The Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) (1996), commonly referred to as welfare reform, reflects strong sentiments that mothers seeking public assistance should work whenever possible. With the introduction of the Temporary Assistance for Needy Families (TANF) program, mothers who do not meet limited exemption criteria must find jobs within two years of receiving benefits. Indeed, the 4.4 million families receiving TANF in August, 1996, dropped by more than 50% nationwide to 2.1 million in March, 2001. Many mothers found employment, with 75% of welfare leavers being employed some time in the year after exit (Acs and Loprest 2001).

Employment among unmarried mothers is important because of its potentially positive contributions over welfare receipt including higher incomes, lower poverty rates, and lower levels of material hardship (Danziger et al. 2000; Simmons et al. 2007a, b). Increasing employment rates in the 1990s also accounted for the largest share in the recent decline of child poverty rates (Lichter et al. 2005). Although employment does not dictate self-sufficiency or the ability to escape poverty (for example, see Bok and Simmons 2002), potential benefits from employment and stringent work requirements under TANF call into question what facilitates unmarried mothers’ employment.

Given employment’s increased importance under TANF, this study attempts to better understand how social support contributes to employment for unmarried mothers who no longer have unlimited access to a cash safety net. Social support is often conceptualized as an “intuitive sense of a broader phenomenon” (Brownell and Shumaker 1984, p. 5) leading to different operationalizations, depending on a study’s focus. Measuring social support is also complex because receipt often depends on level of need. Individuals who receive support may face additional hardships, creating a negative relationship between social support receipt and wellbeing (Cutrona 1986). In other words, measuring received support can introduce endogeneity when examining social support’s influence on employment outcomes. Alternatively, perceived support captures availability of support without requiring that the support be utilized. Although this eliminates the problem of endogeneity, perceived support could measure perception and personality, such as one’s locus of control, rather than the actual availability of support (Sarason et al.1990).

The Fragile Families and Child Wellbeing dataset used in this study documents the social, psychological, and economic lives of unmarried mothers and provides an opportunity to examine how perceived and received family supports, received community supports, and individual and family background supports contribute to employment in the post-welfare reform era for this primarily-disadvantaged group of women.Footnote 1 Accordingly, this study examines (1) the differences in employment rates among Hispanic, non-Hispanic Black, and non-Hispanic White unmarried mothers of young children after the passage of welfare reform,Footnote 2 (2) the influence of family, community, individual, and family background supports on employment status, and (3) how the relationship between supports and employment status differs by race/ethnicity.

Literature Review

Race/Ethnicity, Employment, and Social Support

Approximately 51% of never-married mothers of children older than one year and 34% of never-married mothers of infants (under one year) were employed full-time in 2002. Approximately 16% of both groups were employed part-time (Downs 2003). The recent rise in employment among unmarried mothers is well-documented, and the rate of increase varies by race/ethnicity (Cancian and Reed 2001). Employment increased most dramatically among Black unmarried mothers, from 53% in 1990 to 75% in 1999. The increase was less dramatic for Whites (from 71 to 79%) and Hispanics (from 50 to 59%) (Cancian and Reed 2001). Studies examining the influence of race/ethnicity on employment for different groups of women, however, lack consensus and call for additional study. Although Cancian and Reed (2001) document the highest rates of employment among Whites, Harknett’s (2006) examination of welfare-to-work unmarried mothers indicates that non-White mothers worked more quarters over a seven year period in the 1990s than Whites. Still, other evidence indicates that Hispanics, Blacks, and Whites work at similar rates (Holzer and LaLonde 2000; Yoon and Waite 1994). In analyses using the National Longitudinal Survey of Youth (NLSY) (1978–1993), race/ethnicity was not a factor in young women’s employment entrances and exits (Holzer and LaLonde 2000).

The relationship between race/ethnicity and social support also warrants further study. Several studies indicate that support networks may be stronger among Hispanics and Blacks than Whites (Hogan et al. 1990; Hao 1994; Keefe 1996). In an analysis of kin networks, childcare, and financial support using NLSY (1984) data, Blacks were more involved in support networks and had greater access to kin than Whites (Hogan et al. 1990). However, kin support may be declining among these groups (Brewster and Padavic 2002; Roschelle 1997). Analyses of nationally-representative data from 1977 to 1994 indicate that although Blacks wanted to provide support, financial and social constraints limited the material and in-kind support that they could give (Brewster and Padavic 2002). Hispanics’ and Blacks’ kinship support advantage over Whites may have turned into a kinship disadvantage.

