A Longitudinal Examination of Housing Hardships Among Urban Fathers

Abstract

Stable housing is widely recognized as a prerequisite for the functioning of individuals and families. However, the housing stability of fathers is under-studied, particularly for fathers living apart from their children. This analysis measures the extent and nature of fathers’ housing insecurity using the Fragile Families and Child Wellbeing Study, a national longitudinal survey of urban families. Housing insecurity affects a substantial portion of fathers, with 25 percent experiencing insecurity at least once in their child’s first nine years. However, few fathers report persistent insecurity that spans consecutive waves. Data also indicate significant differences in rates of housing insecurity between fathers living with, and apart from, the mothers of their children, with nonresident fathers far less likely to report secure housing and more likely to experience incarceration. The nature of insecurity experienced by nonresident fathers is also qualitatively different than that experienced by their coresident counterparts.

Housing stability and circumstances are both determinants and indicators of family well-being. Housing insecurity has the potential to undermine children’s wellbeing not only when their own household experiences insecurity, but also through hardships experienced by a nonresident parent. Most nonresident parents are fathers, most of whom are involved in the lives of their children (Argys et al., 2006; Waller & Swisher, 2006). Unstable housing can create a barrier to visitation and the father-child relationship; however, little is known about the housing security of fathers, particularly those who live apart from their children.

We use population-based, longitudinal data to assess the prevalence and types of housing hardships faced by urban fathers over an eight-year period. We measure the extent, nature and trajectory of fathers’ housing insecurity over time, and differences in the nature of housing insecurity and potential socioeconomic correlates of this insecurity, experienced by coresident and nonresident fathers. We find that nonresident fathers face more and qualitatively different insecurity than coresident fathers, as well as greater socioeconomic disadvantage.

Housing Security and Family Life

Housing security and family relationships are inextricably linked. Research on family formation for lower income families is contextualized by significant economic instability (Edin & Kefalas, 2005; M. Reid & Golub, 2015), with implications for housing. Relationship changes may put housing security at risk, particularly when couples separate and leave coresidential arrangements (Moschion & van Ours, 2017; O’Flaherty, 2009). When couples with children separate, the housing security of both parents has implications for their children’s wellbeing. While fathers are rarely primary caregivers post-separation, they are still a direct input into the lives and wellbeing of their children. Most unmarried fathers maintain contact with their young children (Argys et al., 2006), and many are involved with their children’s daily activities (Waller & Swisher, 2006). A father’s ability to remain stably housed enables this ongoing parental involvement. A nonresident father in precarious housing likely faces tradeoffs between meeting his own survival needs and attending to his parental responsibilities.

Little is known about fathers’ housing conditions or the types of hardships they face. National statistics on homelessness identify homeless people as “in families” or “individuals” (National Alliance to End Homelessness, 2014). Nonresident fathers are not categorized as part of their children’s family unit even if they are involved and spend considerable time with them. To the extent that nonresident fathers are included in “individual” statistics but differ from those who are not fathers, it is difficult to measure homelessness among nonresident fathers.

Even less is known about other domains of housing insecurity which may manifest “the same underlying relationship between housing costs and housing resources” as homelessness (Honig & Filer, 1993) but are more prevalent. Researchers have measured housing hardships in various ways, including eviction, frequent moves, difficulty paying rent, mortgage, or utilities, spending more than 50% of household income on housing, living in overcrowded conditions, doubling up, living with others without paying rent, or homelessness. (Drake, Wallach et al. 1991; Phinney et al. 2007; Gilman, Kawachi, Fitzmaurice, and Buka 2003; Kushel et al. 2005; Ma, Gee, and Kushel 2008, Pavao et al. 2007). National statistics track four populations at heightened risk of homelessness: those in poverty, those unemployed, poor renter households experiencing severe housing cost burden, and poor households living doubled up (National Alliance to End Homelessness, 2014). However, these statistics capture only cross-sectional measures of a few domains of housing insecurity, and say little about fathers. Given the interdependence of fathers’ wellbeing and that of the rest of their families (Geller, Garfinkel, & Western, 2011; Tach, Mincy, & Edin, 2010; Waller & Swisher, 2006), our limited understanding of fathers’ housing insecurity represents a significant gap in the literature on family life. The current analysis advances our understanding of fathers’ maintenance or loss of secure housing using population-based, longitudinal data.

