Neighborhood Disadvantage, Residential Stability, and Perceptions of Instrumental Support among New Mothers

Abstract

Using longitudinal data from the Fragile Families and Child Wellbeing study (N = 4,211), we examine neighborhood disadvantage and perceptions of instrumental support among mothers with young children. We find that: (a) living in a disadvantaged neighborhood is associated with less instrumental support, particularly financial assistance, from family and friends; (b) residential stability is associated with stronger personal safety nets irrespective of neighborhood quality; and (c) mothers who move to a more disadvantaged neighborhood experience a small but significant decline in perceived instrumental support compared to those who do not move. In interpreting these results, we suggest instrumental support may be either a cause or consequence of living in an advantaged neighborhood, but, in either case, neighborhood and social network disadvantages go hand in hand.

The quality of one’s neighborhood is associated with a wide range of outcomes for adults and children. Previous research shows that neighborhood quality is associated with better employment opportunities, increased prospects for social mobility, and higher levels of educational attainment (Jencks & Mayer, 1990; Sampson, Morenoff, & Gannon-Rowley, 2002). Living in an advantaged neighborhood may produce social capital benefits, including a range of instrumental supports such as financial, housing, or child care assistance. In this paper, we examine the relationship between where people live and their ability to access these types of supports from friends and family.

Perceptions of instrumental support – defined as mothers’ perceptions that, if needed, they can rely on friends or family members for various forms of instrumental assistance such as in-kind support or financial assistance – can serve as an important safety net over the life course.Theoretically, having perceived social support, which is different than support one actually receives, indicates that mothers will have resources available for coping with everyday stresses and emergencies (Thoits, 1986). Perceived social support may also facilitate mothers’ upward social mobility (Briggs, 1998; Henly, Danziger, & Offer, 2005).

Having available in-kind and financial assistance can be particularly important for new mothers, as the transition to parenthood is a turning point in one’s life and may be accompanied with emotional and financial stress (Cowan & Cowan, 1992; Melson, Windecker-Nelson, & Schwarz, 1998; Mulsow, Caldera, Pursley, Reifman, & Huston, 2002). Available support may reduce such stress, promote positive parenting behaviors, and facilitate social mobility, all of which are directly associated with maternal and child well-being. Thus, having someone available to provide child care, a place to live, or small amounts of monetary assistance when needs arise could make a considerable difference in the lives of new mothers. Indeed, prior research suggests that instrumental support, particularly from family members, is strongly linked to economic stability and psychological well-being (Harknett, 2006; Henly et al., 2005; Ladewig, McGee, & Newell, 1990; Turner, 1981; Wethington & Kessler, 1986).

Though perceived support has important implications for well-being, it is an outcome generally ignored in the neighborhood effects literature. Individual-level predictors of instrumental support are well-documented (Eggebeen, 2005; Eggebeen & Hogan, 1990; Hogan, Eggebeen, & Clogg 1993), but researchers have paid much less attention to the importance of place in determining who has and who lacks support from social connections. Yet, we have reason to expect neighborhoods affect instrumental support. Social ties are partly determined by where one lives, and prior research shows people in economically disadvantaged neighborhoods experience social isolation (Wilson, 1987).

To learn more about the relationship between neighborhoods and perceived instrumental support, we use data from the Fragile Families and Child Wellbeing survey, a longitudinal survey of a birth cohort of nearly 5,000 children and their parents. We focus on mothers’ perceived support, because mothers almost universally have custody of children when parents live apart. Mothers in the study live in a large number of neighborhoods that vary widely in their conditions. About half of the mothers moved between the 12- and 30-month waves, facilitating a longitudinal analysis of residential stability and perceived instrumental support.

BACKGROUND

Social Isolation and Neighborhood Socioeconomic Disadvantage

A large body of literature documents the importance of neighborhood conditions for individual well-being. In The Truly Disadvantaged (1987), Wilson posited that residents of poor communities are socially isolated from mainstream social networks, resources, and institutions, and this seminal work spawned a great deal of research on the relationship between neighborhoods and social behavior (for reviews, see Ellen & Turner, 1997; Sampson et al., 2002; Small & Newman, 2001). Wilson’s theory implies a spatial nature of behavior; individuals’ actions are shaped by where they live, and neighborhood characteristics–such as high rates of unemployment and crime, inadequate housing conditions, and a lack of adult role models–shape normative climates that define acceptable and unacceptable behavior. The social isolation that many people in high-poverty neighborhoods experience can ultimately prohibit social mobility (Anderson, 1990, 1999; Massey & Denton, 1993; Roschelle, 1997; Wilson, 1987, 1996). Although much of the neighborhood effects literature focuses on child or adolescent outcomes (Sampson et al., 2002), recent research has looked at the role of neighborhood context in shaping adult outcomes such as family structure (South & Baumer, 2000; South & Crowder, 2000; South, 2001), social networks (Kleit, 2001; Kissane & Clampet-Lundquist, 2005), and employment (Reingold, 1998; Elliott, 1999; Kling, Liebman, & Katz, 2006).

