Children and Youth Services Review 34 (2012) 14401447

Contents lists available at SciVerse ScienceDirect

Children and Youth Services Review

journal homepage: www.elsevier.com/locate/childyouth

The role of supportive housing in homeless children's well-being: An investigation of child welfare and educational outcomes
Saahoon Hong , Kristy Piescher
Center for Advanced Studies in Child Welfare, School of Social Work, University of Minnesota, United States

a r t i c l e

i n f o

a b s t r a c t
Research has demonstrated that homeless children have disproportionate negative academic experiences, including absenteeism, high rates of mobility, grade repetition, and the need for special education services, which may all contribute to poor academic performance. Homeless children are also more exposed to violence and social isolation due to their often dangerous living environments, past histories of victimization, and trauma experienced by their mothers. The main purpose of this study was to investigate the impact of family supportive housing service receipt on children's well-being, including the academic functioning and child protection involvement of homeless children. A total of 183 children's supportive housing records were sequentially linked to data from the Minnesota Departments of Education and Human Services, including the Minnesota Automated Reporting Student System, MCA-II database, and Social Services Information System through Minn-LInK. Generalized Estimating Equation analysis was implemented to examine the three-year longitudinal role of supportive housing for homeless children's educational and child protection outcomes. Signicantly positive effects of recipients of supportive housing services were found in school mobility, school attendance, and math achievement. The proportion of children with child protection involvement for the supportive housing group sharply decreased over time. Recommendations for policy and future research are made; study limitations are addressed. 2012 Elsevier Ltd. All rights reserved.

Article history: Received 17 February 2012 Received in revised form 28 March 2012 Accepted 29 March 2012 Available online 5 April 2012 Keywords: Supportive housing Homeless children Child welfare Child protection School outcomes Education

1. Introduction Homelessness is a critical social challenge for the United States. The U.S. Department of Housing and Urban Development (HUD) (2010) estimated that over 650,000 individuals experienced homelessness in 2010, of which nearly 20% experienced long-term homelessness (NAEH, 2011). However, the recent downturn of the economic climate has created a shift in the composition of the homeless population over time. Although the overall number of homeless individuals has remained fairly stable over the past few years, family homelessness increased by approximately 7% between 2008 and 2009 and 30% from 2007 to 2009 (HUD, 2010). Currently, children comprise 22% of the homeless population in the United States on any given night (HUD, 2010). The challenge of homelessness is evident for Minnesota as well. The Wilder Research Center's survey of Minnesota's homeless families in 2009 noted that family homelessness has risen to its highest
This project was funded by the Robins, Kaplan, Miller and Ciresi Foundation. The authors thank Richard A. Hooks Wayman and Benjamin Van Hunnik at Hearth Connection for their support of this study through data provision and interpretation assistance. Corresponding author at: Center for Advanced Studies in Child Welfare, College of Education and Human Development, University of Minnesota, 217 Peters Hall, 1404 Gortner Avenue, St. Paul, MN 55108, United States. E-mail address: hong0083@umn.edu (S. Hong). 0190-7409/$ see front matter 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.childyouth.2012.03.025

level since 1991 (Wilder Research, 2011). In October of 2009, on the night of the survey, 1455 families were sleeping in emergency shelters or transitional housing options (including 2857 children age 17 and younger) and another 200 families (with 394 children) were found sleeping outside or in cars. Disturbingly, the Wilder Research Center concluded that homeless families and children have made up the greatest segment of homeless people in Minnesota since the statewide homeless survey began in 1991. Furthermore, the number of children impacted by homelessness has not always been apparent by those who access emergency centers and shelters. During the survey, Wilder Research Center found that 50% of homeless adults were parents of children age 17 or younger, yet only 30% had a child with them. It is clear that relatives and guardians are keeping care of children during housing crises for families, which contributes to unstable home life and mobility between residences and schools. Analysis of multiple research studies result in the following consistent ndings (Rog & Buckner, 2007): 1) the most common prole of a homeless family is one headed by a single woman in her late 20s with approximately two children, one or both under six years of age; 2) those at greatest risk belong to ethnic minority groups; 3) The residential histories of homeless families typically reveal high mobility and instability, including living in a variety of doubled up and own housing arrangements; 3) family separation (removal of child) are common occurrences, both before and after the homelessness episode; 4) homeless families are typically extremely poor, and most who are