Method

Data Source

The Fragile Families and Child Wellbeing Study (FFCWS, http://www.fragilefamilies.princeton.edu/) is a national survey providing longitudinal information about 3,712 children born to unmarried parents, and 1,186 children born to married parents, between 1998 and 2000. Families were sampled from twenty U.S. cities with populations of 200,000 or more. The FFCWS systematically oversampled unmarried births, including a large sample of racial and ethnic minorities, and socioeconomically disadvantaged families. When the data are weighted they are nationally representative of urban births in the timeframe. Parents were interviewed shortly after their child’s birth, and approximately one, three, five, and nine years later (hereafter, the Y1, Y3, Y5, and Y9 waves). These data allowed us to focus on the long-term housing security of diverse urban fathers with young children.

Measuring Housing Insecurity

We constructed a categorical measure of fathers’ housing status based on several opportunities they had at each follow-up wave to indicate housing insecurity in the year leading up to the survey: We coded fathers as insecure if they were not incarcerated at the time of the survey, and reported experiencing at least one of the following hardships in the past year: skipping a rent or mortgage payment, doubling up, eviction, homelessness (having spent at least one night in the past year sleeping in a shelter, car, an abandoned building, or another place not meant for residence). Fathers were also identified as insecure due to “frequent moves” if they reported moving more than one per year over the past wave (Adam & Chase-Landsdale, 2002; Weinreb, Goldberg, Bassuk, & Perloff, 1998; Wood, Valdez, Hayashi, & Shen, 1990).

We coded fathers interviewed in jail or prison as “incarcerated”, even if they also reported housing hardships. We coded fathers as secure if they were not incarcerated and reported that they had not experienced any of the hardships above, and coded their status as unknown if no hardships were reported and at least one hardship status was unknown, or the father was not interviewed at the wave. In some analyses, we expanded the “insecure” category to separately consider the reported hardships as mutually exclusive dimensions of housing insecurity. In this expansion, fathers reporting more than one hardship at a wave were coded as experiencing “multiple” insecurities.

Measuring Parental Coresidence

We measured whether fathers were living with, or apart from, their “focal partners”, the mothers of the Fragile Families children, at each wave. Coresidence measures were based primarily on mother reports, supplemented with father reports when mothers’ were unavailable.

Correlates of Housing Security

To contextualize fathers’ housing insecurity, we measured fathers’ race/ethnicity, nativity, age, employment status, poverty status, and educational attainment at the time of the focal child’s birth. Each of these was based primarily on fathers’ self-reports, supplemented with maternal reports when needed and feasible. We identified fathers as in “deep poverty” if their household income was less than 50% of the poverty line, in “poverty” if income was 50-99% of the poverty line, “near poverty” if income was between 100% and 199% of the poverty line, and out of poverty if income was 200% of the poverty line or more. We also tracked census data on the poverty rates of the census tracts where fathers lived. Finally, among the couples coresident at baseline, we distinguished the married from unmarried couples. The FFCWS assumed married couples lived together, as a result only unmarried mothers are asked about co-residence. Marital status was based primarily on mothers’ reports, supplemented with fathers’ when needed.

Samples: Use of Weights and Measuring Attrition

We retained all 4,898 FFCWS families in our unweighted sample and the 3,442 fathers from the weighted national sample. Many fathers are not interviewed in all four waves and both parents could be lost to attrition. We retain these families in the analysis in order to measure the extent of this “unknown” status. A father’s lack of participation in the FFCWS may be tied to insecurity in his housing circumstances, suggesting that fathers with unknown status may differ substantively from a father whose status is known. We model demographic, socioeconomic, and contextual differences between fathers with known and unknown status in Appendix A.

Table 1 provides a description of our analysis sample. The two leftmost columns present unweighted statisics that reflect the FFCWS’s systematic oversample of nonmarital births. 61% of couples were coresident at the birth of their child; the majority (60%) of these were cohabiting but unmarried. Weighting the sample statistics to reflect the FFCWS sampling strategy suggests that of new urban parents with children born between 1998 and 2000, 80% are coresident at the time of the birth. Approximately three-quarters of coresident parents are married. Given the oversample of unmarried parents, who tend to be socioeconomically disadvantaged, it comes as little surprise to find less insecurity and neighborhood poverty in the weighted sample than in the unweighted FFCWS sample. Subsequent analyses are based on the weighted sample, reflecting the prevalence and types of housing insecurity experienced by urban fathers nationwide.