There are several reasons to explain why neighborhood conditions may be associated with available support. First, individuals living in poor communities may lack the resources needed to provide certain types of instrumental support. Mothers in poor neighborhoods may exhaust support from their resource-poor networks, especially when these mothers are constrained in their ability to reciprocate support because of young children and poverty (Lye, 1996, Gallagher & Gerstel, 2001). Additionally, in poor neighborhoods, it may be more difficult to develop supportive relationships with non-kin, and these relationships may take longer to cultivate because of the distrust that arises in neighborhoods with scarce resources (Ross, Mirowsky, & Pribesh, 2001; Furstenberg, (1993). Though this distrust may not apply to family members, mothers who live in poor neighborhoods may be isolated from kin networks who live outside their neighborhood. For instance, across-neighborhood relationships may be weakened if safety concerns restrict mobility in and out of the neighborhood. Finally, another possibility is that neighborhoods are associated with instrumental support because of selection: The types of people who select into good neighborhoods also happen to be those who have supportive social networks. Those who lack supportive social networks may be more likely to end up in poor quality neighborhoods as a consequence of lacking instrumental support.

Despite compelling theoretical ideas about a lack of social support in poor neighborhoods, little empirical research has examined the extent to which individuals in poor neighborhoods are isolated from networks of family and friends. Few studies have directly analyzed the relationship between neighborhoods and available support in the United States. Two exceptions found mixed results: Fernandez and Harris (1992) found Black women in poor neighborhoods are cut off from network ties, and Rankin and Quane (2000) found neighborhood poverty is not related to social isolation among Black mothers in racially-segregated Chicago neighborhoods. The inconsistent findings may be attributed to several factors. First, the two studies employ different measures of social isolation. Additionally, the Fernandez and Harris study only looks at individuals in poor Census tracts and does not consider length of time in neighborhood.

Residential stability is important when considering how neighborhood quality matters (Jencks & Mayer, 1990; Tienda, 1991). Length of exposure to the ecological setting plays an important role in one’s feelings of attachment to the community (Kasarda & Janowitz, 1974). Additionally, neighborhood-based friendship ties, positive neighborhood evaluations, and social activities are positively associated with time spent in neighborhood (Sampson, 1988). Another study finds that community-level residential stability interacts with neighborhood disadvantage in predicting support. Black women in disadvantaged neighborhoods, for example, receive more support than their counterparts in more advantaged neighborhoods, but only in neighborhoods with high levels of residential stability (Schieman, 2005). Thus, neighborhoods with greater residential stability may foster closely-knit communities, complete with social cohesion and trust, where women can rely on neighbors when confronted with hardship.

Evidence that residential instability can disrupt social ties comes from the Moving to Opportunity (MTO) experiment, a federal housing mobility program that moved people from public housing in disadvantaged neighborhoods to mixed-income communities. Moving, even to a low-poverty neighborhood through the MTO program, disrupted social connections (Kissane & Clampet-Lundquist, 2005). Also, parents who moved to less poor neighborhoods through MTO were less likely to know the parents of their children’s friends (Pettit & McLanahan, 2003). This is consistent with other work that finds individuals who move to advantaged neighborhoods may lose social capital because they do not frequently exchange information and resources with their new neighbors (Clampet-Lundquist, 2004; Kleit, 2001; Turney, Clampet-Lundquist, Edin, Kling, & Duncan, 2006).

Perceived Instrumental Support

A separate literature documents the importance of social support for individual well-being. Social networks can serve as a private safety net by providing instrumental assistance such as loans, housing, or child care (Collins, 2000; Edin & Lein, 1997; Harknett, 2006). This private safety net is particularly important for the everyday survival of low-income families, as recent changes in welfare legislation have made it increasingly difficult for individuals to rely on public assistance (Haider & McGarry, 2005; Henly et al., 2005).

Support from network members tends to increase with parenthood (Belsky & Rovine, 1984; Nomaguchi & Milkie, 2003), and this support may be especially important for mothers who recently had a child (Belsky & Rovine, 1984; Nichols, Elman, & Feltey, 2006). For many, parenthood poses significant financial and emotional challenges (Cowan & Cowan, 1992; Mulsow et al., 2002). Studies of mothers’ instrumental support sometimes measure instrumental support that was received during a given period. One drawback of this approach is that mothers’ receipt of support is highly correlated with their need for support. Measures of received support cannot distinguish between two very different types of mothers: those who need instrumental support from their networks but are not receiving it and those who have no need for support (Taylor, 1990). An alternative approach, and the one used in this paper, is to measure perceived support, which captures potential support that mothers can draw on when needs arise (Wethington & Kessler, 1986). The latter approach provides a measure of the personal safety net that mothers have to fall back on in times of hardship and avoids the problem of conflating need for support with availability of that support.

Prior research shows that perceptions of instrumental support are correlated with economic, physical, and psychological well-being. Perceiving greater levels of support is associated with a reduction in the likelihood of living in poverty and less perceived economic hardship among low-income families, as well as higher levels of employment and earnings and less reliance on welfare among unmarried mothers (Harknett, 2006; Henly et al., 2005). This could be particularly important, as single mothers often face more substantial economic and psychological challenges than their married counterparts (Sigle-Rushton & McLanahan, 2004). Importantly, perceived support can improve physical and mental well-being (House, Umberson, & Landis, 1988; Wethington & Kessler, 1986), and plays an important role in mediating or moderating other outcomes such as child development, health conditions, or marital adjustment (Baron & Kenney, 1986; Graham, Fischer, Crawford, Fitzpatrick, & Bina, 2000; Holmbeck, 1997; Rose, Holmbeck, Coakley, & Franks, 2004).