S. Hong, K. Piescher / Children and Youth Services Review 34 (2012) 14401447

1441

homeless lack human capital, like useful skills and abilities, with respect to both education and employment; 5) conict, trauma, and violence gure prominently in the lives of homeless families, as they do with equally poor but domiciled families; and 6) reports of substance abuse, though likely underestimated, are higher for mothers who are homeless than for other woman in poor families, but lower than for single adults who are homeless. 1.1. Impacts on children The negative effects of homelessness on children are well documented. Children who experience homelessness also experience a myriad of negative encounters with educational and child welfare systems. Research has demonstrated that homeless children have disproportionate negative academic experiences, including absenteeism (Larson & Meehan, 2009; Rafferty & Rollins, 1989; Rubin et al., 1996; Zima, Wells, & Freeman, 1994), high rates of mobility (Buckner, Bassuk, & Weinreb, 2001; Masten, Miliotis, Graham-Bermann, Ramirez, & Neemann, 1993; Rafferty, Shinn, & Weitzman, 2004), grade repetition (Buckner et al., 2001; Masten et al., 1997; Rafferty et al., 2004; Rubin et al., 1996), and the need for special education services (Masten et al., 1997), which may all contribute to poor academic performance (Larson & Meehan, 2009; Masten et al., 1997; Obradovic et al., 2009; Rafferty et al., 2004; Rubin et al., 1996; Shinn et al., 2008; Zima et al., 1994). In addition, they are more exposed to violence and social isolation due to their often dangerous living environments, past histories of victimization, and trauma experienced by their mothers. All these factors increase homeless families' probability of involvement with the child welfare system (Anooshian, 2005). The little research that examines homeless families' contact with child welfare indicates that homeless families have higher rates of child protection involvement than non-homeless families in terms of receipt of child protection services (Culhane, Webb, Grim, Metraux, & Culhane, 2003; Dufeld & Lovell, 2008; Park, Metraux, Brodbar, & Culhane, 2004), substantiations of maltreatment, and out-of-home placements (Larson & Meehan, 2011; Masten et al., 1993; Wilder Research, 2010; Zima et al., 1994). It may be important to note that homelessness and its negative impact on children are not equitably dispersed among all cultural and ethnic communities. In Minnesota, African American and American Indian children are more likely to experience homelessness than their White peers. According to the Wilder Research Center's 2009 Survey of Homeless Children and Families, Although American Indians are only one percent of all Minnesota parents, they are 11% of homeless parents. African Americans are four percent of Minnesota parents but 48 percent of homeless parents in Minnesota. 1.2. Availability of services to homeless families A myriad of services are available to assist families struggling with homelessness. Available services follow a continuum of care, ranging from emergency shelters to transitional housing to permanent supportive housing. Emergency and transitional housing are timelimited programs, while permanent supportive housing programs do not have a specied time limit. Supportive housing programs focus on those households with signicant barriers (e.g., health, disabilities, history of abuse, and violence) to housing stability and long histories of homelessness. The former programs rely on families moving on to subsidized or unsubsidized permanent housing. The latter programs help people move into mainstream permanent housing by offering permanent housing subsidies coupled with services. Families and children perhaps most at risk of experiencing negative impacts of homelessness are families that experience long-term homelessness. These families often nd their way into more intensive programs of support, such as supportive housing programs. In the supportive housing model, families are offered social services in conjunction with housing, such as services as job training, life skills

training, alcohol and drug abuse programs, and case management, for example. Supportive housing encompasses a range of approaches including single sites (housing developments or apartment buildings in which units are designated as supportive housing) or scattered site programs in which participants often use rent subsidies to obtain housing from private landlord and supportive services may be provided through home visits. Services in supportive housing are exible and primarily focused on the outcome of housing stability. By improving housing stability, the negative impacts of homelessness on families and children may be mitigated. Research has shown that coupling permanent housing with supportive services is highly effective at maintaining housing stability; it also helps improve health outcomes and decreases the use of publicly-funded institutions (Harburger & White, 2004). 1.3. Reason for the study It is clear that the homeless population is changing; the population is shifting to include a larger proportion of children; and children are negatively impacted by homelessness. However, little is known about children's experiences of homelessness as it relates to child well-being over time. Even less is known about the impact of housing support on child outcomes, as most research on the effect of housing support focuses on adult (e.g., employment) or family-level outcomes (e.g., housing stability). The current understanding of homeless children's encounters with educational and child welfare systems is mainly driven by studies that explore the experience or state of homelessness, rather than change in academic experience and child welfare outcomes longitudinally. More importantly, we know little about how supportive housing services work to keep children safe and produce the best educational outcomes for children over time. Thus, this study sought to better understand the impact of supportive housing services on homeless children's well-being over time. This study's longitudinal design reects calls for more rigorous evaluation of outcomes for homeless children, such as the U.S. Department of Health and Human Services' call for further longitudinal research in the 2007 National Symposium on Homelessness Research. The report dedicated to homeless families and children concluded that most studies to date, with a few recent exceptions, have had cross-sectional designs. Longitudinal studies are needed to explore the course of residential instability and homelessness over several years, and the individual, contextual, and intervention factors that inuence this course. Research conducted to date on children who are homeless has illuminated a fair amount of knowledge on current needs and the impact of homelessness. The main purpose of the current study was to investigate the impact of family supportive housing service receipt on children's wellbeing, including child protection involvement and the academic functioning of homeless children. Specically, the following questions were investigated: 1) Does receipt of supportive housing services impact children's school mobility, attendance rates, academic achievement, and rates of Individualized Education Plans? 2) Does receipt of supportive housing services reduce child protection involvement over time? 2. Method 2.1. Participants This study relied upon secondary data from the Minn-LInK project a project that utilizes statewide administrative data from multiple agencies, including the Minnesota Departments of Human Services, Education, and Public Health, to answer questions about the impacts of policies, programs, and practice on the well-being of children in