Table 1.

Demographic Description and Housing Insecurity Summary, FFCWS

	Raw (Unweighted) Sample (N=4,898)		Weighted Sample (N=3,442)
Baseline Residence Status	Nonresident (N=1928)	Coresident (N=2,970)	Nonresident	Coresident
Percent of Sample	39%	61%	20%	80%
Father Race
White***	7.3%	25.4%	11.9%	42.5%
Black***	66.4%	37.9%	52.6%	18.8%
Hispanic	21.7%	31.5%	27.8%	31.9%
Other	3.3%	5.2%	5.9%	6.7%
Unknown**	1.3%	0.0%	1.7%	0.0%
Father Foreign Born***	8.6%	22.5%	10.8%	22.6%
Father Age at Baseline***	26.0 years	28.8 years	25.7 years	30.7 years
	[SD=7.4]	[SD=7.0]	[SD=9.2]	[SD=6.6]
Baseline Education
< HS***	32.9%	29.8%	35.6%	21.0%
HS or GED	38.3%	32.2%	31.4%	27.4%
Some College***	16.0%	22.9%	15.4%	28.4%
College Grad***	2.9%	14.7%	3.4%	22.4%
Unknown Education***	9.8%	0.4%	14.2%	0.9%
Employed at Baseline***	59.9%	85.1%	53.6%	91.5%
BL Employment Unknown***	19.1%	1.0%	25.3%	1.0%
Married at Baseline?***	N/A	40.0%	N/A	75.2%
Father Housing Insecurity
Any Insecurity, Y1-Y9	32.7%	34.1%	25.9%	27.3%
No Insecurity, Y1-Y9***	10.7%	25.6%	10.1%	33.1%
Insec. Unknown in 1+ wave***	56.5%	40.4%	64.0%	40.0%
Father Baseline Poverty
No Poverty (200%+ of PL)***	30.2%	45.0%	24.1%	56.8%
Near Poverty (100-199% of PL)	14.1%	20.4%	13.8%	15.7%
Poverty (50-99% of PL)	8.0%	12.7%	8.2%	9.5%
Deep Poverty (0-49% of PL)*	8.7%	11.3%	11.9%	6.5%
Poverty Status Unknown***	39.0%	10.6%	42.0%	11.6%
Neighborhood Poverty
Average BL tract poverty rate***	21.6%	16.8%	21.6%	13.8%
	[SD=.135]	[SD=.133]	[SD=.139]	[SD=.136]
In high poverty tracts (BL)***	25.1%	15.1%	24.1%	12.6%

Note:

^*P<.05,

^**P<.01,

^***P<.001 in comparisons of resident and nonresident fathers in the weighted sample. In the unweighted sample, the significance of race, education, and poverty differences are more pronounced. Results may not add up to 100% due to rounding. 2 mothers did not provide baseline relationship status and are assumed to have been nonresident. Age data are unavailable for 1,066 fathers, and tract characteristics are unavailable for 832 fathers. All married couples are assumed to be coresident, leading to a 0% marriage rate among nonresident fathers. “High poverty” tracts are defined as having family poverty rates of 30% or more.

Analysis Strategy

To better understand the housing hardships experienced by urban fathers and their families, we first computed the extent of housing insecurity among Fragile Families fathers at each wave, both overall and by type. We next assessed the persistence of housing insecurity by examining within-father trajectories of insecurity across the study’s four waves, identifying the extent to which fathers experience insecurity in consecutive waves, move between insecurity and either incarceration or “insecurity unknown” status, or report insecurity in multiple waves, with and without periods of security between hardships.

We next measured differences in the prevalence of housing insecurity between fathers living with, and apart from, their focal partner, and variation in insecurity and coresidence status across waves of the FFCWS. We note both the relative prevalence of coresidence and nonresidence at each wave, and the extent to which fathers’ housing coresidence status is unknown over time (in most cases reflecting attrition from the FFCWS). We also measured the specific types of insecurity reported by coresident and nonresident fathers, respectively. We hypothesized that nonresident fathers would face more, and qualitatively different insecurity than coresident fathers.

Finally, we compared the demographic descriptors of fathers across co-residence and insecurity status using a series of regression models averaging respondent characteristics by housing and coresidence status across waves. These estimates included year fixed effects to control for secular trends that may influence housing security over time. Observed relationships are likely bidirectional, as housing insecurity both reflects, and has the potential to exacerbate, socioeconomic disadvantage. We therefore hypothesized housing-insecure fathers would be more disadvantaged than their secure counterparts across and within coresidence status.