Predictors of Instrumental Support

Although support from friends and family members is important for adult well-being, these resources are not evenly distributed across the population. For example, qualitative research demonstrates the strength of familial support within low-income Black communities (Aschenbrenner, 1973; Hannerz, 1969; Harrington, 1962; Newman, 1999; Stack, 1974). Nationally representative studies generally suggest that Whites are more likely than Blacks and other minority groups to receive financial, and emotional support (Cooney & Uhlenberg, 1992; Eggebeen, 1992; Eggebeen & Hogan, 1990; Hofferth, 1984; Hogan et al., 1993), though Blacks may be more likely to receive child care support (Benin & Keith, 1995). Additionally, many immigrant families are characterized by the presence of extended family members, and immigrants often have exchange relationships with both household and non-household members (Glick, 1999; Glick & Van Hook, 2002). Hao (2003), however, found that immigrant families have less access to support than native families. Age is also associated with support from kin networks; individuals in their 20s are more likely than their older counterparts to receive support as they make the transition to adulthood (Eggebeen & Hogan, 1990).

Additionally, socioeconomic factors including education and income are highly and positively correlated with receipt of support (Eggebeen & Hogan, 1990; Jayakody, 1998; for inconsistent findings, see Benin & Keith, 1995). Mental and physical health may also be important determinants of support, and lower psychological well-being may negatively affect perceptions of available support (Blazer, 1982; Sarason, Sarason, & Pierce, 1990).

Other predictors of instrumental support include family structure characteristics, including marital status and number of children. Single mothers are more likely than their married counterparts to engage in multiple dimensions of exchange (Hogan et al., 1993). Individuals in cohabiting relationships, though, report receiving less support than their married counterparts (Eggebeen, 2005; Hao, 1996; Marks & McLanahan, 1993).

Hypotheses

Based on prior research, we test the following hypotheses: (a) Mothers in disadvantaged neighborhoods have less perceived instrumental support than their counterparts in advantaged neighborhoods. Although mothers do select into neighborhoods, we expect mothers in more socioeconomically disadvantaged neighborhoods to have less perceived instrumental support even after controlling for individual-level characteristics. Furthermore, we expect mothers in poor neighborhoods, compared to their counterparts in more advantaged neighborhoods, will have resource-poor networks and will be especially unlikely to access large amounts of financial support. (b) Residential stability increases perceptions of instrumental support. We expect that greater individual-level residential stability will be associated with stronger social ties and community attachment that leads to higher levels of perceived support. A recent move may disrupt social ties. Residential stability may be more beneficial for those living in higher quality neighborhoods. For this reason, we also consider whether residential stability increases perceptions of support in disadvantaged and advantaged neighborhoods alike. A positive effect of residential stability may be muted in disadvantaged neighborhoods if crime and feelings of insecurity inhibit the development of relationships with neighbors. (c) Moving to a more disadvantaged neighborhood decreases perceived instrumental support. We expect mothers who move to a lower-quality neighborhood between the two waves to report less perceived instrumental support than mothers who do not move. Mothers who move to less advantaged neighborhoods may face challenges developing ties in their new community, potentially because of fear and distrust in these communities or because their neighbors are simply unable to provide assistance. As moving can potentially disrupt existing social ties, we do not expect mothers who move to advantaged neighborhoods to experience an increase in support.

METHOD

Participants

We use data from the Fragile Families and Child Wellbeing survey, a longitudinal study of nearly 5,000 new and mostly unmarried parents in 20 U.S. cities. Mothers completed a 30- to 40-minute in-person interview at the hospital after the birth of their child, between February 1998 and September 2000, and were interviewed by telephone when their child was approximately 12, 30, and 60 months old. This paper uses data from the first three waves of data collection to look at mothers’ perceptions of instrumental support while their children are still quite young. See Reichman, Teitler, Garfinkel, and McLanahan (2001) for more information.

Our analytic sample for the analyses that rely on the baseline and 12-month follow-up is the 4,211 mothers who completed the baseline and 12-month survey and were not missing data on perceptions of instrumental support (8 observations) or neighborhood conditions (146 observations). For the longitudinal analyses, our analytic sample is 3,871 mothers, mothers who participated in all three waves of data collection. Survey response rates at baseline were 82% for married and 87% for unmarried mothers. Of those, 91% of married and 90% of unmarried mothers completed the 12-month survey and 89% of married and 87% of unmarried mothers completed the 30-month survey. Mothers in the sample lived in 2,586 different Census tracts, which provides substantial variation in neighborhood conditions. Importantly, Census tracts are not synonymous with neighborhoods, but this is the closest we can come to approximating neighborhood conditions. Most research on neighborhood conditions, in fact, relies on Census tract data (Mayer & Jencks, 1990; Sampson et al., 2002).

Although survey response rates were relatively high, it is possible that those who moved between waves and those most lacking perceived instrumental support would be more challenging to locate and, therefore, underrepresented in the sample. About 46% of mothers moved between the baseline and 12-month waves and 49% of mothers moved between the 12- and 30-month waves, though, suggesting that researchers were able to track down many movers in the later waves of the survey. Also, we compared the baseline characteristics of respondents and non-respondents at the 12- and 30-month waves and found that respondents and non-respondents lived in neighborhoods of similar quality and had similar levels of perceived instrumental support at baseline. Mothers in the analytic sample are older than their counterparts in the full sample (p < 0.001), though little other significant differences exist. Finally, analyses that use only the first two waves of data (baseline and 12-month) are consistent with the results we present, which are based on mothers who participated in all three waves of data collection.