In addition to explaining variation in social support and in employment, race and ethnicity may condition the relationship between supports and employment. In a NLSY analysis of employment, Black women without high school diplomas were less likely to enter the labor market than Black women with diplomas, while the reverse was true for White women (Hao 1994). Additional NLSY analyses indicate that high school diplomas played a lesser role in securing employment among Spanish speakers than English speakers (Taniguchi and Rosenfeld 2002). In terms of living arrangements, nationally representative data analyses show that cohabitation increased employment for Blacks mothers, but not for Whites (Park 2005).

Conceptual Framework

With complex relationships among race/ethnicity, social support, and employment, it is important to understand how these variables relate to one another. To predict employment among unmarried mothers, this study combines elements of both human capital theory, popular among economists, and social capital theory, popular among sociologists. Employment decisions are based upon the availability of individual, family, and community supports. Human capital theory posits that mothers with higher education levels and employment experience will be more likely to be employed than mothers with lower levels (Becker 1993). Social capital created through relationships with others can provide critical, often informal, resources that help mothers achieve employment, such as employed role models or a relative to provide child care (Coleman 1988).

Modeled after Edin and Harris’ (1999) framework, unmarried mothers’ employment is conceptualized as a decision-making process in which mothers decide how to combine various resources in order to fulfill roles as both mother and provider. The present framework expands social capital’s operationalization to include received community supports, such as rent assistance or job training because both individual and community resources are important in the employment calculus among low-income populations (DeBord et al. 2000; Monroe et al. 1999). In addition to acknowledging the influence of social and human capital on employment, the framework recognizes that their influence on employment may be contingent upon race/ethnicity. Separate models for Hispanics, Blacks, and Whites will illuminate these differences. Outlined in the following sections, individual, family, and community supports are proposed to influence unmarried mothers’ employment decisions.

Individual and Family Background Supports

Individual supports are widely recognized to facilitate employment among unmarried mothers (Edin and Harris 1999; Hao 1994; Parish et al. 1991; Park 2005; Taniguchi and Rosenfeld 2002). Additional years of formal education strongly increase employment opportunities (Gault et al. 1998; Hao 1994; Harris 1993; Holzer and LaLonde 2000; Parish et al. 1991; Park 2005). In an examination of unmarried mothers leaving and returning to welfare using the Panel Study of Income Dynamics (PSID) (1984–1986), for example, Harris (1993) found that the probability of leaving welfare for a new job was 2.5 times greater if the mother had a high school diploma.

The influence of individual supports extends to family-of-origin and health characteristics (Acs and Loprest 2001; Baydar et al. 2007; Corcoran et al. 2004; Edin and Harris, 1999; Gault et al. 1998; Hao 1994). In a study of employment patterns among low-income mothers using PSID (1983–1988) data, Edin and Harris (1999) showed that family background characteristics, such as increased parental education levels, were positively related to employment among low-income unmarried mothers. Not surprisingly, health problems negatively influence employment. In Acs and Loprest’s (2001) review of welfare-to-work projects, a significant minority (15–25% depending on the state) of mothers suffered from health problems that hindered employment.

Family Supports

Family resources and in-kind support often provide critical resources that enable maternal employment (Abroms and Goldscheider 2002; Berry et al. 2008; Edin and Harris 1999; Edin and Lein 1997; Gault et al. 1998; Harknett 2006; Henly et al. 2005; Livermore and Powers 2006; Parks-Yancy et al. 2007; Simmons et al. 2007a, b). In examinations of the survival strategies of 379 low-income, unmarried mothers, Edin and Lein (1997) found that although off-the-books work and agency support did not help mothers enter employment, the assistance of informal networks often did. Perceived social support also promotes economic wellbeing and higher levels of employment and earnings (Harknett 2006; Henly et al. 2005). In a recent analysis of the Fragile Families Study, the data used here, Livermore and Powers (2006) found that social support reliance (receiving money from others, borrowing money to pay bills, or moving in with others) increased employment. However, in another analysis of the same data, social capital (access to a variety of emergency supports) was not related to employment (Ciabattari 2007).