Results

Housing Insecurity Prevalence and Trajectories

Housing insecurity is prevalent among urban fathers in their children’s early and middle childhood. As shown in Table 1, our weighted sample suggests that more than one quarter of urban fathers report experiencing some form of housing insecurity in their child’s first nine years. Table 2 presents rates of fathers’ housing insecurity at each follow-up wave of the FFCWS, and indicates that between 7 and 13% of respondents indicate having recently experienced some form of housing insecurity at each wave. If fathers with missing status experience housing insecurity at the same rates as observed fathers, overall rates of insecurity would range from 10 to 21 percent. To the extent that missing fathers are more insecure than those observed, the true rates would be even higher. Approximately 1% are incarcerated, with reported incarceration declining at Y9 as in-facility interviewing stops.^.

Table 2.

Insecurity Prevalence by Wave

Insecurity Status	Y1	Y3	Y5	Y9
No Insecurity	64.42	63.47	57.73	47.35
Any Insecurity	7.15	9.01	8.40	12.61
Father Incarcerated	1.22	1.18	1.59	0.01
Insecurity Unknown	27.21	26.31	32.29	40.03
Insecurity Type
Homeless in past year	0.36	0.20	0.13	0.16
Doubled up in past year	0.71	1.90	0.67	2.49
Skipped rent or mortgage payment in past year	0.80	4.47	4.90	6.90
Eviction in past year	0.01	0.05	0.02	0.10
Frequent moves	3.85	0.99	0.78	0.38
Multiple Insecurities	1.42	1.43	1.89	2.57
Father Incarcerated	1.22	1.18	1.59	0.01
Status Unknown	27.21	26.31	32.29	40.03
N	3,442	3,442	3,442	3,442

Note: Descriptive statistics are weighted to be representative of fathers with children born in large cities between 1998 and 2000.

The nature of fathers’ insecurity shifts throughout the life course of the focal child. At Y1, the most prevalent insecurity reported is the experience of frequent moves (more than one in the year). These moves are likely to reflect changes in family structure immediately following the focal child’s birth, and are rarely reported in subsequent waves. In subsequent waves, the most prevalent insecurity reported is having skipped a rent or mortgage payment in the past year, with “doubling up” a distant second. Eviction and homelessness are exceedingly rare, reported by less than 1% of fathers. Notably, reported housing insecurity is significantly more prevalent in Y9 than in earlier waves; this likely reflects changing macroeconomic conditions and the Great Recession of 2007-2010, as well as changes in fathers’ individual and family circumstances.

Examining fathers’ housing status over time underscores the challenges of tracking fathers over eight years. As shown in Table 3, more than half of fathers have unknown housing status at one or more measurement period, with over 10% of fathers with unknown housing status at every follow-up wave. Of fathers whose housing status is consistently observed, nearly two-thirds are always secure (29% of the full sample). Of the full sample, another 9% are secure in three out of four waves, and 30% have periods of unknown housing status, but are secure in all waves their status is observed. These three groups comprise more than two-thirds of fathers.

Table 3.

Trajectories of Housing Security Over Time

Summary of Trajectory	Prevalence Among Analysis Sample
Always secure	29%
Secure whenever known	30%
Secure 3 waves, Insecure 1 wave	9%
Secure at least once, but insecure at least twice in a row	4%
Always insecure, incarcerated, or unknown (but not always insecure)	4%
Always insecure	<1%
Some other arrangement	12%
Always unknown	12%
N	3,442

Note: Summary statistics weighted to be nationally representative of fathers with children born in large cities from 1998-2000. Housing trajectories are measured over four FFCWS survey waves (Y1, Y3, Y5, Y9)

Relatively few fathers report housing insecurity in consecutive waves. Fewer than 1% are insecure in all waves, 4% fail to report secure housing at any point in the survey (instead having a mix of insecurity, incarceration, and unknown status), and 4% report secure housing at least once but report at least two periods of insecurity in a row. The remaining 12% of fathers report some other trajectory of housing conditions mixing security, insecurity, and incarceration. This group contains 50 different trajectories, none of which stand out as more prominent than the others. Housing insecurity does not appear to be clustered among a few chronically insecure individuals. Rather, Table 3 suggests housing insecurity is often experienced episodically, by a broader group of fathers who largely report insecurity in a single wave at a time.