Variables

Perceptions of social support

The dependent variable in the analyses measure mothers’ perceptions that they have various types of instrumental assistance available. Mothers were asked if they could count on someone, during the next year, for the following: to loan them $200, to loan them $1,000, to help with babysitting or child care, to provide them with a place to live, to cosign a bank loan for $1,000, and to cosign a bank loan for $5,000. Each of these questions comprise a dichotomous dependent variable: 1= perceived instrumental support and 0 = no instrumental support. These six items are all of practical, substantive importance for new mothers, but are not exhaustive of all types of instrumental support that mothers may need or receive. Data limitations, for example, preclude us from examining perceptions of emotional support. .

A principal components factor analysis informed the construction of our three scales. The factor analysis identified two separate factors from the six items. All six items loaded positively onto the first factor, and all six had similar factor loadings. The similar factor loadings suggest that each of the six types of support should be given about equal weight in the scale. For ease of interpretation, we present a count of types of support rather than the scale derived through factor analysis. Nevertheless, our results are consistent in magnitude and significance when we substitute the count scale with the factor.

For the second factor identified through factor analysis, the factor loadings were positive for a $200 loan, child care, and a place to live, and negative for the $1,000 loan or for cosigning a $1,000 or $5,000 loan. The loadings for this factor suggest a distinction between in-kind support and a small loan, on the one hand, and larger amounts of financial assistance on the other. According to the second factor, we divided support into the in-kind assistance scale and the large financial assistance scale described below. Again, for ease of interpretation, we analyze counts of types of assistance.

The results we present use one of three scales: (1) an overall scale of instrumental support, (2) a scale measuring in-kind and small financial assistance, and (3) a scale measuring large financial assistance. Our first scale is a sum of the number of types of support available at the 12-month wave (α = 0.805): 0 = has no instrumental support available, 6 = has all types of instrumental support available. The in-kind and small financial assistance scale, hereafter referred to as in-kind assistance, is a sum of receipt of the following types of support: loan for $200, child care assistance, and housing assistance (α = 0.726). The large financial assistance scale is a sum of responses to the following types of support: loan for $1,000, cosigner for $1,000, and cosigner for $5,000 (α = 0.804). Both the in-kind assistance and large financial assistance scales range from 0 to 3; the higher the value, the more support one has available.

Finally, we analyze change in support over time as a function of change in neighborhood quality (Allison, 1990). This variable is the difference in available support between the 12-month and the 30-month wave, and ranges from –6 to 6. Higher values indicate the respondent gained available support between waves.

Neighborhood socioeconomic disadvantage and residential stability

Our main independent variable is an index of neighborhood socioeconomic disadvantage, taken from tract-level 2000 Census data where the mothers lived at the 12-month and 30-month interview. The index is composed of the following neighborhood-level variables: percent greater than 25 years old without a high school degree, percent unemployed in the civilian labor force, percent living below the poverty line, and percent receiving public assistance (α = 0.833 at the 12-month wave, α = 0.837 at the 30-month wave). We standardize these four variables, and create the index by summing the standardized variables together.

The first part of our analysis uses the continuous neighborhood disadvantage index at the 12-month wave. The higher the index, the greater the level of neighborhood socioeconomic disadvantage. We also divide this index into quintiles to test for non-linearities in the relationship between neighborhood disadvantage and perceived instrumental support, as neighborhood conditions may only matter when neighborhood quality is extremely poor (Duncan & Raudenbush, 1999; Jencks & Mayer, 1990). In Appendix A, we show models predicting perceived instrumental support that replace the index of neighborhood disadvantage with each of the four individual neighborhood characteristics. These results show that the individual components of the index are all consistent, significant predictors of support.

For our next set of analyses, the main independent variable is a measure of change in neighborhood quality. For those who moved, we create a neighborhood change index that represents the difference in the neighborhood disadvantage index between the 12- and 30-month waves, and then create quintiles of change. Using this index, we group mothers who moved into three categories: moved to more advantaged neighborhoods, moved to more disadvantaged neighborhoods, or moved to neighborhoods of similar quality. Because many movers moved to neighborhoods of similar quality, we combine the middle three quintiles of the neighborhood change index into a group that experienced little change in neighborhood quality. Moving to a more advantaged neighborhood is defined as the first quintile of the neighborhood change index, and represents a sizeable increase in neighborhood quality between waves. Moving to a more disadvantaged neighborhood is defined as the fifth quintile of the neighborhood change index, and represents a sizeable decrease in neighborhood quality between waves. Changes in neighborhood quality are a result of an individual moving, not one’s neighborhood becoming more or less disadvantaged over time. We compare these three groups of movers to mothers who did not move.

Our final independent variable is years lived in current neighborhood. We estimate time lived in current neighborhood by using information about how long individuals lived in their neighborhood at baseline, along with whether or not the respondent moved between baseline and the 12-month interview.

Control variables

The multivariate analyses control for characteristics we expect to be correlated with neighborhood socioeconomic conditions and perceived instrumental support. We include a host of variables that control for family background to minimize the selection bias associated with studying neighborhood characteristics, as individuals have a certain amount of choice in deciding what neighborhood they live in and how long they remain there (Jencks & Mayer, 1990; Sampson et al., 2002). We control for the following, all measured at the 12-month wave: mother’s race, mother and father are a mixed-race couple, mother’s immigrant status, mother’s age, grandmother resides in the household, mother’s number of children, birth is the mother’s first child, mother’s education, mother’s employment status, household income, mother’s overall health, mother’s depressive symptoms, mother’s relationship with father, mother’s homeownership, and grandparents’ education.