1442

S. Hong, K. Piescher / Children and Youth Services Review 34 (2012) 14401447

Minnesota. For this study, data from the Minnesota Departments of Education and Human Services were used in accordance with data sharing agreements between the Minn-LInK Project at the University of Minnesota and these State agencies. Data-sharing agreements allowed the use of identied data for the purpose of conducting research on families and children. The University's Institutional Review Board approved the use of this secondary data for these purposes, and all identiers were removed from the data le once cross-system matching was achieved (de-identication). 2.1.1. Supportive housing groups Data from a metropolitan supportive housing provider in Minnesota was used in the current study to identify children whose families received supportive housing services. Data included information about school age children's families who were enrolled in supportive housing services (n= 183) and permanently housed in the 20062007 academic year (n= 180). Four Supportive Housing Cohorts were developed using this sample year. The cohorts consisted of children who were in grades 3, 4, 5, or 6 during the 200607 school year and whose enrollment in supportive housing services began in 2007 (see Fig. 1). A cohort structure was chosen to allow for analysis of a relatively homogeneous group of children across all chosen indicators of well-being. Children's supportive housing records were sequentially linked to data from the Minnesota Departments of Education and Human Services, including the Minnesota Automated Reporting Student System (MARSS), MCA II database, and Social Services Information System (SSIS) via Minn-LInK. Registry Plus Link Plus (NCCDPHP, 2010), a probabilistic record matching software developed for matching cancer registry records at the Centers for Disease Control (CDC), was utilized for data linking purposes. The use of LinK Plus resulted in a match rate of 90% of all children receiving supportive housing services to Minnesota educational records. 2.2. Comparison groups To examine the effect of supportive housing service receipt on school outcomes such as attendance, school mobility, and student achievement levels in reading and math, comparison groups were developed using the Homeless Student Flag in MARSS. The federal McKinney Vento Act (U.S. Department of Education, 2004) requires school districts to identify and meet the needs of children and youth who are homeless. The Act denes a homeless student as a student who lacks a xed, regular and adequate nighttime residence or shares the housing of other persons due to loss of housing, economic hardship, or a similar reason. Children in the comparison group were

identied as homeless children (in both 2008 and 2009) not receiving supportive housing services. It is important to note that it is unknown whether children in the comparison groups were homeless in 2007 (at Time 1) because this information has only been included in MARSS since 2008. It is also unknown whether these children's families received other types of housing supports in the community (e.g., vouchers). Some comparison group children's families may have received no housing supports while other children's families may have received a wide variety of housing supports, though children included in the comparison groups maintained a homeless status in educational records throughout the entirety of the study. The use of this type of comparison group allows for evaluations of supportive housing service receipt versus all other options while still maintaining a homeless status in the comparison group. Children in the comparison group identied as homeless in both 2008 and 2009, and were categorized as comparison groups by grades. Descriptive analysis was rst used to describe differences between the supportive housing and comparison groups. Table 1 provides descriptive information about the supportive housing and comparison groups in Year 1 (2007). As can be seen in Table 1, the supportive housing group consisted of a slightly higher proportion of American Indian and White children, and a slightly lower proportion of Black children than the comparison Group. In addition, the supportive housing group consisted of a slightly older population (more middle-school and fewer elementary school children) than did the comparison group. (Statistical testing of differences between was not warranted due to the small number of children in each cell.) 2.3. Outcome measures 2.3.1. School mobility School mobility is an important indicator in predicting academic attendance and achievement, as children who have high mobility may miss school and or educational content with each move. Therefore school mobility was calculated using the Status End code in MARSS, which includes a transfer indicator (MDE, 2011). By using a transfer code, school mobility can be dened as the result of a school change (to another public or non-public school within, or outside of, the same district). School mobility may or may not coincide with residential mobility. The total number of school transfers was calculated for each student in a given school year. 2.3.2. Attendance School attendance contributes signicantly to achievement and educational attainment. Attendance, or lack thereof, is also closely

Number Grade 1
2006 2% 2005 12% 2004 1% 2003 5% 2010 6% 2009 6% 2008 9% 2007 59%

15 13 19 20 18 13 19 19 10 6 2 0 183

Grade 2 Cohort 1 Cohort 2 Cohort 3 Cohort 4 Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 Grade 9 Grade 10 Grade 11 Grade 12 Total

Fig. 1. Number and grade of children in supportive housing cohorts, based on 2007 enrollment in supportive housing services (n = 183).

S. Hong, K. Piescher / Children and Youth Services Review 34 (2012) 14401447 Table 1 Characteristics of the supportive housing total (N = 313), supportive housing cohort groups (N = 70), and comparison group (n = 342). Supportive housing group Total N Gender Male Female Ethnicity American Indian Asian/Pacic Islander Hispanic Black White Grade Pre-K Grade 1 Grade 2 Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 Grade 9 Grade 10 Grade 11 Grade 12 Grade Repetition IEP *p b .05. 157 156 41 12 12 158 90 34 25 18 35 32 36 29 30 32 19 11 3 0 11 49 Percent 50.2 49.8 13.1 3.8 3.8 50.5 28.8 1.0 8.0 5.8 11.2 10.2 11.5 9.3 9.6 10.2 6.1 3.5 1.0 0.0 3.5 15.7 Supportive housing group Cohort groups N 36 34 8 3 2 42 15 Percent 51.4 48.6 11.4 4.3 2.9 60 21.4 Comparison group Total N 168 174 29 16 11 220 66 Percent 49.1 50.9 8.5 4.7 3.2 64.3 19.3