Community Supports

Although few studies examine the influence of community supports on employment, their influence is important in light of the expansion of work as a quintessential component of welfare reform. These supports can provide necessary resources to compensate for the additional costs of working such as job training, child-care, and transportation (DeBord et al. 2000; Gault et al. 1998). Livermore and Powers (2006) found that various community resources influenced employment differently—rent assistance was negatively related to employment while job assistance was positively related.

Summary

TANF’s emphasis on work and self-sufficiency changes the employment calculus of low-income unmarried mothers. This study provides an opportunity to examine various dimensions of social support and their influence on post-welfare reform employment. The uneven distribution of material and social resources among Hispanic, Black, and White unmarried mothers of young children raises the question as to whether supports operate differently for these racial/ethnic groups. Although studies have examined race/ethnicity, social support, and employment, this study uniquely considers all three together after welfare reform.

Methodology

Data and Sample

The analysis uses the Fragile Families and Child Wellbeing Study, a longitudinal study begun in 1998 with the overarching goal of providing more accurate information on unmarried parents and their children. Sponsored by the US Department of Health and Human Services, the National Institute of Child Health and Human Development, the National Science Foundation, and 21 private foundations, the dataset and study information is available through the Fragile Families website (Center for Research on Child Wellbeing, n.d.). Using a cluster sample stratified by policy regime, 16 large cities (population greater than 200,000 residents) were selected (4 additional cities were added for convenience), followed by a selection of hospitals within these cities. Although race/ethnicity and income were not stratifying criteria, the selection of unmarried parents resulted in a largely minority, low-income sample—62% Black, 23% Hispanic, and 70% under 200% of the poverty level at the child’s birth. Baseline interviews were conducted with mothers and fathers (both new parents and those with previous children) in hospitals at the child’s birth and follow-up surveys were conducted one, three, and 5 years later (for complete description, see Reichman et al. 2001). When weighted, the sample is nationally representative of mothers giving birth in large cities at the time of data collection.

Measures collected at Baseline and Year 1 were used to predict Year 3 employment.Footnote 3 Marital status, additional births, and labor market engagement influence employment behavior (Downs 2003); therefore, mothers included in the current study met three criteria: continuously unmarried (not married at any survey point), no additional births after the baseline birth, and engaged in the labor market at some point in the year preceding or at some point after the child’s birth (either employed or looking for a job at a minimum of one point).Footnote 4 From the sample of 3,696 unmarried mothers at baseline, 17% of mothers married after giving birth, 19% had additional children, 3% were not engaged in the labor market at any point, and 12% were lost to attrition, resulting in 1,847 mothers used for analysis in this study. After restricting the sample to Hispanic, Black, and White mothers (N = 1,785), data on one or more of the variables used in the analysis were missing for 204 mothers (or 11.4%) in of the sample. Thus, descriptive and empirical analyses include 1,581 mothers.Footnote 5

Measures

Dependent Variable

The dependent variable of interest is the number of hours worked for which mothers received a regular paycheck. For mothers employed at more than one job, work hours were combined for all regular jobs. Because hours employed was not normally distributed, it was collapsed into three categories: not working, working part-time (1–34 h per week), and working full-time (35 h or more per week). These categories were chosen based on the classifications of the US Census Bureau (Downs 2003) and to capture the variation of hours spent working each week. For separate racial/ethnic analyses, due to the limited number of part-time workers, mothers were considered employed (regardless of the number of hours) or not employed.

Independent Variables

Independent variables include race/ethnicity and various forms of social support and individual and family background characteristics. To measure race/ethnicity, the mother was asked her race, followed by a question regarding whether she was of Hispanic or Latina origin or descent. The categories of Hispanic, non-Hispanic Black, and non-Hispanic White were chosen to capture any variation between Hispanics and Blacks and to compare them to the largest, most advantaged racial/ethnic group in the country, Whites.Footnote 6 Persons of other racial or ethnic origins were not included due to an insufficient number of cases.

