudes, their reports should not have differed from the other income groups - who, we presume, would have also been reporting attitudes that conformed to the national norm. Second, low-income respondents reported more conservative values for items that seemingly reflect the modern norm for accepting nontraditional behaviors (e.g., waiting until marriage to have sex, couples cohabiting rather than marrying), suggesting that, compared with people with higher incomes, the values of people with low incomes are bucking the trends toward more acceptance of nontraditional relationship behaviors.
The theory that low-income populations endorse especially high or unrealistically romantic standards for their intimate relationships received little support in the current study: Lowincome respondents' standards for romantic and emotional aspects of marriage were similar to or lower than those of higher income respondents. Lower income couples might think that the romantic and emotional components of marriage are important (Cherlin, 2004), but their romantic standards for marriage do not seem to be any more unrealistic than those of people with higher incomes. Lower income groups did report higher standards for the economic aspects of marriage, confirming with survey data what has previously been suggested by qualitative research on lowincome mothers (Edin, 2000; Edin & Kefalas, 2005). The relative prominence of employment and financial security in low-income respondents' views of successful marriage may stem from the fact that these are aspects of marriage that they cannot take for granted. Higher income respondents, perhaps assuming they and their partner will have good jobs (or that one partner would have an income high enough to support both spouses), may not see these issues as central to their relationships. It is worth emphasizing that, in highlighting the importance of finances and employment to successful marriages, lower income respondents were echoing the results of empirical research indicating that these factors are indeed important for successful marriages (Bramlett & Mosher, 2002; Raley & Bumpass, 2003). Thus, far from being unrealistic, the standards of lower income respondents reflect an accurate understanding of the important predictors of successful marriage in their communities. If this idea is borne out in future research, it implies that accuracy in predicting marital outcomes could actually keep some lowincome couples from marrying in the first place. Indeed, although researchers have devoted much attention to trying to accurately predict marital outcomes (e.g., Gottman, 1994), less attention has been paid to the impact of the couples' own predictions of potential marital success on their relationship behaviors.
With respect to difficulties managing intimacy, the current research demonstrated that, for most specific relationship problems, lower income partners are no more likely to struggle than are higher income partners. The relationship problems that low-income respondents do experience as more severe than higher income respondents included problems that are generally more common among low-income populations (e.g., problems with money, substance abuse) but also included problems with fidelity and friends. It is important to note that, as in previous research (Amato & Previti, 2003; Karney & Bradbury, 1995), these problems are largely related to external stressors (i.e., financial problems, friends) and problem behaviors (i.e., substance abuse) rather than relationship -centered problems, raising questions about the appropriateness of interventions targeting low-income couples that focus primarily on interpersonal processes (e.g., communication and problem solving). To be sensitive to the unique challenges that may be associated with higher vulnerability in this population, interventions may need to expand their focus to how couples negotiate the demands and temptations of their circumstances. Some state programs have already instituted this type of comprehensive intervention program to improve marriage, incorporating drug and alcohol treatment as well as job training into their programs (Ooms et al., 2004). The current research suggests that this type of intervention should be the norm rather than the exception. Similarly, programs that promote economic stability in low-income communities (e.g., programs to increase steady employment or assist with debt relief and housing) may have significant effects on marital outcomes in those communities, even if those programs never target marriages or relationships directly. Whatever bolsters the financial prospects of lowincome couples may remove barriers to marriage and/or forestall divorce for couples struggling with financial problems (Levin-Epstein, Ooms, Parke, Roberts, & Turetsky, 2002). Although the effect of financial assistance on marriage and divorce rates is a source of controversy in the literature (Gennetian & Knox, 2003), the current research suggests that these strategies would help relieve stress on low-income relationships, allowing low-income couples to better follow through with their desires for stable, healthy marriages.