1443

19 20 18 13

27.1 28.6 25.7 18.6

89 83 95 75

26 24.3 27.8 21.9

placement. Reports made to local child protective service agencies are rst screened to determine whether they meet the criteria to be assigned for a child protection response. Once a report is accepted, it is assigned to one of two response types Family Investigation or Family Assessment (Differential Response). Reports of child maltreatment that allege substantial child endangerment must receive an investigation. Depending on the circumstances of a report, the local child protection agency may also decide to assign a report not involving substantial child endangerment for an investigation. Reports that do not allege substantial child endangerment may receive a Family Assessment, which is the preferred response to reports not alleging substantial child endangerment. The total number of reports, whether Family Investigation or Family Assessment, was summed to calculate the number of reports in which a child was involved. It is also important to note that children in out-of-home placement may have been placed there for reasons other than maltreatment (e.g., via juvenile justice) since most children who were involved in an investigation/assessment of child maltreatment did not enter outof-home placement. All out-of-home placements were summed for this analysis, regardless of whether they originated in child protection or elsewhere. 2.4. Analysis All supportive housing cohorts received supportive housing services from September 2006 to August 2007, when they were in grades 36. Those 3rd, 4th, 5th, and 6th graders were tracked for three years. During the 200809 school year, they were respectively 5th, 6th, 7th, and 8th graders. Child well-being indicators in 2007 were considered independent variables in the analysis, while 2008 and 2009 indicators served as dependent variables. Two-year data sets were analyzed using Generalized Estimating Equations (GEE). The GEE approach is a statistical method to analyze longitudinal data, especially when dependent variables are binary or continuous (Liang & Zeger, 1986). This analysis was specically chosen to allow for the investigation of changes in outcomes over time in response to receipt of supportive housing services. GEE with continuous variable response was administered for the school mobility and attendance rate variables, while GEE with binary variable response was conducted for the MCA II achievement level and special education status variables. In other words, GEE was used to determine whether the supportive housing service group and its comparison groups child well-being outcomes were changing at different rates over the two-year investigation period while controlling for child well-being in 2007 (Time 1). In the GEE model, a response variable Y could be either continuous or categorical where Y = (Yij), response for each subject i, was measured at two different time points, j = 1, 2. A GEE model with continuous responses is below,   g ij ij 0 1 Preij 2 X ij : The term, g(), Pre, and X , is the link function, a baseline data, and a group variable, respectively, where (ij) = 1 and V (yij) = . A GEE model with binary responses looks like below,  3 2  # P Y ij 1 ij  5 0 1 Preij 2 X ij ; log log 4 1 ij 1P Y 1 "
ij

2 22

2.9 31.4

10 80

2.9 23.4

associated with involvement in child welfare, as children, aged ve to 11, who miss more than the allotted seven unexcused absences in Minnesota are required to be reported to child welfare (Maltreatment of Minors Act, 1993). Within MARSS, the attendance rate for each student was derived by totaling the Average Daily Attendance (ADA; the days the student actually attended) and dividing it by the total Average Daily Membership (ADM; the required days of enrollment) for each student. Use of this ratio as opposed to another measure of attendance allowed for comparisons of children across school districts whose school year lengths vary in Minnesota. The attendance ratio could range from .01 (very low, or almost no attendance) to 1.0 (perfect attendance). 2.3.3. MCA II: student achievement levels A student's achievement level on the MCA-II falls into one of four categories: Does Not Meet Standards, Partially Meets Standards, Meets Standards, and Exceeds Standards. Among them, the Meets Standards and Exceeds Standards are considered procient. In this study, student achievement was recoded into a binary prociency variable (1 = procient, 0 = not procient) and considered as a key outcome. 2.3.4. Individualized education plan (IEP) An IEP is a written commitment of resources and a management tool that enables children with disabilities to receive needed special education and related services in a way that is appropriate to their unique learning needs (IDEA, 1997). In MARSS, the Special Education Evaluation Status code (1 = yes, 0 = no) was used to identify children receiving special education services via an IEP. 2.3.5. Child protection involvement Three indicators were developed to measure child protection involvement for 2007, 2008, and 2009: 1) involvement in a child protection report for a specied year (coded as a binary variable, 1 = yes, 0 = no), 2) the number of accepted reports in which a child was involved, and 3) the number of children experiencing an out-of-home

where variance specication for binary responses is (ij) = ij (1-ij) and = 1. IBM SPSS Statistics 19 was mainly used to conduct this analysis. In the results section, a signicant association or signicant p-value indicated that the two-year child well-being outcomes were dependent on the receipt of supportive housing services (see Table 3).

1444

S. Hong, K. Piescher / Children and Youth Services Review 34 (2012) 14401447 Table 3 Regression estimates from GEE models of group differences. B SE 95% CI Lower School mobility
a

3. Results Mean levels of child well-being indicators under investigation and results of GEE analyses are presented in Tables 2 and 3, respectively. Changes in school mobility of the supportive housing and comparison groups were examined to identify potential relationships between receipt of supportive housing services and school mobility. Table 2 provides average school mobility between supportive housing and comparison cohorts over time. In general, results revealed that school mobility rates were decreasing in supportive housing cohorts over time while their comparison cohorts were increasing or uctuating. The results of GEE analysis conrmed this nding for Cohort 3 and revealed that this supportive housing cohort's school mobility rate over time was signicantly different from its comparison cohort ( = 0.405, 95% CI = 0.755, 0.055, p b .01; see Table 3). Average attendance for each of the four cohort groups was calculated and examined over a period of three years. Table 2 summarizes student attendance rates across the supportive housing and comparison cohorts over time. Results revealed that the supportive housing cohorts' attendance rates were generally higher than those of the comparison cohorts. Results of GEE analysis revealed the supportive housing group's attendance rate over time was signicantly different from its comparison group in Cohort 4 ( = .075, 95% CI = .016, .134, p b .05), with generally higher attendance rates for the supportive housing cohort (see Table 3). To investigate whether receipt of supportive housing services was associated with academic achievement, comparisons between the proportion of children passing the MCA-II in the supportive housing and comparison cohorts were made. Table 2 indicates the proportion of children who were procient in MCA II-Reading across cohorts. Results revealed mixed ndings; some supportive housing cohorts displayed greater prociency than their comparison cohorts and some comparison cohorts displayed greater prociency than their corresponding supportive housing cohorts. It is important to note that generally low levels of prociency were demonstrated and that some cohorts experienced decreasing rates of prociency over time. None of GEE analyses showed a signicant relationship between receipt of

