The Relation Between Home Visiting Program Enrollment and Adolescent Mothers’ Depression:
How it Differs Based on Mothers’ Past and Present Social-Emotional Experiences Emily Morton
Tufts University
Abstract
The purpose of the current study was to examine how the influence of enrollment in a home visiting program on adolescent mothers’ depression varied due to their social-emotional experiences. Home visiting programs, evidence-based family-centered services in which a
trained home-visitor regularly meets with the mother, have become a popular method of reaching and supporting this at-risk population, and may improve mothers’ mental health. However, little is known about how the social-emotional experiences that predict depression affect the ability of the home visiting program to decrease mothers’ depressive symptomatology. Two-hundred eighty-nine (289) mothers were interviewed over the phone and in-person four times between pregnancy and when their first child reached 5 years of age. Interviewers collected data on mothers’ depressive symptomatology, various types of social support, quality of their childhood parent-child relationship, and reports of childhood maltreatment. Analyses included 2x2 mixed- model ANCOVAs and multiple regression analyses. Statistical trends emerged such that, among mothers enrolled in the home visiting program, their depressive symptomatology differed as a result of the dependability of their social support from friends and family and their childhood experience of maltreatment. Implications of the study include the home visiting program’s ability to benefit the mental health of mothers who have better support and mothers who have not had traumatic experiences of maltreatment in childhood. Thus, the study expands upon previous research on adolescent mothers, home visiting programs, and depression, by identifying
subgroups of mothers for whom the home visiting program serves as a protective factor for better mental health.
Keywords: home visiting program, adolescent mothers, depression, social support, parent-child relationships, child maltreatment
The Relation Between Home Visiting Program Enrollment and Adolescent Mothers’ Depression:
How it Differs Based on Mothers’ Past and Present Social-Emotional Experiences Maternal depression has serious health consequences for both mothers and their children.
Among adolescents and adults, depression is related to problems with anxiety, substance abuse, maintaining employment, educational achievement, risky sexual behaviors, and overall
functioning (Fergusson & Woodward, 2002). For mothers, these consequences impair not only their personal well-being, but also their ability to provide sensitive parenting (Lovejoy, Graczyk, O'Hare, & Neuman, 2000). Unsurprisingly, children of depressed mothers are at particularly high-risk for delayed cognitive, socioemotional, and motor development (Bernard-Bonnin, 2004;
Cummings & Davies, 1994; Petterson & Albers, 2001). Adolescent mothers are especially at-risk for depression, as clinical levels of postpartum depression (PPD) are nearly twice as common among adolescent mothers in comparison to adult mothers (Kim, Connolly, & Tamim, 2014).
Adolescent mothers are also more at risk for depression than the general population of
adolescents, as up to 38.1% of a sample of adolescent mothers had clinical levels of depressive symptoms one year postpartum (Easterbrooks, Kotake, Raskin, & Bumgarner, 2015; Birkeland, Thompson, & Phares, 2005), whereas only an estimated 8.7% of U.S. adolescents aged 18-25 experienced at least one Major Depressive Episode (MDE) in 2013 (United States Department of Health and Human Services, 2014). Home visiting programs, which are evidence-based family- centered services that provide mothers with regular meetings with trained home-visitors, have become a popular method of reaching and supporting this at-risk population.
In the present study, I specifically aimed to examine differences in depressive
symptomatology at one point in time between adolescent mothers who were randomly assigned to participate in a home visiting program and adolescent mothers who did not participate in the
program and were assigned to a comparison condition. Furthermore, I considered how additional factors related to depression, such as social support and relational experiences in childhood, as well as home visiting program participation, were related to mothers’ depressive
symptomatology approximately 4 years after program participation.
Adolescent Motherhood Rates and Predictors
Over the past twenty years, rates of teen births in the United States have continually declined. Likely due to both cultural changes and the implementation of many effective teen pregnancy prevention programs, the number of adolescent girls aged 15-19 who gave birth decreased from 61.8 per thousand girls in 1991 to 26.5 per thousand girls (a total of 273,105 births) in 2013 (Martin, Hamilton, Osterman, Curtin, & Mathews, 2015). Nevertheless, the rate of teenage births in the United States remains the highest of all industrialized countries (United Nations Statistics Division, 2015).
Although the percentage of teens giving birth has declined over time, adolescent
motherhood is still a serious concern in the United States; adolescent mothers are a highly at-risk population, not only because of the stress of young motherhood, but also because of the
environments they lived in before they became pregnant. Teens are more likely to become pregnant when they live in poverty, drop out of high school, are in a racial minority group, use welfare, live in a single-mom household, have poor or abusive parent-child relationships, and have a history of demonstrating other risky or deviant behaviors, such as smoking or substance use (Cavazos-Rehg et al., 2010; Coley & Chase-Lansdale, 1998; Miller, Benson, & Galbraith, 2001; Mullen, Martin, Anderson, Romans, & Herbison, 1996). Thus, even before adolescent mothers become pregnant, they have often experienced difficult home environments with little
support, which places them at risk for developing mental health issues, such as depression, in the future.