Finally, although this study did not focus on differences between men and women of different racial backgrounds and income levels, some important differences did emerge from our analysis; specifically, low-income Black and Latino men reported higher levels of problems with drinking/drug use than did any other race, gender, and income subgroup. This suggests that these subgroups would especially benefit from an intervention to treat individuals with alcohol or drug dependency, which may have the concomitant effect of improving their marriages. In addition, low-income Black and Latino men reported more relationship problems with being faithful than did any other race, gender, and income subgroup. Although it is difficult to explain why Black and Latino men reported more problems with these issues than did other groups, these findings do point to the difficulties that Black and Latino couples face in establishing and maintaining marital relationships. It is possible that Black and Latino men experience more stressors from outside the marriage that then affect relationship processes (e.g., from discrimination; Lincoln & Chae, 2010; Murry, Brown, Brody, Cutrona, & Simons, 2001) and that these stressors are exacerbated by financial stress experienced by low-income couples. If that is true, then one would expect that racial differences would exist within income groups as well. Indeed, some research has revealed racial differences in marital attitudes among low-income mothers (Edin, 2000), but more needs to be done to uncover the scope and magnitude of these differences.
Strengths and Limitations
A number of strengths in the methodology and design of this study enhance our confidence in the results. First, respondents were sampled by income and race so that substantial representative samples of these groups could be compared. Second, participants were asked about a wide range of values, marriage standards, and relationship problems that had not been assessed in previous large-scale surveys. Third, as opposed to previous research that looked only at lowincome populations, we were able to make direct comparisons between income groups while statistically controlling for potential confounding variables (e.g., age, marital status).
Despite these strengths, this study is also limited in several ways. First, all of the data examined here were obtained through selfreports; therefore, it is possible that people's responses did not match their actual values, standards, and problems. Of course, in order to compromise our conclusions, this tendency would have to have been more pronounced among lower income respondents than among higher income respondents, and we know of no evidence to support this possibility. Second, respondents provided data at a single assessment (i.e., all of these data are cross-sectional), so we cannot make conclusions about the causal direction of our effects.
In addition, although the states sampled in the current study were selected to be broadly representative of different regions of the United States (South, Northeast, Midwest, and West), it is possible that residents of other states would respond differently to the survey. As noted earlier, the sampling strategy used in the current study was unique in that we specifically sampled by race/ethnicity and income, and we broadly focused on men and women, with and without children, who may or may not have been in a relationship at the time. Because other research on the state of marriage among low-income populations has focused on specific subpopulations (e.g., unmarried couples with children, divorced couples), it is difficult to assess the generalizability of our findings by directly comparing our results with those obtained in other studies (e.g., the Fragile Families study). Nonetheless, our results generally replicate the findings of these other studies while also providing important extensions, suggesting that our results are generalizable to populations living in states not sampled in our study.
Finally, although the questions assessing traditional values, marriage standards, and relationship problems were all taken from previous research on these issues, the items here did not exhaust the possible universe of items. It remains possible that a different set of questions would have led to different results.
So, what's wrong (or not wrong) with lowincome marriages? This is an important question for American policymakers and for American society. The answer is not simple. Low-income couples marry less, divorce more, and have children out of wedlock more than higher income families, yet they hold largely traditional family values, do not have unrealistic standards for marriage, and do not experience more problems with basic relationship processes than do higher income couples. How are we to reconcile these two portraits of lowincome marriages? The current research offers a glimpse at a possible explanation: People with lower incomes experience relationship vulnerabilities that fall largely outside the realm of the relationship itself - and outside the realm of most interventions that aim to improve marriages. Thus, in order to be successful, efforts to save low-income marriages must directly confront the economic and social realities these couples face.
The preparation of this article was supported by a research grant (R01HD061366) from the National Institute of Child Health and Human Development awarded to Benjamin R. Karney. The data set described in this article was collected through a contract between Benjamin R. Karney and the Florida Department of Children and Families. We gratefully acknowledge the crucial assistance of Cynthia Garvan, Wei Hou, and Michael Thomas in preparing earlier versions of these results. We also thank Jeff Johnson, currently Director for Family and Community Services at the Texas Health and Human Services Commission, for his stewardship of the original survey.
Amato, P. R. (1988). Parental divorce and attitudes toward marriage and family life. Journal of Marriage and the Family, 50, 453 - 46 1.
Amato, P. R., & Previti, D. (2003). People's reasons for divorcing: Gender, social class, the life course, and adjustment. Journal of Family Issues, 24, 602-626.
American Association for Public Opinion Research. (2011). Standard definitions: Final dispositions of case codes and outcome rates for surveys. Retrieved from http://www.aapor.org/standard_ definitions2.htm
Axinn,W. G., & Thornton, A. (2000). The transformation in the meaning of marriage. In L. J. Waite(Ed.), The ties that bind (pp. 147-165). Hawthorne, NY: Aldine de Gruyter.