Housing Security and Fathers’ Relationship Status

Differences in fathers’ housing security by relationship status are presented in Figure 1. These pie charts illustrate the relative prevalence of fathers’ coresidence and nonresidence, housing insecurity for both groups, and changes in these statuses over time. Each row illustrates differences between coresident and nonresident fathers in a given wave, and each column illustrates changes over time among fathers who are, respectively, coresident, nonresident, and those who have unknown coresidence status.

Figure 1. Fathers’ Housing Security by Coresidence Status and Survey Wave.

Note: At each wave, multinomial logistic regression models predicting insecurity status (insecure, not insecure, incarcerated, or unknown) indicate that nonresident fathers are significantly more likely (P<.001) than their coresident counterparts to be insecure or have unknown insecurity status. Nonresident fathers are significantly more likely to be incarcerated at the Y1, Y3, and Y5 waves, with no significant differences in incarceration at Y9.

Each pie is sized to reflect the number of fathers who are coresident or non-resident at each wave with the survey focal child and mother. The diminishing size of the pies in the “coresident” column and increasing size of those in the “nonresident” column reflects the diminishing proportion of fathers residing with the focal child and mother from birth to nine years; the increasing size of the “residence unknown” charts reflects attrition of both resident and nonresident fathers over the course of the survey. Within coresidence status, the prevalence of insecurity is relatively consistent over time; more than three-fourths of coresident fathers report secure housing, with fewer than 10% reporting housing insecurity in the first three follow-up waves, and a peak of 16% reporting insecurity at Y9. “Unknown status” is relatively rare (approximately 10% of coresident fathers have unknown status at each wave). Reports of incarceration among the co-resident are quite low. Among coresident fathers less than .5% are reported as incarcerated at Y1, Y3, and Y5.

The housing circumstances of nonresident fathers are considerably less stable. The percent of nonresident fathers reporting insecurity is only slightly greater than the percent of coresident fathers reporting insecurity (10%-17% across waves). However, more than 40% of nonresident fathers have unknown housing status in each wave. Many of these fathers are not interviewed in a wave, or have left the FFCWS entirely. If these “unobserved fathers” experience housing insecurity at rates comparable to nonresident fathers whose insecurity status is known, insecurity rates approach 20% in Y3 and Y5, and nearly 30% in Y1 and Y9. In Y1, Y3, and Y5, between 4% and 5% percent of nonresident fathers are reported as incarcerated. Incarcerated fathers are particularly vulnerable to housing insecurity (Geller & Curtis, 2011), suggesting that these fathers may face an elevated risk upon their release.

In addition to the observed differences in rates of housing security across fathers’ coresidence status, coresident and nonresident fathers’ housing insecurity experiences also differ in kind. Figure 2 presents the distribution of insecurity types experienced by coresidence status over time, and shows that coresident fathers experiencing housing insecurity most often experience this insecurity in the form of a missed rent or mortgage payment. One exception is the Y1 wave, in which 80% of coresident housing-insecure fathers experience insecurity in the form of frequent residential moves (multiple moves in the year since the child’s birth).

Figure 2.

Fathers’ Insecurity Types by Coresidence Status and Survey Wave (Known Status Only)

Nonresident fathers experience a different, more diverse set of insecurities. While skipping a rent or mortgage payment varies in prominence across waves, it is consistently less prominent among nonresident fathers than coresident insecure fathers. Doubling up, a more precarious housing situation, is more common among the nonresident insecure fathers. In addition, more than 20% of nonresident insecure fathers report multiple insecurities at each wave. In sum, nonresident fathers are not only more likely to report insecurity, the kind of insecurity that they report is often more severe than that reported by coresident fathers.

Estimated demographic and socioeconomic differences across fathers’ coresidence and insecurity status are provided in Table 4. Notably, the differences between secure and insecure fathers within coresidence status (i.e., secure vs. insecure coresident, and secure vs. insecure nonresident) are less salient than differences between coresidence status. Coresident fathers are significantly more likely to be white, and less likely to be black, than nonresident fathers, regardless of housing security status. They are also significantly older and more educated than nonresident fathers, underscoring the social disadvantage nonresident fathers face.

Table 4.