Upper 0.198 0.104 0.055 0.430 0.044 0.049 0.051 0.134 1.109 1.500 1.552 1.690 .967 0.097 2.784 1.295 1.911 1.148 0.243 1.033

Attendancea

MCA II-Readingb

MCA II-Mathb

IEP

Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort Cohort

1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4

Group Group Group Group Group Group Group Group Group Group Group Group Group Group Group Group Group Group Group Group

0.062 0.092 0.405 0.030 0.006 0.010 0.007 0.075 0.072 0.492 0.040 0.064 0.089 1.340 1.124 1.244 0.083 0.288 1.102 0.285

0.1326 0.1001 0.1785 0.2350 0.0190 0.0199 0.0223 0.0300 0.6030 0.5144 0.8121 0.8295 .5390 0.6340 0.8469 1.2958 1.0172 0.7326 0.6864 0.6725

0.322 0.289 0.755 0.491 0.031 0.029 0.037 0.016 1.254 0.517 1.631 1.562 1.145 2.582 0.536 3.784 2.077 1.724 2.447 1.603

p b .05. p b .01. a Indicates GEE with continuous response variable. b Indicates GEE with binary response variable.

supportive housing services and performance on the MCA IIReading over time. However, a signicant difference among cohort groups was found for performance on the MCA II-Math. In particular, receipt of supportive housing services was associated with MCA II-Math Performance for Cohort 2. The proportion of the supportive housing group children who were procient in MCA II-Math increased from 29.4% to 31.3%, while that of its comparison group decreased from 30.5% to 16.9%. The supportive housing group's prociency rate, in Cohort 2, was signicantly different from its comparison group ( = 1.340, 95% CI = 2.582, 0.097, p b .05) (see Table 3), even though both groups were

Table 2 Mean (SD) levels of child well-being indicators for supportive housing (n = 70) and comparison cohorts (n = 342) over time. Supportive housing 2007 School mobility Cohort Cohort Cohort Cohort Attendance Cohort Cohort Cohort Cohort MCA II-reading Cohort Cohort Cohort Cohort MCA II-Math Cohort Cohort Cohort Cohort IEP Cohort Cohort Cohort Cohort 1 2 3 4 1 2 3 4 0.94 0.84 0.88 0.46 0.90 0.91 0.90 0.87 (0.87) (0.90) (0.70) (0.66) (0.10) (0.08) (0.11) (0.13) 2008 0.63 0.60 0.67 0.77 0.89 0.92 0.89 0.90 (0.68) (0.68) (0.59) (0.83) (0.10) (0.06) (0.10) (0.09) 2009 0.58 (0.77) 0.15 (0.49) 0.44 (0.70) 0.69 (0.95) 0.88 (0.08) 0.87 (0.14) 0.84 (0.18) 0.90 (0.07) Comparison 2007 0.49 0.54 0.46 0.62 0.89 0.89 0.90 0.90 (0.77) (0.71) (0.69) (0.96) (0.10) (0.11) (0.10) (0.10) 2008 0.39 0.28 0.57 0.57 0.89 0.89 0.89 0.85 (0.89) (0.57) (0.93) (0.87) (0.11) (0.11) (0.11) (0.14) 2009 0.63 0.59 0.73 0.91 0.89 0.87 0.88 0.84 (0.83) (0.81) (1.02) (1.18) (0.12) (0.14) (0.12) (0.16)

1 2 3 4 1 2 3 4 1 2 3 4

0.39 0.50 0.24 0.33 0.47 0.29 0.12 0.10 0.17 0.30 0.12 0.23

(0.50) (0.51) (0.44) (0.49) (0.51) (0.47) (0.33) (0.32) (0.38) (0.47) (0.33) (0.44)

0.26 0.47 0.24 0.25 0.33 0.33 0.13 0.27 0.21 0.45 0.28 0.31

(0.45) (0.51) (0.44) (0.45) (0.49) (0.49) (0.34) (0.47) (0.42) (0.51) (0.46) (0.48)

0.21 (0.42) 0.53 (0.51) 0.21 (0.43) 0.15 (0.38) 0.17 (0.38) 0.31 (0.48) 0.0 (0.00) 0.18 (0.40) 0.26 (0.45) 0.40 (0.50) 0.39 (0.50) 0.38 (0.51)

0.55 0.33 0.37 0.32 0.45 0.31 0.18 0.19 0.18 0.17 0.14 0.24

(0.50) (0.48) (0.49) (0.47) (0.50) (0.46) (0.39) (0.39) (0.38) (0.38) (0.34) (0.43)

0.35 0.36 0.34 0.21 0.26 0.19 0.26 0.13 0.22 0.25 0.20 0.28

(0.48) (0.48) (0.48) (0.41) (0.44) (0.39) (0.44) (0.34) (0.42) (0.44) (0.40) (0.45)

0.37 0.35 0.26 0.33 0.26 0.17 0.21 0.16 0.24 0.25 0.22 0.31

(0.49) (0.48) (0.44) (0.48) (0.44) (0.38) (0.41) (0.37) (0.43) (0.44) (0.42) (0.46)

Note. Missing values were excluded when calculating percentages.