Family supports include household living arrangement at Year 1, created from a household roster (living alone, living with others, or cohabiting (reference)), whether or not mothers used relative care at Year 1, and whether they perceived access to $200 in an emergency both at Baseline and Year 1. To measure the influence of the child’s father at Year 3, mothers’ ratings of their relationships were coded as excellent/very good/good or fair/poor/no relationship. To measure financial support, mothers were coded as receiving formal child support, informal support only, or no child support. Community support measures included whether or not mothers received rental assistance from the federal, state, or local government, including Section 8 vouchers or public housing; employment assistance, such as job training; or child care referral assistance. To measure individual and family background characteristics, respondent education level was categorized as less than a high school diploma, a diploma or G.E.D. only, or more than a high school diploma. Respondents’ mothers were coded as having at least a high school diploma or not. To measure the presence of health problems, mothers specified whether they had a job-limiting health problem. Whether or not mothers lived with both parents at age 15 captured family-of-origin structure. To measure value orientation, mothers rated how much they agreed with two statements: “The important decisions in the family should be made by the man of the house,” and “It is much better for everyone if the man earns the main living and the woman takes care of the home and family.” To construct the measure for value orientation, mothers’ scores were recoded on a 0–3 scale where 0 equaled strongly disagree and 3 equaled strongly agree. Higher scores indicate higher levels of traditional values. Mothers born outside the US or US territories were coded 1, otherwise 0. In addition to these explanatory supports, due to a possible relationship with employment, multivariate analyses control for the number of toddlers (children under 5 years of age) in the household at each survey point, maternal age, and region of the country.

Analytic Methods

First, frequencies and bivariate statistics for each variable by race/ethnicity are presented. In the descriptive analyses, weighted data are used that adjust for age, race, and education (see Carlson 2006 for a description of weighting procedures). Second, multivariate analyses include both a model to detect how race/ethnicity and social supports influence employment and separate models for Hispanics, Blacks, and Whites to detect whether supports operate differently for each group. Multivariate models are unweighted because the variables for which weights adjust are included in the model and the unweighted model allows for the inclusion of four additional cities not in the nationally-representative sample. Replication of the multivariate findings with weighted data yields similar findings.

To model race/ethnicity and employment, I employ multinomial logistic regression models. This type of regression is useful in order to estimate how a multi-category outcome variable (in this case, part-time and full-time employment) is influenced by a set of continuous or categorical predictor variables relative to a reference category (not employed). Due to few part-time workers, binomial regressions models for racial/ethnic subsets of the data were conducted to distinguish employed mothers from those who were not. Separate binomial logistic regression models for Hispanics, Blacks, and Whites include the same variables in the full sample except in the case of nativity due to few Black and White immigrants. For ease of interpretation, the results for all analyses are presented in odds ratio form with corresponding confidence intervals (Powers and Xie 2000). Models including each block of variables (family support, father support, community support, and individual and family background) added separately were conducted. Log-likelihood ratio tests are presented for each block of supports to consider whether each block of supports significantly improved the model fit in predicting employment both for the full sample and among each racial/ethnic group.

Results

Bivariate Results

Table 1 displays descriptive and bivariate statistics. More than one-half of unmarried mothers were employed full-time at Year 3 and 64% were employed at least part-time. Although a smaller percentage of Hispanics worked full-time in comparison to Whites (49.1% and 53.0% respectively), there were no significant racial/ethnic differences in employment. However, many family, community, and individual supports did vary by race/ethnicity. More Black mothers lived alone, while Hispanic and White mothers had significantly higher levels of cohabitation. The majority of mothers believed that they had access to $200, however Hispanics and Blacks were less likely to report access to this emergency support compared to Whites. Regardless of race/ethnicity, 30% of unmarried mothers received no financial support from their child’s father and almost half had a fair, poor, or no relationship with him. In regard to community support, 11–23% received agency supports. Blacks, and Hispanics with regard to rent assistance, were more likely to receive support than Whites.

Table 1 Weighted distribution of variables used in models by race/ethnicity
Full size table

Unmarried mothers were disadvantaged in terms of their levels of individual supports and their family backgrounds, and this disadvantage is highlighted among Hispanic and Black mothers. Formal education among unmarried mothers was low, with Hispanics faring worse than Blacks or Whites. Over 40% of Hispanics lacked a high school diploma compared to less than one-third of Blacks or Whites. Respondents’ mothers also had low levels of formal education with Hispanics, again, less likely to have a high school diploma (41% versus 75% and 70%, respectively). In addition to low education levels, 7% of mothers had job-limiting health conditions. Blacks experienced a disadvantage in family-of-origin household composition, with only 28% living with both parents at age 15, compared to 48% of Hispanics and Whites. Not surprisingly, mothers reported liberal values (mean = 0.97) with Hispanics significantly more traditional than Whites. Nearly one-fourth of Hispanic unmarried mothers were immigrants compared to very few Blacks or Whites. Together, descriptive analyses reflect high levels of employment, despite weak safety nets and few individual resources. Although not reflected in their levels of employment, Black and Hispanic unmarried mothers experienced distinct disadvantages in comparison to Whites.