The decision to become a mother often exacerbates the challenging environment that preceded the pregnancy. For example, in comparison to pregnant teens who have an abortion or give up their children for adoption, those who become mothers have less stable partner
relationships, more nonmarital births, greater poverty, greater health problems, lower life
satisfaction, and a greater number of stressful life events (Coley & Chase-Lansdale, 1998). Coley and Chase-Landsdale (1998) suggest that these adverse consequences are a product of
simultaneously managing the stress of being both an adolescent and a parent. Furthermore, balancing these two roles can force mothers to shorten or skip the “identity crisis,” or identity exploration, normative to adolescence (Coley & Chase-Lansdale, 1998; Erikson, 1968). This period of exploration entails reexamining values and choices and exploring alternatives, and, according to psychosocial developmental theorists Erikson (1968) and Marcia (1991), is necessary to develop healthy psychological functioning. Shortening or skipping this period of exploration can lead to identity foreclosure, when an adolescent is not able to complete his/her identity exploration and is forced to commit to an identity prematurely, or identity diffusion, when an adolescent has an unstable identity and never develops a sure sense of self (Marcia, 1991). Both foreclosed and diffused identity statuses are related to psychological problems such as depression.
Therefore, understanding and evaluating interventions and programs, such as home visiting programs, that provide support to teenage mothers is of utmost importance. It is critical to investigate whether there are aspects of mothers’ social-emotional life that could impede or enhance the ability of a program to improve mothers’ mental health. Understanding these
relationships would help to identify the populations of mothers for whom home visiting
programs are the most effective, as well as the populations of mothers who may need additional or alternative support. Gaining this understanding would allow policymakers and practitioners will to target these programs at the groups who are most likely to benefit from them.
Home Visiting Programs and Depressive Symptomatology
Home visiting programs are one strategy within the family-centered support approach.
Programs within this approach aim to help families meet their essential needs through the use of a mentoring model (Chaffin, Bonner, & Hill, 2001). The family-centered approach has been empirically shown to improve many different family outcomes, from reducing the prevalence of children’s behavioral and emotional problems to reducing childhood obesity and improving parental mental health (Sanders, 1999; Davison, Lawson, & Coatsworth, 2012; Lim Brodowski, 2012). Maternal home visiting programs are an evidence-based type of family-centered support service in which a trained home-visitor regularly meets with the mother and family to promote positive parent-child interactions and child development by providing regular support, coaching, and information. Though the implementation of home visiting has become a popular method of intervention, research findings on the general efficacy of home visiting at improving mothers’
mental health, as opposed to the broader family outcomes, is mixed.
Whereas some studies find that home visiting outcomes include improved maternal psychological well-being and parent-child relations (Lim Brodowski, 2012; Mckelvey et al., 2015), others find very limited or minimal maternal effects (Sierau et al., 2015; Easterbrooks et al., 2015). Easterbrooks et al. (2015) argue that the limited effects they observed can be at least partially attributed to the fact that the home visiting program they examined did not focus specifically on reducing maternal depression, and that much of the final variation they found in
mothers’ depressive symptomatology was due to differences that existed before mothers entered the program. Despite the minimal size of some of the effects found, it is still important to
acknowledge that almost all studies find positive effects of home visiting on various aspects of at least some mothers’ psychological outcomes, such as parenting stress (Sierau et al., 2015;
Easterbrooks et al., 2015). Furthermore, due to the multiplicity of the factors that contribute to depression, it is not surprising that programs would not have the same effect for all mothers or be equally effective at reducing depressive symptomatology in all mothers.
Nevertheless, several studies support the positive influence of home visiting programs on maternal depression. For example, in a longitudinal analysis of changes in parental depression among a diverse sample of 569 parents involved in the Family Connections Prevention
Replications Project, which used home visiting services to implement a child maltreatment prevention program, Lim Brodowski (2012) found that depression of all participants decreased over time. This finding is limited by the lack of a control group that was not enrolled in the home visiting program, and, thus, cannot show causation because the effect could be moderated by many factors including time; nevertheless, the study affirms that enrollment in such a program and depression are at least correlated within that study (Weissman et al., 2006). Mckelvey et al.
(2015) found support for the causal effect of a home visiting program, specifically the Infant Mental Health home-based Early Head Start program, on maternal depression. Their longitudinal study compared the outcomes of low-income families enrolled in the program to those of low- income families in a comparison condition. Program enrollment was found to be predictive of decreased maternal depression, decreased parenting stress, and better family functioning.
However, the generalization of this study to all home visiting programs is limited because this
intervention focused more specifically on mental health than other home visiting programs (Mckelvey et al., 2015).
Overall, the evidence for the potential positive influence of home visiting on maternal depression is compelling; the recurring positive, though sometimes minimal, effects suggest that there may be subgroups of mothers whose mental health is differentially impacted by the
program, and that there may be other factors moderating the impact of the program. This explanation is based on a developmental-systems theory framework, as it accounts for the interaction of different environmental and relational environments on an individual’s development.