S. Hong, K. Piescher / Children and Youth Services Review 34 (2012) 14401447

1445

decreasing. However, for other cohorts, results were mixed mirroring results found in prociency in MCA II-Reading. Descriptive analysis was used to examine the proportion of children having an IEP on le at school for children receiving supportive housing services and children in the comparison group . The proportion of children with an IEP increased over time for children with and without supportive housing services (see Table 2). None of the GEE analyses showed a signicant relationship between receipt of supportive housing services and the proportion of children with IEPs (see Table 3). Although differences between two groups were not statistically signicant when controlling for the number of children with an IEP in 2007, the number of children with an IEP in the supportive housing group increased at much higher rates than that in the comparison group. Table 4 indicates that the number of children with an IEP on le increased over the years for children receiving supportive housing services. To investigate potential associations between receipt of supportive housing services and child protection involvement, the proportion of children involved in a child protection (CP) report, the total number of CP reports, and the proportion of children in out-of-home placement were compared for the supportive housing and comparison groups over time (see Table 5). Descriptive statistics were primarily used to describe differences between two groups; GEE analysis was not adopted because of a small sample size. For this cohort group analysis, a total of seven children (10%) from the Supportive Housing group were involved in at least one accepted report of child maltreatment between 2007 and 2009, whereas 27 children (8%) in the Comparison group had at least one accepted report of child maltreatment during the same time period. As can be seen in Table 5, the Supportive Housing group started out with higher rates of child protection involvement than the Comparison group. However, while child protection involvement decreased overall between 2007 and 2009 for the Supportive Housing group (from

Table 5 Descriptive information about the number of children with child protection involvement for supportive housing (N = 70) and Comparison (N = 342) cohorts over time. Supportive housing (20072009) N % Comparison (20072009) N 8 12 10 15 20 10 6 12 8 4 8 8 % 2.3 3.5 2.9 33.3 44.4 22.2 13.3 26.7 17.8 1.2 2.3 2.3

# of children with CP Involvementa 2007 6 8.6 2008 3 4.3 2009 1 1.4 # of CP report (duplicated)b 2007 7 58.3 2008 3 25 2009 2 16.7 # of CP report determinations (duplicated)b 2007 4 33.3 2008 2 16.7 2009 1 8.3 # of children with out-of-home placement 2007 2 2.9 2008 0 0 2009 1 1.4

a Reects involvement in CP each year; children may be involved in multiple years and would therefore be duplicated across years. b % is calculated by dividing the number of CP reports in a given year by the total number of CP reports during 200709.

approximately (9% to 1%), the same was not true of the Comparison group (whose involvement rates increased from approximately 2% to 3%). This pattern was evident across measures of child protection involvement with the exception of number of CP reports. During the investigation period, it appeared that the number of children with CP involvement, the number of CP reports, and the number of accepted child maltreatment reports with determination

Table 4 Primary disability diagnoses for children receiving supportive housing services over time (n = 70). 2007 Number Cohort 1 Non-disabled student Speech/language impairment Developmental cognitive disabilities: Physically impaired Specic learning disabilities Emotional/behavioral disorders Other health disabilities Total Non-disabled student Speech/language impairment Developmental cognitive disabilities: Physically impaired Specic learning disabilities Emotional/behavioral disorders Other health disabilities Total Non-disabled student Speech/language impairment Developmental cognitive disabilities: Physically impaired Specic learning disabilities Emotional/behavioral disorders Other health disabilities Total Non-disabled student Speech/language impairment Developmental cognitive disabilities: Physically impaired Specic learning disabilities Emotional/behavioral disorders Other health disabilities Total 11 1 1 13 15 1 1 1 18 14 1 4 1 20 15 2 17 Percent 84.6 7.7 7.7 100 83.3 5.6 5.6 5.6 100 70.0 5.0 20.0 5.0 100 88.2 11.8 100 2008 Number 14 3 1 1 19 11 1 6 2 20 12 4 2 18 9 1 1 1 1 13 Percent 73.7 15.8 5.3 5.3 100 55.0 5.0 30.0 10.0 100 66.7 22.2 11.1 100 69.2 7.7 7.7 7.7 7.7 100 2009 Number 12 6 2 20 11 5 2 18 8 1 1 2 1 13 13 2 1 2 1 19 Percent 60.0 30.0 10.0 100 61.1 27.8 11.1 100 61.5 7.7 7.7 15.4 7.7 100 68.4 10.5 5.3 10.5 5.3 100

mild-moderate

Cohort 2

mild-moderate

Cohort 3

mild-moderate

Cohort 4

mild-moderate

Note. The total numbers of children uctuated because of student disruption.