Demographic Description of Secure and Insecure Fathers, by Coresidence Status

Row Labels	Coresident		Nonresident
	Insecure	Secure	Insecure	Secure
% White	35%	49%	30%	26%
% Black	23%*	16%	46%	45%
% Hisp	39%**	30%	18%	24%
% Other Race	3%	6%	6%	4%
% Unknown Race	0%	0%	0%	<1%
	29.7***	31.2	26.7*	28.9
Age	[0.3]	[0.3]	[0.3]	[0.3]
% < HS	26%*	17%	39%	32%
% HS Only	37%	21%	39%	37%
% Some College	30%	33%	17%	21%
% College Grad	6%	29%	4%	8%
% Education Unknown	<1%	<1%	1%	2%
% Married at BL	62%**	79%	29%	34%
% Employed at BL	90%	94%	81%	78%
% in High Poverty Tracts	15%	10%	17%	19%

Note: Entries are stratified average predicted values from regression models predicting each characteristic with fathers’ coresidence and housing security status at each wave, and wave FE.

^*P<.05,

^**P<.01,

^***P<.001 in within-coresidence comparisons by security. Co-residence differences in race (black/white), age, and education are statistically significant at P<.05 or less.

Within coresidence status, coresident fathers experiencing insecurity are more disadvantaged than their secure counterparts. Insecure coresident fathers are significantly younger than their secure counterparts, less likely to have completed high school, more likely to be black and Hispanic, and more likely to have been unmarried at baseline. Among nonresident fathers, the demographic differences between the housing-secure and housing-insecure are less pronounced. The housing-insecure are significantly younger than their more secure counterparts; differences in racial composition, education, and marital status are statistically insignificant.

CONCLUSION

Summary of Findings

While housing insecurity touches the lives of many fathers and families, a comprehensive understanding of this insecurity remains elusive. Between seven and twelve percent of fathers are reported as insecure in any given survey wave, and more than a quarter have unknown insecurity status at each wave. Given that fathers with unknown status are more disadvantaged than observed fathers, overall rates of insecurity are likely even higher. However, few fathers report insecurity in multiple waves, suggesting that insecurity is distributed broadly across fathers, rather than concentrated within a small group who are consistently insecure. More than 25% of fathers report housing insecurity at least once.

The insecurity observed among urban fathers differs between coresident and nonresident fathers. Most coresident fathers report secure housing at each wave, and after Y1, the majority of coresident fathers experiencing housing insecurity report missing a rent or mortgage payment. Missing a rent or mortgage payment has the potential to compromise family wellbeing and suggests rather broadly distributed economic distress among urban fathers of younger children. The housing status of nonresident fathers is considerably more precarious with a notable proportion reporting doubling up and frequent residential moves. Additionally, the more than, 40% of non-resident fathers with unknown status at each wave suggests a significant underreport of housing hardships among these fathers.

Limitations

The current analysis is limited by significant amounts of missing data. Survey response rates tend to be lower among socioeconomically disadvantaged individuals (Groves, 2004), and relying on reports of respondents that are easiest to observe has been shown to attenuate estimates of population disadvantage (Western, Braga, Hureau, & Sirois, 2016). To estimate the extent to which this is the case in our analysis, we modeled the probability that a given person-wave observation is observed (rather than missing), using logistic regression models based on family characteristics observed at baseline (i.e., when most fathers are interviewed and all mothers are interviewed). Results, presented in Appendix A, indicate that observation of fathers’ housing status was associated with their demographic background, baseline socioeconomic status and human capital. These results suggest that the rates of housing insecurity observed in our analysis sample are likely to be conservative estimates of the true population rates. One reason fathers may not participate in a survey wave is because they have moved from their last known address and cannot be located by the research team. However, the magnitude of the disparity between “missing fathers” and observed fathers is unclear, as are the types of insecurity experienced by fathers whose status is unknown.

Our analysis does not address the potential causal relationships between fathers’ coresidence and housing security statuses. Our findings in Table 4 indicate that nonresident fathers tend to be more socioeconomically disadvantaged than their coresident counterparts, regardless of housing security status. Causal modeling to illuminate how coresidence (or nonresidence) and any associated disadvantage influences housing security would be a fruitful area for research. The current analysis does not address the role of social services in protecting against housing insecurity, or connections of fathers’ housing security to their relationships with their children. Housing insecurity is likely to adversely affect fathers’ ability to engage in parenting activities. The prevalence and severity of hardship among nonresident fathers suggests that housing insecurity may further complicate their own well-being and parental capacity. Future research is needed to understand the implications of housing insecurity for father-child contact and child wellbeing. Finally, the FFCWS sample was selected based on the birth of one “focal” child, but many of the fathers have additional children, some with women beyond their focal partner. Future work that considers fathers’ housing in the context of multipartner fertility and family complexity would enhance the literature.