Developmental Systems Theory Framework
The present study considers adolescent development and well-being from the approach of developmental systems theory (Lerner & Castellino, 2002). DST, embedded within the greater Relational Developmental Systems metatheory, is a holistic and transactional perspective that considers development to be the interaction of individual and environmental factors over time (Lerner, Theokas, & Bobek, 2005; Overton, 2013). Furthermore, DST acknowledges that individuals have an active role in shaping their environments and their development by regulating and integrating the influences of their biology (e.g., genetic traits) and their various relational contexts (e.g., family, school, society). These relational contexts comprise multiple levels of the ecology of human development whose interaction and integration with individual factors is the key process in adolescent development (Lerner, 2005). Figure 1 illustrates the influence of these various social systems (arrows pointing toward the individual) as well as of the individual’s perception of and agency within these systems (arrows pointing away from the individual).
Adolescence is a time at which the developmental systems approach is especially applicable, as adolescents’ roles are changing and evolving in each of these contexts, and their growing autonomy increases the impact of their personal agency in their roles, contexts, and overall development (Lerner et al., 2005). Furthermore, DST’s transactional approach focuses on the influence of the interaction of all of the systems within the ecology of human development, rather than that of any individual system. Thus, the theory rejects the previous idea that the developmental system could be dichotomized or split into multiple levels of organization, and claims that the systems are inherently integrated in the individual’s experience (Lerner, 2005).
The present study operates from a DST framework by considering the interactive effect of three relational systems on adolescent mothers’ mental health. The three relational systems include (1) a mother’s experience of social support in various contexts, (2) a mother’s childhood relational history, and (3) a mother’s enrollment in a home visiting program. The potential integrated effect of these relational systems is explored through their relation to mothers’
depressive symptomatology. DST suggests that any effect that the home visiting program has on maternal depression will not be independent of the strength of the mothers’ other relational contexts. Mothers’ relational contexts that have been closely linked to higher depressive symptomatology include low social support and negative childhood relational experiences (Bailey, Moran, & Pederson, 2007; Brown, Harris, Woods, Buman, & Cox, 2012; Green et al., 2010; Silver, Heneghan, Bauman, & Stein, 2006). Thus, the DST approach suggests that these negative relational experiences would interact with mothers’ home visiting program enrollment to affect their depressive symptomatology; mothers with more positive perceptions of their past and present relational contexts may benefit more from a home visiting program and have lower
depressive symptomatology than those who have less positive perceptions of their relational contexts.
Social Support for Adolescent Mothers
Though there are many existing definitions of social support, the present study defines social support as the degree to which an individual feels both that he/she is adequately cared for by others and that he/she is integrated within a social network (Schlecker & Fleischer, 2013).
When measuring social support as individuals’ perceived support, two dominant, empirically- based models explain the connection between social support and health: the buffering hypothesis and the direct effects hypothesis. The buffering hypothesis argues that high social support
generally protects against the negative experience of stress associated with aversive life events (e.g., divorce, death of a relative) (Cohen & Willis, 1985). This hypothesis is based on the
finding that, among people who experience more aversive life events, those who have high social support are less likely to have poor health than those who have low social support (Cohen &
Willis, 1985). Alternatively, the direct effects hypothesis suggests that high social support
predicts better mental and physical health outcomes, regardless of one’s stressful life experiences (Cohen & Willis, 1985; Lakey & Cronin, 2008; Uchino, 2009). Moreover, because the
participants in the present study are high-risk and likely to have experienced many stressful life events, both the buffering hypothesis and the direct effects hypothesis suggest that the
participants who have greater social support will experience less stress, and should have lower levels of depressive symptomatology.
Categories of support. Sources of social support can include a variety of people and systems, such as family, friends, neighbors, and organizations. Both informal supports and formal supports have been shown to protect against developing depressive symptomatology
(White & Hastings, 2004; Rodrigo, Martín, Máiquez, & Rodríguez, 2007; Thompson & Peebles- Wilkins, 1992). Informal supports generally include family members, partners, and friends, and are often individuals’ primary source of support. Formal supports, alternatively, include
hospitals, childcare centers, and other private and public organizations. The research comparing the difference in the effects of formal and informal support is somewhat controversial: though some studies claim that professional services that integrate both types of support are the most effective way to help highly at-risk adolescent mothers (Thompson & Peebles-Wilkins, 1992), other researchers find there is little evidence for the meaningful influence of organizations and professionals on the lives of at-risk adolescents, and argue that support from friends and family is much more impactful (Rodrigo et al., 2007). Additional research examining parents facing challenges in other contexts, such as parenting children with disabilities, further suggests that informal support is more important to individuals’ daily functioning than formal support (White
& Hastings, 2004). Moreover, formal support ideally supplements informal support, though it occasionally provides primary support for individuals who have limited or no informal support.
Thus, formal support can be especially important for adolescent mothers, as their informal support is often unreliable (Boath, Henshaw, & Bradley, 2013).
Continuing to expand on the distinction between formal support and informal support, Ceci and Papierno (2005) discuss what formal support specifically is most helpful and effective.