1446

S. Hong, K. Piescher / Children and Youth Services Review 34 (2012) 14401447

in which children were involved decreased sharply each year for the Supportive Housing group, whereas these measures of child protection involvement for children in the Comparison group uctuated over time. Out-of home placements decreased by approximately 50% over time for the Supportive Housing group, but they increased by approximately 50% for the Comparison group. 4. Conclusion This study was conducted to explore the relationship between receipt of supportive housing services and children's well-being. Specifically, this study investigated the impact of supportive housing services on children's educational outcomes, including school attendance, school mobility, student MCA-II performance, having an IEP, and child protection involvement. Findings revealed that supportive housing services had positive impacts on child well-being across educational and child welfare outcomes. Positive effects of receipt of supportive housing services were found in school mobility, school attendance, IEP status, and math achievement, even though levels of signicance were not reached for all cohort groups/grades. In particular, signicant effects revealed that fth grade children receiving supportive housing services showed less school mobility than their homeless peers; sixth grade children receiving supportive housing services attended school at signicantly higher rates than children from the comparison group; and more fourth grade children receiving supportive housing services were procient at MCA II-Math than children from the comparison group. Other cohorts/grades showed similar patterns, though a level of significance was not reached for all cohorts. Although these ndings suggest positive effects of supportive housing receipt, it is important to note that less than half of all children in this study demonstrated prociency on the MCA II. While supportive housing may help to stabilize children in their academic environment (as seen by lower levels of school mobility and higher attendance rates for supportive housing cohorts), the negative effects of homelessness on academic achievement (Larson & Meehan, 2009; Masten et al., 1997; Obradovic et al., 2009; Rafferty et al., 2004; Rubin et al., 1996; Shinn et al., 2008; Zima et al., 1994) may still be present for these children. The increasing number of children with an IEP over the years for children receiving supportive housing services possibly suggests that increased attendance and reduced school mobility represents an opportunity for a) school records from previous schools (in which an IEP and/or disability had been identied) to "catch up" to the student, resulting in an opportunity for the new school to serve the student through an IEP or b) school staff has more opportunity to identify existing (previously undiagnosed) disabilities and/or address the overall needs of the student which becomes formalized by coding of an IEP in administrative systems which may or may not involve a new diagnosis. These increases could be also explained by an increasing number of children diagnosed with specic learning disabilities with increasing age. In special education, specic learning disabilities may not be apparent early in student performance in academic subjects. As these disabilities become noticeable and evident over the course of 5th and 6th grade, the number of children with specic learning disabilities may increase. It is also important to note that children receiving supportive housing services had declining levels of child protection involvement while their homeless peers showed no reduction in child protection involvement over time. Children receiving supportive housing services had declining involvement in child maltreatment reports, determinations of maltreatment, and out-of-home placements, whereas their homeless peers had increasing involvement in these areas over time. Though a small sample size limited the analysis that could be conducted, descriptive analysis suggests that supportive housing services are benecial in reducing children's need for public child welfare system intervention.

These ndings are consistent with the focus of services provided to the supportive housing group. The supportive housing service's primary focus is providing permanent, stable housing for all enrolled families. Families also then receive other social services from which they may benet. It was hypothesized that the reduction of school mobility for the supportive housing group was a direct effect of housing service receipt, which also was presumed to have an impact on student attendance (e.g., Crowley, 2003). Once children are attending school on a regular basis, researchers theorize that children may come to the attention of educators, which then may lead to increased rates of referral to special education, receiving an IEP, and academic achievement. Other supportive services provided in the supportive housing model are presumed to assist with academic outcomes. It is important to note that although the study used a multisystemic view to understand the impact of supportive housing services on children's well-being, important limitations exist. First, sample sizes in each cohort group were too small to identify any possible mediation that might cause an indirect effect on the observed relationships. Such mediation can be identied by examining changes in family structures and systems that are the central focus of momentum in human development (Bronfenbrenner, 1979). Information about family structures and systems may include parenting skills, parent's education level, school-level data, child cognitive functioning, and certain types of child welfare services provided to families. Second, compared to the supportive housing group, the history of homeless children in the comparison group was not available in 2007. Therefore, only the comparison group's homeless data in 2008 and 2009 was used. This raises a question about the equality of variances in different samples since homogeneity of variance is critical in comparing two different groups in inferential statistics. In addition, it is beyond the scope of this study to disentangle the independent effects of housing from other supports provided to families in the supportive housing model. Rather, this study sought to understand the impact of supportive housing as a model of housing support on children's well-being. 5. Recommendations Findings of this study lead to recommendations for future research as well as important policy implications. First, additional research is needed to develop more rigorous approaches to understand the effect supportive housing services have on children's well-being. Further investigations could include families histories of homelessness, larger sample sizes, experimental designs, and advanced methodologies in statistics to identify mediator variables and moderators that may help explain differences in outcomes that may be attributed to receipt of supportive housing services. Additionally, it will be important to understand the impact of supportive housing on children's well-being as compared to other types of housing service receipt, such as emergency and transitional housing services. Findings also reveal implications for policy at the local and national level. At the local level, agencies providing supportive housing services may want to consider the array of services offered to families they serve. Though a focus on providing housing services is critical, additional services targeted toward child well-being may be warranted, especially as they pertain to academic achievement. Inclusion of tutoring services and increased access to academic resources (such as computers, printers, etc.) may assist children in their academic achievement, thus leading to higher prociency on standardized achievement tests, lower rates of grade retention, and higher rates of graduation. Agencies and schools may also wish to re-consider how they collaborate with one another. Stronger communication between school systems and supportive housing services will allow both parties a better understanding of children's contexts, which may impact student well-being. In addition, given that the population of families served by supportive housing agencies is comprised largely of particular cultural groups,

S. Hong, K. Piescher / Children and Youth Services Review 34 (2012) 14401447

1447

agencies may wish to consider partnering with culturally-specic service providers to assist children in their academics. At the national level, several recommendations also come to light. First, given the fact that there is an increased number of children and families who struggle with homelessness and the fact that the ndings of this study support the benet of supportive housing receipt, it will be important for policymakers to nd ways to create funding avenues for both the provision of supportive housing services and on-going evaluation of these programs. Funding allocated to the provision of supportive housing services may reduce the need for other, more costly services (such as child protection intervention, juvenile justice, etc.) in the future though future research is warranted. In addition, improved efciency for enrolling eligible children in IEPs may be warranted. Finally, increased funding for educational specialists to work with homeless families entering housing assistance programs may alleviate some of the negative effects of homelessness on children's well-being in the educational system.