Implications

Our work identifies housing as an important domain of social disadvantage facing urban fathers. Among the most advantaged fathers, difficulty with housing payments may introduce considerable stress into family functioning. This suggests attention to families housing circumstances, particularly in tight urban markets, is warranted. Among less advantaged, non-resident fathers, housing hardships are particularly pronounced and introduce considerable challenges for basic family functioning. Non-custodial parents have limited access to social welfare supports, housing subsidies and limited access to work supplements like Earned Income Tax Credit with small cash benefits for non-custodial parents (Berger, 2017). This research contributes to the evidence base linking father’s housing circumstances to the literature on families. Further, we identify non-custodial parents as a population of policy interest, whose housing security and broader wellbeing must be considered and could be strengthened through attention to existing social welfare policy levers.

Appendix A: Modeling Inclusion in the Analysis Sample

To assess the probability that fathers were included in our analysis sample, we estimated two logistic regression models, presented below. Our first model indicates that fathers’ participation in the baseline survey (compared to inclusion in the FFCWS due to partner participation at baseline) is the dominant predictor of their reporting housing status in follow-up waves. We therefore estimated a second model that limited our sample to those fathers observed at baseline. Results are presented in Table A.1. In both models, we see that fathers who were unmarried (both cohabiting and nonresident) at the time of the focal child’s birth were less likely to be observed, as were black, Hispanic, and “other race” fathers. Education is also associated with fathers’ probability of being observed, with high school graduate fathers more likely to be observed than those with less than a high school degree, and less likely to be observed than those completing some college. Fathers reporting employment at baseline are more likely to be observed than those unemployed. Finally, the coefficients on wave fixed effects suggest that fathers are less likely to be observed with each passing wave, reflecting attrition from the FFCWS.

Table A.1.

Inclusion Model, Predicting Fathers Status Observed

	Model 1			Model 2
	Coeff.	SE		Coeff.	SE
Parents’ Baseline Relationship Status (Reference = Married)
Cohabiting	−0.265	[0.071]	***	−0.296	[0.080]	***
Nonresident	−0.417	[0.077]	***	−0.481	[0.087]	***
Father Interviewed at Y1	1.439	[0.080]	***
Father’s Baseline Poverty Status (Reference = 200% of Poverty Line or more)
Near Poverty (100-199% of PL)	0.025	[0.070]		0.041	[0.072]
Poverty (50-99% of PL)	−0.034	[0.083]		−0.013	[0.085]
Deep Poverty (0-49% of PL)	−0.137	[0.089]		−0.113	[0.092]
Father’s Baseline Education (Reference = HS Grad or GED)
< HS	−0.123	[0.059]	*	−0.123	[0.066]	+
Some College	0.149	[0.069]	*	0.182	[0.079]	*
College Graduate	0.101	[0.102]		0.135	[0.118]
Education Unknown	−0.233	[0.173]
Father’s Race (Reference = Non-Hispanic White)
Non-Hispanic Black	−0.157	[0.076]	*	−0.204	[0.088]	*
Hispanic	−0.376	[0.079]	***	−0.426	[0.090]	***
Other Race	−0.49	[0.123]	***	−0.473	[0.143]	***
Race Unknown	−0.345	[0.446]
Father’s Baseline Tract Poverty (Reference: Father in “Low Poverty Tract”)
High Poverty Tract at Baseline?	0.078	[0.068]		0.109	[0.074]
Baseline Tract Poverty Unknown	−0.539	[0.087]	***	−0.509	[0.121]	***
Baseline Employment	0.151	[0.064]	*	0.166	[0.069]	*
Baseline Employment Unknown	−0.188	[0.138]		−1.4	[0.819]	+
Survey Year (Reference = Y1)
Y3	−0.097	[0.035]	**	−0.285	[0.044]	***
Y5	−0.255	[0.039]	***	−0.536	[0.047]	***
Y9	−0.747	[0.040]	***	−1.067	[0.046]	***
Constant	0.235	[0.127]	+	1.923	[0.122]	***
N	19592			15320