They conclude that, contrary to societal expectations, intervention programs tend to help the individuals who have more resources and the least risk, the “haves,” much more than those who have fewer resources and are most at-risk, the “have nots.” Thus, though controversial, the research supports this perspective that formal support alone is not likely to be an effective support system for highly at-risk adolescent mothers (Ceci & Papierno, 2005). Ideally, however,
formal support, such as home visiting programs, is used to supplement informal support in order to increase adolescent mothers’ total support and improve both maternal and child well-being.
Social support, depressive symptomatology, and home visiting programs. Extensive evidence suggests that lower social support predicts greater depressive symptomatology (Silver et al., 2006; Brown et al., 2012). Although most of this research on mothers’ social support and depressive symptomatology focuses on mothers during only their first year post-partum,
Schmidt, Wiemann, Rickert, and Smith (2006), found that depressive symptoms continued to be prevalent in adolescent mothers during the first four years postpartum. Thus, teen moms are likely to remain at risk for depression for at least several years after giving birth, and their depressive symptomatology is likely linked to their levels of social support. Because of this connection between levels of social support and depressive symptomatology, I suggest that mothers’ social support could play an integral role in the association between home visiting program enrollment and low depressive symptomatology.
Limited research explores the effect of informal social support on home visiting program outcomes, and especially on outcomes related to maternal mental health. Easterbrooks, Kotake, Raskin, and Bumgarner’s (2015) study specifically considered mothers’ relationship satisfaction with the father of their baby and their home visiting program status (enrolled or control) as factors possibly contributing to their depression. The researchers found that both mothers’
program status and mothers’ satisfaction with father support significantly related to mothers’
experience of depression within the first two years postpartum. More specifically, being enrolled in the program was most effective at helping mothers who were mentally healthy at enrollment remain mentally healthy, rather than helping mothers recover from depression. The study further highlighted the protective role that father support, a type of informal support, can have on the
likelihood of mothers’ depression to remit over the course of the home visiting program.
However, this study considered only father support and acknowledged the need for additional research to investigate the relationships between other sources of support and mothers’
depression throughout the program.
Much of the research that explores the detrimental effect of low social support on the effectiveness of home visiting programs focuses on outcomes related to child health. For example, Le Roux, Rotheram-Borus, Stein, and Tomlinson (2014) examined the relationship between home visiting program enrollment and infant health outcomes. The researchers
implemented a home visiting program in Cape Town, South Africa, that followed women from pregnancy through 18 months postpartum and aimed to improve infant health outcomes. The study measured participants’ informal social support from baseline to six months postpartum, and based the measure on the quantity of close relationships mothers reported and the frequency of support mothers received. The results showed that, although the program did not have a direct effect on infant health, it did have a significant indirect effect when moderated by informal social support. The existence of this moderation effect such that families with more support benefitted more from the program suggests that there could be similar group differences in the effects of home visiting programs on mothers with different levels of social support. Furthermore, because improved maternal mental health has been found as an outcome of home visiting programs in conjunction with improved infant development, it is possible that the relationship between adolescent mothers’ program enrollment and depressive symptomatology would depend on mothers’ perceived amount of social support (Haskins, Paxson, & Brooks-Gunn, 2009).
From a developmental systems framework, it is important to consider not only how a mother’s immediate social environment extrinsically impacts her experience of and benefit from
the home visiting program, but also how her history and perceptions of past relational
experiences intrinsically influences her experience of the program. Therefore, the present study will also investigate how the relationship between home visiting program enrollment and depressive symptomatology may vary based on mothers’ social-emotional childhood experiences.
Social-Emotional Experiences in Childhood
Children’s social-emotional, or relational, experiences have serious consequences for their future relationships and self-concepts. Attachment theory, as formulated by Bowlby (1969) and expanded upon by Ainsworth (1978) explains that children’s relationships with their primary caregivers crucially impact the children’s ability to regulate their emotions and develop
expectations of others in future relationships. When a caregiver is sensitive, responsive, and caring, a child will form a secure attachment to that person and learn to trust others. When a caregiver’s behavior is consistently insensitive or unpredictable, the child can develop an insecure or disorganized attachment style, developing negative or fearful expectations of others based on his/her interactions with his/her caregiver. Though these expectations can change over time, research shows that adult attachment styles are often consistent with those developed during infancy (Waters, Merrick, Treboux, Crowell, & Albersheim, 2000). Children who experienced major trauma or abuse were most likely to develop a disorganized attachment style that became an Unresolved attachment pattern in adulthood. Individuals with Unresolved attachment patterns typically struggle to maintain emotional closeness with other people, lack trust in their partners, have very low self-esteem, and are at increased risk for developing psychopathology. Unfortunately, predictors of Unresolved attachment patterns, including a history of childhood maltreatment and parent-child relationships that lacked sensitivity and/or
limited the development of autonomy, are common among adolescent mothers (Bailey et al., 2007). Research shows that adolescent mothers who experienced such abuse in childhood and develop an Unresolved adult attachment status are especially likely to have experienced dissociation, confusion surrounding identity, relationship problems, and mental health issues (Bailey et al., 2007; Stovall-McClough & Cloitre, 2006). Additional research further links adults’
childhood experiences of maladaptive parenting and family functioning, such as maltreatment, family violence, and parental mental illness, to outcomes of psychopathology and depression (Green et al., 2010). Thus, the present study operates from the perspective that adolescent
mothers continue to be affected by their childhood relationships, expecting that mothers who had aversive relational experiences during their childhood would have greater depressive
symptomatology than those who had positive experiences.