Bramlett, M. D., & Mosher, W. D. (2002). Cohabitation, marriage, divorce, and remarriage in the United States. Retrieved from http://www.cdc.gov/ nchs/data/series/sr_23/sr23_022.pdf
Cahn, N., & Carbone, J. (2010). Red families v. blue families: Legal polarization and the creation of culture. New York: Oxford University Press.
Cherlin, A. J. (2004). The deinstitutionalization of American marriage. Journal of Marriage and Family, 66, 848 -861.
Cherlin, A. J., Cross-Barnet, C, Burton, L. M., & Garrett-Peters, R. (2008). Promises they can keep: Low-income women's attitudes toward motherhood, marriage, and divorce. Journal of Marriage and Family, 70,919-933.
Choo.H. Y., & Ferree, M. M. (2010). Practicing intersectionality in sociological research: A critical analysis of inclusions, interactions, and institutions in the study of inequalities. Sociological Theory, 28,129-149.
Cole, E. R. (2009). Intersectionality and research in psychology. American Psychologist, 64, 170-180.
Costello, E. J., Compton, S. N., Keeler, G, & Angold, A. (2003). Relationships between poverty and psychopathology. Journal of the American Medical Association, 290, 2023-2029.
Curtin,R., Presser, S., & Singer, E. (2000). The effects of response rate changes on the Index of Consumer Sentiment. Public Opinion Quarterly, 64,413-428.
Curtin, R., Presser, S., & Singer, E. (2005). Changes in telephone survey nonresponse over the past quarter century. Public Opinion Quarterly, 69, 87-98.
Cutrona, C. E., Russell, D. W., Abraham, W. T., Gardner, K. A., Melby, J. N., Bryant, C, & Conger, R. D. (2003). Neighborhood context and financial strain as predictors of marital interaction and marital quality in African American couples. Personal Relationships, 10, 389-409.
Cutrona, C. E., Russell, D. W., Brown, P. A., Clark, L. ?., Hessling, R. M., & Gardner, ?. A. (2005). Neighborhood context, personality, and stressful life events as predictors of depression among African American women. Journal of Abnormal Psychology, 114, 3-15.
Dennis, J. M., Chart, C, Li, R., Motta-Stanko, ?., & Pulliam, P. (2005). Data collection mode effects controlling for sample origins in a panel survey: Telephone versus Internet. Retrieved from http://www.knowledgenetworks.com/ganp/ rtimode.html
Edin, K. (2000). What do low-income single mothers say about marriage? Social Problems, 47, 112-133.
Edin, K., & Kefalas, M. (2005). Promises I can keep: Whypoorwomenputmotherhoodbeforemarriage. Los Angeles: University of California Press.
Fein, D. J., Burstein, N. R., Fein, G. G., & Lindberg, L. D. (2003). The determinants of marriage and cohabitation among disadvantaged Americans: Research findings and needs. Retrieved from http://www.abtassociates.com/reports/ES-lit rev_abt.pdf
Fomby, P., Estación, A., & Moffitt, R. (2003). Welfare, families, and children: Characteristics of the three-city sample: Wave 1 and Wave 2. Retrieved from http ://web jhu.edu/threecitystudy/images/ CharacteristicsReport.pdf
Gennetian, L., & Knox, V. (2003). Staying single: The effects of welfare reform policies on marriage and cohabitation. Retrieved from http://www.mdrc.org/ng_working_papers/ ng_wkgppr_l 3/ng_l 3_wkgppr.pdf
Gibson-Davis, C. M., Edin, K., & McLanahan, S. (2005). High hopes but even higher expectations: The retreat from marriage among low-income couples. Journal of Marriage and Family, 67, 1301-1312.
Gottman, J. M. (1994). What predicts divorce? Hillsdale, NT: Erlbaum.
Groves, R. M. (2006). Nonresponse rates and nonresponse bias in household surveys. Public Opinion Quarterly, 70, 646-675.