Regression Analysis

Analysis of the Full Sample

Without significant racial/ethnic differences in employment in the descriptive analyses, multivariate analyses emphasize relationships between social support and employment and how these vary for Hispanic, Black, and White unmarried mothers. In general, more advantaged characteristics including higher levels of family support and individual supports were associated with full-time employment; only education level and a job-limiting health problem were significantly associated with part-time employment (see leftmost columns in Table 2). Relative care users had 72% higher odds of full-time employment and those mothers who perceived access to $200 had 39% higher odds of full-time employment. Yet, in some instances, such as relationship with child’s father, disadvantage promoted employment. Mothers without a good relationship with their child’s father had 32% higher odds of working full-time. With regard to community support, mothers who received a child care referral had 46% higher odds of full-time employment, whereas mothers who received rent or employment assistance had lower odds of full-time employment. Not surprisingly, education level was positively related to employment. Relative to post-high school education, less than a high school degree and, to a lesser extent, only a high school degree decreased the odds of full-time employment (OR = 0.28 and 0.70, respectively). Poor health was also associated with a lower likelihood of employment. A mother with a job-limiting health problem had 64% lower odds of full-time employment and 48% lower odds of part-time employment.

Table 2 Multinomial and binomial regression models of employment as a function of race/ethnicity
Full size table

Results indicate that access and use of supports generally increased employment except in cases where mothers have few other options, such as mothers without a good relationship with their child’s father or those receiving rent assistance. To further test how supports influenced employment, log likelihood tests were performed on successive models for each set of predictors (family, father, community, and individual and family background) to assess whether these variables improved the model fit. Table 3 (column 1) summarizes the findings and indicates that the introduction of family, community, and individual and family background supports improved the model fit. Although findings including the full sample presented above represent unmarried mothers, separate regression models are required to illustrate how social supports may operate differently among Hispanics, Blacks, and Whites.

Table 3 Multinomial and binomial regression models of employment as a function of race/ethnicity
Full size table

Analyses by Race/Ethnicity

Overall, social supports were related to employment for Hispanics and Blacks but not for Whites (see right-most columns in Table 2). The strong influences of child care stand out for Hispanics—relative care users and those who received child care referrals had more than twice the odds of employment. Among Blacks, rent assistance had a particularly negative influence (OR = 0.63). Less than a high school diploma and a health problem decreased employment for both Hispanics and Blacks. Holding traditional values significantly reduced odds of employment for Hispanics only (OR = 0.52). No support was significantly related to employment among Whites.

Log-likelihood ratio results, displayed in Table 3 (columns 2–4), indicate that all predictor groups improved the model fit for Hispanics and Blacks with the exception of the child’s father’s characteristics. No set of attributes improved the log likelihood fit for Whites.Footnote 7

Discussion

The Fragile Families and Child-Wellbeing data provided the opportunity to examine racial/ethnic differences in employment among unmarried mothers with young children, the relationship between social support and employment, and how this relationship differs for Hispanic, Black, and White mothers. Indeed, complementing previous work, this study finds high levels of employment, despite unmarried mothers’ often difficult social circumstances and low levels of human capital (Downs 2003; Livermore and Powers 2006; Teitler et al. 2004). In additional analyses to better understand mothers not employed (35.6% of formerly engaged unmarried mothers), nearly two-thirds were actively looking for work. Those not employed and not seeking work were most commonly homemakers (24%) or students (17%). Regardless of race/ethnicity, unmarried mothers of toddlers reflect a highly engaged workforce. Employment rates support evidence of converging employment patterns among Hispanics, Blacks, and Whites (Yoon and Waite 1994).

Although employment was similar among the three racial/ethnic groups, social support was not. Blacks generally had less family support and more community support than Whites with Hispanics falling between the two groups. This finding supports earlier work indicating informal family support among Hispanics and Blacks may not be available (Brewster and Padavic 2002; Roschelle 1997). Similar to earlier findings (Yoon and Waite 1994), this study found unique sources of disadvantage for Blacks and Hispanics. Blacks’ disadvantage arose from few family supports, while Hispanics’ arose from low levels of education.