Childhood maltreatment. The term childhood maltreatment encompasses any incidence of neglect or physical, emotional, or sexual abuse during one’s childhood (Korbin & Krugman, 2014). An abundance of research shows the connection between a history of maltreatment in childhood and increased depressive symptomatology in adolescence and adulthood (Shea,
Walsh, MacMillan, & Steiner, 2005; Brown, Cohen, Johnson, & Smailes, 1999). It is particularly important to study the relationship of depressive symptomatology and childhood maltreatment in adolescents, as studies find up to 55% of sampled adolescent mothers experienced some sort of physical or sexual abuse in childhood (Noll, Shenk, & Putnam, 2009; Bailey et al., 2007; Tufts Interdisciplinary Evaluation Research, 2015). When considering these reported rates of
maltreatment, it is important to consider that the number of substantiated reports of maltreatment in a population are often dramatically lower than the number of reports ascertained through self- report, and that neglect is reported even more infrequently (MacMillan et al., 2003). Specifically
considering the relationship between childhood abuse and depression in adolescents, Fergusson, Horwood, and Lynskey (1996) found that 18-year-olds who reported being previously abused had significantly greater rates of clinical depression than their peers who had not been abused.
Childhood maltreatment, depressive symptomatology, and home visiting programs.
Very limited research specifically investigates the role of child maltreatment in mothers’
histories as a moderator of the effectiveness of home visiting programs. Moreover, no research considers the possible moderation effect of child maltreatment specifically pertaining to the impact of home visiting programs on mothers’ depressive symptomatology. Bagley and Young (1999) studied a group of women with a history of child abuse and found that enrollment in social work group therapy that focused on childhood sexual abuse issues predicted decreased depressive symptomatology for all participants six years after enrollment. However, when looking at subgroups of women who were most depressed or who had multiple abuse histories (experiences of multiple types of abuse simultaneously or sequentially), Bagley and Young (1999) found that these groups of women showed less improvement in their mental health. Of course, there are certainly differences between group therapy and home visiting programs, as well as differences between a clinical population of women and a population of adolescent mothers. However, based on these limited related findings, I expected that home visiting programs would be less effective at decreasing depressive symptomatology for mothers who experienced child maltreatment than for those who did not.
Mothers’ childhood relationships with parents. As explained previously, a healthy parent-child relationship is present when a child learns to expect a sensitive and caring response from his/her caregiver, thereby developing a secure attachment. Less healthy parent-child relationships are characterized by insecure attachment patterns, when the child experiences
inconsistent behavior from or a lack of interaction with his/her caregiver. Adolescent mothers are especially likely to have developed insecure attachment bonds to their caregivers when one considers that early pregnancy, as a developmental outcome, is associated with having experienced a stressful childhood environment in which there was a negative parental relationship (Abass, 2014). For example, 37% of Bailey et al. (2007)’s sample of adolescent mothers were found to have an Unresolved attachment style, which typically originates from childhood experiences of abuse, trauma, or neglect. In adults, Unresolved attachment
classifications predicted greater experiences of dissociation, self-identity confusion, and
relationship difficulties. Such prevalence is very concerning among a population of mothers, as mothers’ own attachment styles are often passed on to their children through their interactions (Mazzarello, 2007). More specifically, mothers who reported receiving less positive care in childhood were found to be more neglectful of their children (Bartlett, Raskin, Kotake, Nearing,
& Easterbrooks, 2014). Overall, adolescent mothers are especially likely to have low ratings of positive parental care in childhood and, therefore, especially likely to develop an Unresolved or insecure attachment status and to parent their children in a similar way.
Additional studies have looked at the specific connection between adolescent mothers’
childhood parental care and their depressive symptomatology. Unsurprisingly, studies find that mothers who recalled more caring parental relationships and had evidence of their own secure attachments in childhood had less depressive symptomatology (Kenny & Rice, 1995; Nuttbrock- Allen, 2002). Kenny and Rice’s (1995) study also specifically found that adolescent mothers who reported positive attachments to their parents had less parenting stress and less depressive symptomatology. Similarly, a study of 74 young African-American moms found that adolescents who ranked their relationships with their parents positively had lower levels of depressive
symptomatology five months later than those who ranked their relationships with their parents poorly (Nuttbrock-Allen, 2002). Overall, the research suggests that adolescent mothers with low levels of positive parental care during their childhood would have greater depressive
symptomatology. Furthermore, the strength of their parent-child relationships in childhood will likely predict adolescent mothers’ depressive symptomatology strongly enough that the home visiting program would be less effective at alleviating these symptoms for the mothers who recalled more negative parent-child relationships.