References
Anooshian, L. J. (2005). Violence and aggression in the lives of homeless children: A review. Aggression and Violent Behavior, 10, 129152. Bronfenbrenner, U. (1979). The ecology of human development: Experiments in nature and design. Cambridge, MA: Harvard University Press. Buckner, J. C., Bassuk, E. L., & Weinreb, L. F. (2001). Predictors of academic achievement among homeless and low-income housed children. Journal of School Psychology, 39, 4569. Crowley, S. (2003). The affordable housing crisis: Residential mobility of poor families and school mobility of poor children. The Journal of Negro Education, 72(1), 2238. Culhane, J., Webb, D., Grim, S., Metraux, S., & Culhane, D. (2003). Prevalence of child welfare services involvement among homeless and low-income mothers: A veyear birth cohort study. Journal of Sociology and Social Welfare(3), 7997. Dufeld, B., & Lovell, P. (2008). The economic crisis hits home: The unfolding increase in child & youth homelessness. : The National Association for the Education of Homeless Children and Youth and First Focus December, 2008. Retrieved on September 8, 2011 from http://www.rstfocus.net/Download/TheEconomicCrisisHitsHome.pdf Harburger, D. S., & White, R. A. (2004). Reunifying families, cutting costs: Housingchild welfare partnerships for permanent supportive housing. Child Welfare, 83(5), 493508. Individuals with Disabilities Education Act (IDEA) Amendments of 1997, 20 U.S.C. 1414(d) (Supp. III 1997). Larson, A., & Meehan, D. (2009). Minn-LInK Child Welfare Special Topic Report No. .7 Homeless and highly mobile students: A description of the status of homeless students from three school districts in Minnesota. Retrieved on September 17, 2011 from http://www.cehd.umn.edu/ssw/CASCW/attributes/PDF/minnlink/ ReportNo7.pdf Larson, A. M., & Meehan, D. M. (2011). Homeless and highly mobile students: A population-level description of the status of homeless students from three school districts. Journal of Children and Poverty, 17(2), 187205.

Liang, K. Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 1322. Maltreatment of Minors Act (1993). MN Stat. Ann. 626.556 subd. 2. Masten, A., Miliotis, D., Graham-Bermann, S., Ramirez, M., & Neemann, J. (1993). Children in homeless families: Risks to mental health and development. Journal of Consulting and Clinical Psychology, 61, 335343. Masten, A., Sesma, A., Si-Asar, R., Lawrence, C., Miliotis, D., & Dionne, J. (1997). Educational risks for children experiencing homelessness. Journal of School Psychology, 35, 2746. Minnesota Department of Education (MDE) (2011). MARSS Appendices. Retrieved September 3, 2011, from http://education.state.mn.us/MDE/Accountability_Programs/ Program_Finance/MARSS_Student_Accounting/MARSS_Instruction_Manual/ Appendices/index.html National Alliance to End Homelessness (NAEH) (2011). State of homelessness in America 2011. Retrieved on September 2, 2011 from http://www.endhomelessness.org/ content/article/detail/3668 National Center for Chronic Disease Prevention and Health Promotion (NCCDPHP) (2010). Registry Plus. Retrieved on September 2, 2011 from http://www.cdc. gov/cancer/npcr/ Obradovic, J., Long, J. D., Cutuli, J. J., Chan, C. -K., Hinz, E., Heistad, D., et al. (2009). Academic achievement of homeless and highly mobile children in an urban school district: Longitudinal evidence of risk, growth, and resilience. Development and Psychopathology, 21, 493518. Park, J. M., Metraux, S., Brodbar, G., & Culhane, D. P. (2004). Child welfare involvement among children in homeless families. Child Welfare, 83, 423436. Rafferty, Y., & Rollins, N. (1989). Learning in limbo: the educational deprivation of homeless children. Long Island City, NY: Advocates for Children (ERIC Document Reproduction No. ED 312363). Rafferty, Y., Shinn, M., & Weitzman, B. C. (2004). Academic achievement among formerly homeless adolescents and their continuously housed peers. Journal of School Psychology, 42, 179199. Rog, D. J., & Buckner, J. C. (2007). Homeless Families and Children. Paper presented at Toward Understanding Homelessness: The 2007 National Symposium on Homelessness Research Retrieved on October 13, 2011 from http://aspe.hhs.gov/hsp/ homelessness/symposium07/rog/ Rubin, D. H., Erickson, C. J., San Augustin, M., Cleary, S. D., Allen, J. K., & Cohen, P. (1996). Cognitive and academic functioning of homeless children compared with housed children. Pediatrics, 87, 289294. Shinn, M., Schteingart, J. S., Williams, N. C., Carlin-Mathis, J., Bialo-Karagis, N., BeckerKlein, R., et al. (2008). Long-term associations of homelessness with children's well-being. American Behavioral Scientist, 51, 789809. U.S. Department of Education (2004). Education for homeless children and youth program: Title VII B of the McKinney-Vento homeless assistance act, as amended by the No Child Left Behind Act of 2001. Washington DC: US Department of Education. U.S. Department of Housing and Urban Development (HUD) (2010). The 2009 Annual Homeless Assessment Report. Wilder Research (2010). Homelessness in Minnesota 2009: Results of the Wilder statewide survey. Retrieved on September 17, 2011 from http://wilder.org/download. 0.html?report=2339 Wilder Research (2011). 2009 Minnesota homeless study: Homeless children and their families. Retrieved on October 13, 2011 from http://www.wilder.org/download.0. html?report=2399 Zima, B., Wells, K., & Freeman, H. (1994). Emotional and behavioral problems and sever academic delays among sheltered homeless children in Los Angeles County. American Journal of Public Health, 84, 260264.