Race is represented by a series of dummy variables: White (reference category), Black, Hispanic, and other race. We measure education by the following: less than high school diploma (includes mothers with a GED and is the reference category), high school diploma, and post-secondary education. Employment status is a dummy variable representing whether the mother worked in the past two weeks. Poor or fair health is a dummy variable indicating the mothers’ self-reported health status. We control for depression in the past year, measured with the Composite International Diagnostic Interview-Short Form (CIDI-SF) Version 1.0. Relationship status is measured by a series of dummy variables: married (reference category), cohabiting, dating (romantically involved but not living together), and separated. We also include a dummy variable indicating the mother is a homeowner to control for wealth. This indicator may also indicate residential stability, as homeowners move less frequently than those who rent. Finally, we include a dummy variable indicating whether the mother’s parents graduated from college to control for the SES of her family of origin. Few observations are missing control variables, and we impute these missing values using a regression-based approach. Results using listwise deletion (not presented) produce similar coefficients.

Procedures

We conduct three sets of multivariate analyses. First, we use Poisson regression models to estimate the relationship between neighborhood disadvantage and perceptions of instrumental support. Poisson regression is appropriate because the dependent variable is a count variable and skewed to the left (Kennedy, 1998). Next, we look at the extent to which neighborhood disadvantage predicts in-kind assistance and large financial assistance. Finally, we examine how moving to a more advantaged or disadvantaged neighborhood might influence one’s available support. This analysis allows us to see whether a change in neighborhood context triggers a change in perceived support. If moving to a new neighborhood is associated with a change in perceived support, the relationship between neighborhood and support is not confounded by time-invariant characteristics of the mothers.

All multivariate analyses use the cluster option in Stata to adjust for non-independence of mothers who live in the same Census tract, though standard errors are similar when this option is not used. In analyses not presented, we use the survey command in Stata; this technique produces coefficients consistent with the presented results.

RESULTS

Table 1 shows descriptive statistics for all variables. Between 83% and 88% of mothers have small amounts of monetary assistance, housing, and child care support available. Fewer mothers report having large amounts of monetary assistance. Less than half of mothers have someone to loan them $1,000, and only two in five have someone who would cosign a $5,000 loan. On average, mothers have 4 of the 6 types of perceived instrumental support. Mothers report slightly more support when their children are about 12 months old, compared to when their children are 30 months old, but the difference is negligible. About 40% of mothers experienced no change in support between waves, and the remaining mothers are about equally split between those who experienced an increase in support and those who experienced a decrease in support (descriptives not shown).

Table 1.

Descriptive Statistics of Variables Used in Analyses (N = 4,211)

Variable name	Mean	SD	Min	Max
Perceived instrumental support: 30-month
Loan for $200	0.829	n/a	0.000	1.000
Loan for $1,000	0.484	n/a	0.000	1.000
Child care	0.868	n/a	0.000	1.000
Housing	0.828	n/a	0.000	1.000
Cosigner for $1,000	0.575	n/a	0.000	1.000
Cosigner for $5,000	0.386	n/a	0.000	1.000
Sum of perceived support	3.972	1.869	0.000	6.000
Change in support between waves	0.058	1.591	−6.000	6.000
Neighborhood characteristics: 30-month
Neighborhood disadvantage	0.000	3.482	−8.487	15.369
Proportion without college degree	0.828	0.148	0.106	1.000
Proportion below poverty line	0.180	0.137	0.000	0.927
Proportion receiving public assistance	0.076	0.068	0.000	0.850
Proportion unemployed	0.103	0.073	0.000	0.664
Moved since 12-month interview	0.488	n/a	0.000	1.000
Change in disadvantage	0.003	2.247	12.366	15.369
Covariates
White	0.215	n/a	0.000	1.000
Black	0.493	n/a	0.000	1.000
Hispanic	0.251	n/a	0.000	1.000
Other race	0.039	n/a	0.000	1.000
Mixed-race couple	0.173	n/a	0.000	1.000
Immigrant	0.158	n/a	0.000	1.000
Age	26.438	6.093	12.000	49.000
Grandmother in household	0.187	n/a	0.000	1.000
Number of children in household	2.306	1.329	0.000	10.000
First birth	0.385	n/a	0.000	1.000
Less than high school diploma	0.387	n/a	0.000	1.000
High school diploma	0.254	n/a	0.000	1.000
Post-secondary education	0.359	n/a	0.000	1.000
Employed	0.531	n/a	0.000	1.000
Household income (median)	22,000	35,997	0.000	500,000
Poor or fair health	0.137	n/a	0.000	1.000
Depressive symptoms	0.157	n/a	0.000	1.000
Married	0.302	n/a	0.000	1.000
Cohabiting	0.276	n/a	0.000	1.000
Dating	0.098	n/a	0.000	1.000
Separated	0.324	n/a	0.000	1.000
Home owner	0.173	n/a	0.000	1.000
One parent graduated college	0.178	n/a	0.000	1.000
Years in neighborhood	4.185	6.551	0.379	42.528

Table 1 also shows mothers’ neighborhood conditions at the 30-month wave. When their children are about one year old, mothers, on average, lived in neighborhoods where about 18% of people lived at or below the poverty line. About 8% of their neighbors were receiving public assistance and about 10% of their neighbors were unemployed. Average neighborhood characteristics are similar at the 12-month wave.