Mothers’ childhood relationships with parents, depressive symptomatology, and home visiting programs. The role of mothers’ recollection of caring relationships with their own parents as a moderator of the effectiveness of home visiting programs is understudied. The most relevant finding is from Moran, Pederson, and Krupka’s (2005) study of a home visiting program that aimed to improve parent-child interactions and the attachment relationship for adolescent mothers and their children. Researchers randomly assigned participants to the home visiting program or a control comparison condition. The study found that the ability of the program to improve parent-child relationships was moderated by mothers’ adult attachment statuses. More specifically, mothers who had Unresolved adult attachment statuses were found to benefit less from the home visiting program than mothers with other, more positive attachment styles (Moran et al., 2005). Because adults who have Unresolved attachment statuses often experienced neglectful or abusive parent-child relationships in their childhood, it is llikely that mothers who recalled negative parent-child relationships also benefited less from the
implemented home visiting program. It is important to acknowledge that this study measured the benefit of the home visiting program through its effect on adolescent mothers’ current
relationships with their children as opposed to mothers’ depressive symptomatology.
Nevertheless, both improved attachment relationships and reduced depressive symptomatology are shown to be outcomes of effective home visiting programs for adolescent mothers; thus, I expected that the effect of home visiting program enrollment on mothers’ depressive
symptomatology would depend on their perception of their parent-child relationships as children such that mothers who were enrolled in a home visiting program would have lower depressive symptomatology, but how much lower would depend on the degree to which the mothers recalled negative relationships with their parents.
Hypotheses
Enrollment in the home visiting program, higher social support, and fewer negative social-emotional experiences in childhood will be related to lower depressive symptomatology in a sample of adolescent mothers. Furthermore, focusing on the differences between mothers enrolled in a home visiting program and mothers in a control group, I hypothesize that mothers enrolled in the program will have lower depressive symptomatology when they (1) have higher levels of various categories of social support, such as: informal support, formal support, and partner support; and (2) had positive social-emotional experiences in childhood, such as: no history of childhood maltreatment or positive parent-child relationships.
Method Healthy Families Massachusetts
Participants and data used for the present study were drawn from a longitudinal randomized control trial evaluation of Healthy Families Massachusetts (HFM). HFM is a statewide newborn home-visiting program that serves first-time parents who were under age 21 at childbirth. The program, Healthy Families Massachusetts (HFM) was adapted from the National Healthy Families America program with the aims of (1) preventing child abuse and
neglect by supporting positive, effective parenting; (2) promoting optimal child health, growth, and development; (3) encouraging parental educational attainment, job, and life skills; (4) preventing repeat teen pregnancies; and (5) promoting parental health and well-being.
The program offers meetings with a home visitor weekly or biweekly, additional support via phone and other media, educational curriculum that is developmentally appropriate for the family, goal-setting instructional activities, access to group-based support and education, and referrals and connections to other applicable resources. The program aims to have visits weekly or biweekly with pregnant mothers, then weekly visits for mothers the first six months
postpartum, followed by decreasing visits respective to the family’s needs after six months postpartum. Families continue to be eligible for the program until the child’s third birthday.
Participation in and utilization of the program is voluntary at all times.
Home visitors are paraprofessionals who have a variety of skills that reach from experience working with families to understanding of developmental processes, family dynamics, cultural attitudes, and how to build trusting relationships. All home visitors
additionally complete an intensive six-day training before working with families as well as ten other topical trainings within their first year. Furthermore, professionally trained staff oversees each member of the home visiting staff, including home visitors, supervisors, and program coordinators, on a weekly basis.
Massachusetts Healthy Families Evaluation
The Massachusetts Healthy Families Evaluation (MHFE) aimed to evaluate HFM
through a randomized controlled trial that assigned participants to enroll in the HFM program or to a control group that received referrals to other programs serving young parents and their children. The group enrolled in HFM was called the Home Visiting Services (HVS) group, and
the control group was called the Referrals and Information Only (RIO) group. The eligibility requirements for participation in the study included being female, at least 16 years old, not having received services from HFM in the past, speaking fluent English or Spanish, and being cognitively capable of providing informed consent. 840 participants were recruited and 704 (84%) agreed to participate in the study. 61% of participants were assigned to the home visiting group and 39% were assigned to the control group.
Program evaluation data were collected in three stages, Massachusetts Healthy Families Evaluation Phase 1 (MHFE-1; 1997-2005), Massachusetts Healthy Families Evaluation Phase 2 (MHFE-2; 2008-2012), and Massachusetts Healthy Families Evaluation: Early Childhood (MHFE-2: EC; 2012-present). MHFE-1 assessed the quality and accountability of the program.
MHFE-2 alternatively aimed to evaluate the impact of program enrollment on participants, and data were collected at participant enrollment (T1), one year after enrollment (T2), and two years after enrollment (T3). MHFE-2EC further examined the long-term impacts of the HFM program on children and families when the target child (TC) was approximately 4.5-5 years old (T4) and again when the child was approximately 5.5-6 years old (T5).
Procedures
Data were collected through phone and in-person interviews conducted by the MHFE data collection team. All mothers who participated at T1 completed a phone interview, and a subsample also chose to participate in an in-person interview, which included additional surveys and behavioral observation. This procedure was similar at T4, with an additional child protocol added in for the in-person interview. The in-person interviews took place in a location of the mother’s choice.