Regression models indicate that both human capital and social capital increase employment, supporting earlier findings (Debord et al. 2000; Livermore and Powers 2006; Urban and Olson, 2005). Also supporting earlier work (Alfred 2005; Edin and Lein 1997; Gault et al. 1998), access to $200 during the child’s infancy and child care availability through kin or referrals was associated with full-time employment. The receipt of rent and employment assistance, more common among Hispanics and Blacks, decreased employment, also supporting earlier work (Livermore and Powers 2006). Mothers often use agency supports (or become entitled to them) as a last resort when few other resources are available. Mothers in these precarious situations, then, would have the most difficultly securing employment.

Separate racial/ethnic analyses indicate the importance of recognizing unique needs of Hispanic, Black, and White unmarried mothers in achieving employment. Although social supports were related to employment for Hispanics and Blacks, these supports were less indicative of Whites’ employment. No block of support variables significantly improved the model fit for Whites, however all blocks, with the exception of father support, improved the model fit for Hispanics and Blacks. Relative care promoted employment among Hispanics and Blacks as did assistance with a child care referral among Hispanics. These findings lend support to expanding child care policies and availability to make employment feasible for additional unmarried mothers, particularly mothers of color. Indeed, Herbst & Barnow (2008) analysis of child care availability in Maryland neighborhoods indicates that the presence of child care is related to increases in maternal employment. The positive influence of a child care referral among Hispanic mothers in this analysis suggests child care may be limited in predominantly Hispanic neighborhoods (Hirshberg 2002).

Family background and individual supports did not contribute to employment rates among Whites; however, they were most influential among Hispanics and Blacks. The lack of significant contribution among Whites supports findings that racially advantaged mothers out of the labor force face higher levels of other barriers, including mental health and child health barriers, than their minority counterparts (Corcoran et al. 2004). In light of the strong influence of individual-level supports on employment for Hispanics and Blacks, lower levels of discretionary assistance, such as educational program referrals, among mothers of color are particularly problematic (Gooden 1998). Policy should encourage mothers to obtain their high school diplomas or an equivalent with targeted programs to address the Hispanic-White education gap.

Similar to earlier post welfare reform studies (Acs and Loprest 2001; Gault et al. 1998), a significant minority (7%) of mothers in this study experienced a job-limiting health problem. This finding supports other recent studies on health as a barrier to employment in the post-welfare reform era. Job-limiting health problems are particularly problematic for Hispanics’ and Blacks’ employment. Better responses to these mothers’ needs are required. Only 12% of mothers in this study who reported that they suffered from a job-limiting health condition received Supplemental Security Income (SSI). Although all mothers would not qualify for SSI, the complicated eligibility process and the high likelihood of initial denial of benefits may discourage potentially eligible recipients from the application process.

This study yields an important step in understanding race/ethnicity, social support, and post-welfare reform employment among unmarried mothers with young children, but should be considered in the context of its limitations. The single measure of employment is most apparent. The number of hours employed at one point does not capture employment instability common among unmarried mothers. Second, quantitative analyses of social support are highly susceptible to omitted variable bias. Due to data limitations, for example, this study did not examine transportation, which may be a necessary link to employment. In analyses conducted including one proxy for transportation, car ownership, results were similar to those presented. Future studies should also include additional measures of neighborhood support, emotional support, and instrumental support from the child’s father. Third, although the time-ordering between predictor variables measured at Year 1 and employment measured at Year 3 supports the possibility that social supports are causally related to employment outcomes, there is a potential for reverse causality. A mother’s employment patterns may also influence the range of supports she finds available. A lack of employment, for example, could cause job-limiting health problems rather than the reciprocal as the present model suggests. Statistical techniques, such as structural equation modeling, should be used in the future to clarify causal relationships.

In light of its limitations, this study advances knowledge surrounding the relationship of social support to employment for unmarried mothers of young children following welfare reform. With similar employment rates among Hispanics, Blacks, and Whites, individual and family supports were associated with increased employment while community supports were associated with decreased employment, possibly reflecting pre-existing hardships. Supports influenced employment to a much greater degree for Hispanics and Blacks than for Whites. Thus, in order to promote successful employment post-welfare reform regardless of race/ethnicity, programs and services must be available and flexible to meet the diverse needs of unmarried mothers.