Holbrook, A. L., Krosnick, J. ?., & Pfent, A. (2007). The causes and consequences of response rates in surveys by the news media and government contractor survey research firms. In J. M. Lepkowski, C. Tucker, J. M. Brick, E. D. de Leeuw, L. Japec, P. J. Lavrakas, M. W. Link, & R. L. Sangster (Eds.), Advances in telephone survey methodology (pp. 499-528). Hoboken, NJ: Wiley.
Karney, B. R., & Bradbury, T. N. (1995). The longitudinal course of marital quality and stability: A review of theory, method, and research. Psychological Bulletin, 118, 3-34.
Keeter, S., Miller, C, Kohut, ?., Groves, R. M., & Presser, S. (2000). Consequences of reducing nonresponse in a national telephone survey. Public Opinion Quarterly, 64, 125-148.
Levin-Epstein, J., Ooms, T., Parke, M., Roberts, P., & Turetsky, V. (2002). Spending too much, accomplishing too little: An analysis of the family formation provisions of HR 4737 and recommendations for change. Retrieved from http ://www.clasp .org/admin/site/publications _archive/files/0098.pdf
Lincoln, K. D., & Chae, D. H. (2010). Stress, marital satisfaction, and psychological distress among African Americans. Journal of Family Issues, 31, 1081-1105.
Mauldon, J. G, London, R. ?., Fein, D. J., Patterson, R., & Sommer, ?. (2005). Attitudes of welfare recipients toward marriage and childbearing. Population Research and Policy Review, 23, 595-640.
McLanahan, S. (2009). Fragile families and the reproduction of poverty. Annals of the American Academy of Political and Social Science, 621, 111-131.
Murry, V. M. B., Brown, P. ?., Brody, G. H, Cutrona, C. E., & Simons, R. L. (2001). Racial discrimination as a moderator of the links among stress, maternal psychological functioning, and family relationships. Journal of Marriage and Family, 63,915-926.
The National Marriage Project. (2010). When marriage disappears: The new middle America. Retrieved from http://stateofourunions.org/2010/ SOOU2010.pdf
Ooms, T., Bouchet, S., & Parke, M. (2004). Beyond marriage licenses: Efforts in states to strengthen marriage and two-parent families. A state-by-state snapshot. Retrieved from http://www.aecf.org/ upload/publicationfiles/beyond_marr.pdf
Pew Research Center. (2010). The decline of marriage and rise of new families. Retrieved from http://pewsocialtrends.Org/2010/l 1/1 8/thedecline-of-marriage-and-rise-of-new-families/
Popenoe, D. (2001, May 22). Marriage decline in America. Testimony before the Subcommittee on Human Resources Committee on Ways and Means, U.S. House of Representatives. Washington, DC.
Raley, K., & Bumpass, L. (2003). The topography of the divorce plateau: Levels and trends in union stability in the United States after 1980. Demographic Research, 8, 245-259.
Reichman, N, Teitler, J. O., Garfinkel, I., & McLanahan, S. (2001). The Fragile Families and Child Well-Being Study: Sample and design. Children and Youth Services Review, 23, 303 - 326.
Seefeldt, K. S., & Smock, P. J. (2004). Marriage on the public policy agenda: What do policy makers need to know from research? Retrieved from http ://deepblue.lib .umich.edu/handle/2027 . 42/58003
Small, M. L., Harding, D. J., & Lamont, M. (2010). Reconsidering culture and poverty. Annals of the American Academy of Political and Social Science, 629,6-21.
Smock, P. J., Manning, W. D., & Porter, M. (2005). "Everything's there except money": How money shapes decisions to marry among cohabitors. Journal of Marriage and Family, 67, 680-696.
Thornton, ?., & Young-DeMarco, L. (2001). Four decades of trends in attitudes toward family issues in the United States: The 1960s through the 1990s. Journal of Marriage and Family, 63, 1009-1037.
Waller, M. R. (2001). High hopes: Unwed parents' expectations about marriage. Children and Youth Services Review, 23, 457-484.
Waller, M. R., & McLanahan, S. S. (2005). "His" and "her" marriage expectations: Determinants and consequences. Journal of Marriage and Family, 67, 53-67.
THOMAS E. TRAIL AND BENJAMIN R. KARNEY University of California, Los Angeles
Department of Psychology, Franz Hall, Box 951563, University of California, Los Angeles, CA 90095 (firstname.lastname@example.org).