The Analytic Sample
The present study uses data collected from participants in MHFE-2 and MHFE-2: EC.
This study will draw from a subsample of mothers who were pregnant at T1 and also participated in the study at T4 (N=289). I chose to exclude participants who were parenting at T1 in order to control against the possible effects of postpartum depression, as T1 depression scores may not be an accurate indication of baseline depression for mothers who had recently given birth.
Characteristics of the selected mothers and their children are displayed in Table 1, and are largely representative of the greater MHFE sample.
Measures
Demographic characteristics. Demographic information of the mothers and children was collected through maternal interviews over the phone. Questions concerned topics such as ethnicity, language, child age and sex, relationship status, use of public assistance, and
employment.
Depressive symptomatology. This study measured depressive symptomatology using the Center from Epidemiological Studies-Depression (CES-D) scale, which was administered to mothers during the phone interview. The CES-D scale is a 20-item self-report scale that
specifically measures depressive symptomatology (Radloff, 1997). Each item asks respondents to rank how frequently they experienced a certain depressive symptom in the last week using a 4- point Likert scale (0 = not at all, 1 = little, 2 = occasionally, 3 = a lot). Participants’ total scores were calculated as a sum of their responses to all items (scores could range from 0 to 60), with higher scores representing greater depressive symptomatology. A score of 16 is considered the clinical cutoff score, indicating that a total score of 16 or higher indicates risk for clinical depression (Lewinsohn, Seeley, Roberts, & Allen, 1997). The scale is consistently used in clinical research and has been shown to be reliable in clinical populations (α = .90; Radloff,
1997) non-clinical populations (α = .85; Radloff, 1997), and across populations with a wide variety of demographics (Lewinsohn et al., 1997). Cronbach’s alpha for the continuous T1 and T4 CES-D scale sum score of the full MHFE-2 sample were similarly high, at .89 at T1 and .91 at T4. This measure was administered at all time points of the study.
Program status. Mothers’ program status was defined by their initial assignment to the HVS or RIO condition. The HVS group included all mothers who were assigned to the HFM home visiting program, regardless of the number of actual visits the mother had with the home visitor. The RIO group included all mothers assigned to the control condition, which meant they were not enrolled in the HFM home visiting program and were offered only referrals and
information.
Social support. The present study measured mothers’ social support using the Personal Network Matrix (PNM; Trivette & Dunst, 1988) and the Quality of Relationships Inventory (QRI; Pierce, Sarason, & Sarason, 1991), both of which were administered during the in-person interview. The PNM was used to measure mothers’ informal social support and formal social support, whereas the QRI was used to measure mothers’ partner support.
Informal and formal social support. The PNM evaluates mothers’ overall social support
network based on self-report of their ratings of their frequency of contact with and their dependability on 23 potential sources of support. The original form of the PNM includes “My Child” as a potential form of support. “My Child” has been removed from this study as a potential form of positive support, as research shows that maternal dependence on children can have negative consequences for both the mother and child (Burton, 2007). Thus, the PNM was scored using 22 total possible sources of support. As displayed in Table 2, the measure
categorizes potential supports into three categories: familial supports, extrafamilial supports, and formal supports.
Part I of the PNM assesses mothers’ frequency of contact with each source over the last month. Participants ranked their contact with the individual(s) on a 5-point Likert scale (0 = Not at all, 1 = Once or Twice, 2 = At least 10 times, 3 = At least 20 times, 4 = Almost every day).
Participants’ frequency of contact scores were calculated by summing the minimum total number of contacts they had per month with each source of support (Once or twice = 1 contact, At least 10 times = 10 contacts, At least 20 times = 20 contacts, Almost every day = 30 contacts). For example, a participant who had contact with only two formal supports, ten times with one support in the past month and one time with the other support once in the past month, would have a frequency of contact with formal support score of 11. Thus, the possible range of participants’ frequency of support scores depended on the number of applicable supports in the category of support being studied. Part II of the PNM considers the dependability of each of the sources of support. Participants ranked dependability on a 5-point Likert scale (0 = Not at all, 1 = Sometimes, 2 = Occasionally, 3 = Most of the time, 4 = All of the time). Continuous
dependability summary scores were calculated from the combined mean of all the respondents’
dependability ratings; thus, total scores could range from 0 to 4, with 4 representing the most dependable support. Global reliability and validity of the PNM measure have not been
established. However, for the full MHFE-2 sample, Cronbach’s α was .72 for the T4 frequency of support PNM measure and was .77 for the T4 dependability of support measure. In addition, operationalizing social support by measuring frequency and dependability of support is shown to have high construct validity, as demonstrated by other similar studies of social support (e.g., Brown et al., 2012; Le Roux et al., 2014; Nygren, 2013). The PNM was administered at in-
person interviews at all time points of the MHFE-2 study, but this study uses only the PNM data that were collected at T4.
Scores were assigned to the different types of social support by including and excluding certain possible supports before calculating the frequency or dependability score. The informal social support scale included the 12 supports listed as familial or extrafamilial in the PNM.