Turning briefly to demographic characteristics of the sample, most mothers are minorities–about half are Black and one-quarter are Hispanic. The average mother is 26 years old and, for nearly two in five mothers, this is her first birth. About 30% of mothers are married to the father of their child, 28% are cohabiting, 10% are dating, and 32% are separated.

Is Neighborhood Disadvantage Associated with Perceived Instrumental Support?

Table 2 presents Poisson regression models that predict the number of types of instrumental support mothers perceive as available to them. Model 1 shows neighborhood disadvantage is negatively associated with perceived support. A one-unit increase in mother’s neighborhood disadvantage is associated with a 0.036-point decrease in available support. The magnitude of this relationship is small but statistically significant.

Table 2.

Poisson Regression Analyses Predicting Mothers’ Perceived Instrumental Support (N = 4,211)

	Model 1	Model 2	Model 3	Model 4
Neighborhood disadvantage	−0.036***	−0.009***	−0.009***
First quintile				0.078**
Second quintile				0.066**
Third quintile				0.056*
Fourth quintile				0.035
Fifth quintile (omitted)				---
White (omitted)				---
Black		−0.075***	−0.074***	−0.079***
Hispanic		−0.047*	−0.047*	−0.051**
Other race		−0.089*	−0.087*	−0.089*
Mixed-race couple		−0.040*	−0.041*	−0.041*
Immigrant		−0.053*	−0.048*	−0.047*
Age		−0.005***	−0.006***	−0.006***
Grandmother in household		0.037*	0.029	0.029
Number of children		−0.021**	−0.021**	−0.022**
First birth		0.049**	0.049**	0.051**
Less than high school (omitted)		---	---	---
High school diploma		0.039*	0.035	0.034
Post-secondary education		0.078***	0.076***	0.078***
Employed		0.042**	0.042**	0.043**
Household income (log)		0.062***	0.061***	0.062***
Poor or fair health		−0.160***	−0.159***	−0.160***
Depressive symptoms		−0.135***	−0.133***	−0.134***
Married (omitted)		---	---	---
Cohabiting		−0.042*	−0.043*	−0.044*
Dating		−0.040	−0.046	−0.047
Separated		−0.116***	−0.118***	−0.119***
Home owner		0.113***	0.112***	0.114***
One parent graduated college		0.046**	0.046**	0.048**
Years in neighborhood			0.003***	0.003***
Constant	1.384***	0.998***	1.010***	0.958***
Wald X2	316.11	1475.61	1496.58	1499.09
Log pseudolikelihood	−8637.11	−8338.99	−8335.51	−8336.87

^*p < 0.05.

^**p < 0.01.

^***p < 0.001.

We control for individual characteristics in Model 2, and find that neighborhood disadvantage is still a significant predictor of perceived instrumental support. Mothers experience a 0.009-point decrease in perceived instrumental support for every one-unit increase in neighborhood disadvantage. Again, the magnitude of the coefficient is small but remains statistically significant. Mothers in lower-quality neighborhoods–independent of demographic and socioeconomic characteristics–have slightly lower levels of perceived instrumental support than their counterparts in more advantaged neighborhoods. Holding constant these individual-level characteristics, a change from the best to worst neighborhood is associated with one-quarter of a point increase on the perceived support scale.

In Model 3, we add length of time lived in a neighborhood, which does not change the relationship between neighborhood disadvantage and perceived support. Mothers still experience a 0.009-point decrease in perceived instrumental support for every one-unit increase in neighborhood disadvantage. Length of time in neighborhood is associated with higher levels of perceived instrumental support. For every additional year that mothers live in their neighborhood, they experience a small (0.003-point) increase in perceived instrumental support. In analyses not shown here, we replace the continuous variable of length of time in neighborhood with dichotomous variables: less than one year (reference category), one to five years, or five or more years. Living in one’s neighborhood for between one and five years, compared to living there for one year or less, is associated with more perceived instrumental support, although this coefficient is not significant. But mothers benefit from living in their neighborhood for five years or longer, as this is associated with a statistically significant 0.058-point increase in support.

In Model 4, we replace the continuous measure of neighborhood conditions with quintiles of neighborhood socioeconomic disadvantage. This model suggests a linear relationship: The more advantaged the neighborhood, the higher the levels of perceived support. Living in the most disadvantaged neighborhoods is associated with significantly lower perceived support compared to each of the other neighborhood quintiles. Mothers living in the least disadvantaged neighborhoods, for example, average 0.078 points higher on the perceived support scale compared to those living in the most disadvantaged neighborhoods. Likewise, mothers living in the second best quintile average 0.066 points higher on the perceived support scale, but the difference between the fourth and fifth quintiles is not statistically significant. Neighborhood poverty rate provides an illustration of neighborhood conditions to put results in context. Mothers living in the first quintile have an average poverty rate of 6%, while mothers living in the fifth quintile have an average poverty rate of 34%.

Neighborhood disadvantage is negatively associated with perceived instrumental support for new mothers and residential stability is positively associated with greater perceived support. In results not shown, we test the hypothesis that the positive relationship between residential stability and perceived support is concentrated in advantaged neighborhoods by including interactions between quintiles of neighborhood disadvantage and time spent in a neighborhood. None of these interactions approach statistical significance, suggesting the relationship between residential stability and perceived support does not depend on neighborhood quality.

Although not the primary focus of our analysis, other predictors of perceived support are consistent with prior research. Socioeconomic characteristics, such as employment and household income, are positively associated with perceived instrumental support for new mothers. Mothers who are cohabiting with or separated from their child’s father report less available instrumental support than their married counterparts. Older mothers report less available support, and number of children is negatively associated with perceived instrumental support.

How are In-Kind and Large Financial Assistance Associated with Perceived Support?

Our discussion thus far has focused on the relationship between neighborhood disadvantage and the average number of six types of instrumental support. The previous analyses combine child care, housing, and various amounts of financial support into one scale. Here, we separate the six types of instrumental support into two categories: in-kind assistance and large financial assistance. Table 3 presents Poisson regression results predicting each of these two types of support. All covariates are included in these models, but we only present coefficients for neighborhood socioeconomic disadvantage.

Table 3.

Poisson Regression Analyses Predicting In-Kind and Large Financial Assistance (N = 4,211)

	In-kind assistance		Large financial assistance
	Model 1	Model 2	Model 1	Model 2
Neighborhood disadvantage	−0.003		−0.019***
First quintile		0.019		0.173***
Second quintile		0.027		0.148**
Third quintile		0.030		0.111*
Fourth quintile		0.014		0.082
Fifth quintile (omitted)		---		---
Constant	0.823	0.802	−0.695	−0.816
Wald X2	493.19	494.10	1574.81	1580.47
Log psuedolikelihood	−6556.98	−6556.93	−6175.49	−6176.72

^*p < 0.05.

^**p < 0.01.

^***p < 0.001. Robust standard errors in parentheses. Models include all covariates from Table 2.

Interestingly, neighborhood conditions are not associated with in-kind assistance for new mothers. Mothers in more advantaged and disadvantaged neighborhoods perceive similar amounts of child care, housing, and small financial assistance available. Neighborhood disadvantage is associated with perceived availability of large amounts of financial assistance. Every one-unit increase in neighborhood disadvantage is associated with a 0.022 decrease in large financial assistance. This pattern of results suggests that mothers in poor neighborhoods are not entirely lacking instrumental support, but rather are lacking social connections with the ability to provide large amounts of financial assistance.

In analyses not shown here, we use logistic regression to predict the odds of perceiving each of the six individual types of support. In general, we find an association between neighborhood disadvantage and perceived support. Child care assistance is the one exception to this: new mothers in poor and nonpoor neighborhoods are equally likely to perceive available child care assistance. Unlike the other forms of perceived support we examine, child care assistance does not require economic resources. Thus, while an unemployed friend or family member may not be able to provide financial resources, he or she may be able to assist with child care. Notably, child care support also requires trust in the provider. It may not be, then, that social ties in poor neighborhoods are less supportive and trustworthy, but are instead simply more disadvantaged.

How Does Support Change When Neighborhood Quality Changes?

The prior analyses suggest that living in an advantaged community is associated with advantages in perceived instrumental support, but are limited in that they are subject to potential unobserved heterogeneity. Unobserved characteristics–such as mother’s assertiveness, extrovertedness, and resourcefulnesss–may confound the relationship between perceived instrumental support and neighborhood quality. Further, these analyses cannot provide information on causal direction. Lacking perceived support could be a cause or a consequence of disadvantaged neighborhood conditions.

Table 4 looks at the relationship between change in neighborhood quality and change in perceived instrumental support, which helps address some of the selection associated with neighborhood choice. When a mother changes neighborhoods, many of her characteristics–including personality characteristics we cannot control for in the prior analyses–remain constant. Thus, if she experiences a change in support after moving, it is likely that the change can be attributed to the change in neighborhood context. However, we still cannot account for all time-varying characteristics that may affect moving and changes in support.

Table 4.

Ordinary Least Squares Regression Estimating the Change in Perceived Instrumental Support Between 12- and 30-Month Waves (N = 3,871)

Did not move (omitted)	---
Moved to more advantaged neighborhood	−0.020
Moved to neighborhood of similar quality	−0.026
Moved to more disadvantaged neighborhood	−0.240*
Constant	−0.200

^*p < 0.05. Robust standard errors in parentheses. Models include all covariates from Table 2. “Moved to more advantaged neighborhood” includes mothers in the top quintile of change in disadvantage between the 12-month and 30-month wave. “Moved to more disadvantaged neighborhood” includes mothers in the bottom quintile of change in neighborhood disadvantage. “Moved to a neighborhood of similar quality” includes all other mothers who moved.

Moving to more advantaged neighborhoods, compared to not moving between waves, is not associated with a change in perceived support. Additionally, mothers who move to a neighborhood of similar quality do not experience an appreciable change in perceived support. However, a decrease in neighborhood quality is associated with a reduction in perceived support. Net of individual characteristics, moving to a more disadvantaged neighborhood, compared to not moving, is associated with a 0.240-point reduction in perceived support.

In results not shown, we check the robustness of these results. First, we include a control for neighborhood disadvantage at the 12-month wave into the models, which allows us to look at change in neighborhood quality net of where one begins. Second, we substitute our measures of change in neighborhood disadvantage with measures of change in neighborhood poverty rates as a different way of operationalizing neighborhood quality. Third, we restrict the sample to only movers, and compare those who experienced an increase or decrease in neighborhood quality to those who did not experience a substantial change. Fourth, because it is possible that moving alters the relationship between neighborhood disadvantage and perceived support differently for married and unmarried mothers, we include an interaction term between mother’s relationship status and her residential mobility between waves. However, none of these interaction terms reach statistical significance.

Finally, we incorporate some time-varying characteristics that may confound the relationship between neighborhood disadvantage and perceived support. In particular, we include dummy variables that indicate if the mother separated from her child’s father between waves or if the mother married her child’s father between waves. Neither of these changes in family structure significantly predict perceived support, and do not substantively change the neighborhood quality coefficient. In fact, in each of these model permutations, the general finding persists: Mothers who experience a decrease in neighborhood quality report less available support from their friends and family members than those who do not experience a change in quality.

DISCUSSION

This paper tests the theory that neighborhood disadvantages are associated with social network disadvantages, manifested as a lack of perceived instrumental support. We examine a sample of 4,211 mothers in the Fragile Families and Child Wellbeing study. These mothers resided in more than 2,500 different Census tracts and were exposed to widely varying neighborhood conditions. Consistent with our hypotheses, neighborhood and social network disadvantages are correlated. Though the differences are small, mothers living in more advantaged neighborhoods report access to significantly more types of instrumental assistance than their counterparts in more disadvantaged neighborhoods. In particular, mothers living in poor neighborhoods are disadvantaged when it comes to large amounts of financial assistance.

Our findings suggest that mothers living in disadvantaged neighborhoods may face barriers to social mobility because they lack personal and social resources that can provide substantial financial help. Lacking financial assistance of $1,000 or more may make it difficult or impossible to buy or repair a car, to put down a security deposit on an apartment or a down payment on a house, or to pay tuition for college or vocational training. Our analysis is consistent with Wilson’s discussion of restricted mobility in areas of concentrated poverty because of isolation from mainstream social networks (Wilson 1987). Whereas Wilson focused on social networks as a means of connecting to jobs, we show that those living in disadvantaged neighborhoods may also be constrained in their mobility because they lack access to economic support. These results underscore the important function performed by community banks that make small loans accessible to residents of low-income communities.

Although residents of poor neighborhoods lack access to large financial assistance, they are no less likely to have more modest financial support, housing assistance, and child care assistance than their counterparts in better neighborhoods. Thus, it seems that support networks do exist in disadvantaged neighborhoods, but these networks simply lack the means to provide large monetary assistance. These findings are consistent with studies documenting the daily support that is particularly common in low-income Black communities (Aschenbrenner, 1973; Chatters & Jayakody, 1995; Jayakody & Chatters, 1997; Stack, 1974).

We find that living in a better neighborhood is advantageous in terms of instrumental supports in the cross-section, but moving to a better neighborhood is not associated with improvements in perceived instrumental support, at least in the short term. Further, we find that residential stability is associated with more perceived instrumental support in better and worse neighborhoods alike. This pattern of findings suggests a difficult tradeoff: the advantages of moving to a better neighborhood have to be weighed against the benefits of residential stability.

Our findings on residential mobility differ somewhat from the Moving to Opportunity (MTO) study, which found that moving to a better neighborhood disrupts social ties (Kissane & Clampet-Lundquist, 2005; Pettit & McLanahan, 2003). This disparity may be at least partly due to the fact that MTO participants were encouraged to move to low-poverty neighborhoods by the offer of a housing voucher. The voucher provided an incentive that may have tipped the scales in favor of moving, even though it meant forgoing the benefits of residential stability. At the same time, some families in MTO chose not to move in spite of the voucher incentive. Our findings, documenting the benefits of residential stability and the lack of benefits of moving to a better neighborhood in the short term, may help explain why some MTO treatment families chose not to move.

A few features of our sample and measures should be kept in mind when interpreting our findings. Though monetary, housing, and child care support are all of substantial, practical importance for new mothers, these types of support are by no means exhaustive of the types of assistance mothers may need or receive from their networks. Many prior studies of perceived support focus on available emotional support, which was not available in our study (Sarason, Levine, Basham, & Sarason, 1983; Vaux & Harrison, 1985; Zimet, Dahlem, Zimet, & Farley, 1988). Future research should consider how neighborhood disadvantage is related to available emotional support or other forms of in-kind support such as transportation. We define neighborhoods according to Census tracts, but Census tracts may not correspond with how individuals define their neighborhood. Data limitations also preclude us from including an indicator of neighborhood crime, a characteristic that might influence the support one may be able to receive from friends and family members who live outside the neighborhood. Furthermore, the Fragile Families sample only includes new parents in urban areas, and results might differ for residents of rural areas, mothers of older children, or individuals without children (Belsky & Rovine, 1984).

Our research establishes a relationship between neighborhood and perceived instrumental support that is robust to individual-level controls and controls for time-invariant, unobserved characteristics, but the direction of causality remains unclear. Bad neighborhoods may inhibit the development of supportive social ties; or, lacking instrumental support may lead one to live in a more disadvantaged neighborhood or to move to a more disadvantaged neighborhood. Residential stability may increase support because the longer one lives in a neighborhood, the more time one has to get to know and enter into exchange relationships with neighbors. Alternatively, having a supportive social network may decrease one’s chances of having to or wanting to move. Our analysis cannot distinguish between these possibilities. Further, we cannot rule out the possibility that unobserved, time-varying characteristics–such as the death of a friend or family member–lead mothers to move to a worse neighborhood and to experience a decline in support. Given these methodological complexities, we agree with Entwisle (2007) on the need for more research on the process by which individuals select into particular neighborhoods.