Therefore, the frequency of support scores for informal support could range from 0 (no contact with any social supports) to 360 (daily contact with 12 different social supports). For the purpose of the present study, participants who interacted with less than an average of three total family members or friends/neighbors per day (fewer than 90 total informal contacts per month) were considered to have low frequency of contact with informal support. The formal social support scale included the 10 PNM formal social support sources, making the frequency of support scores range from 0 (no contact with any supports) to 300 (daily contact with 10 different social supports). Participants who interacted with one formal support at most every two days (fewer than 16 total formal contacts per month) were considered to have low frequency of contact with formal support. The dependability of support measure was scored the same way for each
category of support, and could range from 0 (not dependable) to 4 (always dependable).
Participants were considered to have low dependability if their average rating fell in the bottom third of the scale, meaning they were below 1.34.
Partner support. The quality of the support from mothers’ partners was assessed at T4
during the in-person interview using the Quality of Relationships Inventory (QRI; Pierce, Sarason, & Sarason, 1991). Participants completed the measure as it pertained to their
relationship with the father of the baby (FOB) and as it pertained to their partner over the past year, if they were dating someone other than the FOB. Another variable that indicated the
mother’s relationship status to inform whether or not she was romantically involved with the FOB was used to merge the QRI scores into one variable that represented the mothers’ romantic relationships with current partners, regardless of whether that partner was the FOB or someone else.
This study uses the 7-item support subscale of the QRI that measures the availability and reliability of the partner’s support. Each item was answered using a 4-point Likert scale (1 = not at all, 2 = a little, 3 = quite a bit, 4 = very much). Therefore, higher scores indicated more conflict and implied a lower-quality relationship. The QRI has strong reliability, validity, and factor structure in both clinical and nonclinical research (Verhofstadt, Buysse, Rosseel, & Peene, 2006; Reiner, Beutel, Skaletz, Brähler, & Stöbel-Richter, 2012). For the full MHFE-2 sample, Cronbach’s alpha was .97 for the T4 support scale concerning the FOB and was .92 for the T4 support scale concerning the current partner (not including any current partners who were also FOBs). Though the measure has been successfully applied to other types of relationships (e.g., peers, parents, mentors, etc.), it is primarily used in the assessment of romantic relationships (Ptacek, Pierce, Eberhardt, & Dodge, 1999; Reiner et al., 2012). The QRI was scored by
averaging participants’ answers across the seven items in the support scale and was employed in this study as a continuous measure. The QRI was administered at all time points of the MHFE-2 study, but the present study uses only the QRI data that were collected at T4.
Social-emotional experiences in childhood. Mothers’ social-emotional experiences in childhood were operationalized by measuring mothers’ history of childhood maltreatment and the quality of their parent-child relationships.
Perceptions of parent-child relationships. Mothers’ perceptions of their parent-child relationships and the care they received from their parents in childhood were measured using a
modified version of the Care subscale of the Parental Bonding Instrument (PBI; Parker, Tupling,
& Brown, 1979). The scale included 12 items, each of which asked about a behavior of their
“main caretaker” during their childhood. Respondents rated the frequency of each of the various behaviors on a 3-point scale (0 = “never,” 1 = “sometimes,” and 2 = “often”), with higher total scores indicating more positive care. The present study chose to measure PBI through
participants’ sum scores; thus scores ranged from 0-24, and PBI was measured as a continuous variable. The Care subscale has demonstrated strong reliability and validity across several studies, with Cronbach’s alphas ranging from .63 to .92 (Graue, Wentzel-Larsen, Hanestad, &
Søvik, 2005; Hall, Peden, Rayens, & Beebe, 2004; McDonald, Beck, Allison, & Norsworthy, 2005; Todd, Boyce, Heath, & Martin, 1994). For the full MHFE-2 sample, Cronbach’s alpha was .81. The Care subscale also has shown adequate test-retest reliability for up to 20 years
(Wilhelm, Niven, Parker, & Hadzi-Pavlovic, 2005). This measure was administered during the in-person interview at T2, and focused primarily on participants’ relationships with their own mothers.
Childhood maltreatment. The history of child maltreatment in a mother’s past was
assessed by collecting reports of abuse and neglect from state child protective services provided by the Massachusetts Department of Children and Families (DCF). The records included all reports filed between the birth of each participant and May 5, 2011. Mothers were then coded into two groups based on their childhood experiences with maltreatment: (1) no substantiated reports of maltreatment, or (2) any substantiated report of maltreatment (neglect, physical abuse, and/or sexual abuse).
Results
This section includes the results of the analyses conducted to examine the relationships between adolescent mothers’ social support, childhood social-emotional experiences, enrollment in a home visiting program, and depressive symptomatology. All results were evaluated using a p<.05 to indicate statistical significance, and p<.15 to indicate a statistical trend.
Depressive Symptomatology
Depressive symptomatology was assessed using the CES-D scale (Radloff, 1997). The average score across all participants was 12.17 (SD=10.74), just under 4 points below the clinical cutoff level of 16. However, 30.6% (n=88) of the sample of adolescent mothers who completed the CES-D (n=288) received a score greater than 16, indicating clinical levels of depression in these participants. Interestingly and contrary to my hypothesis, although the average depressive symptomatology of the HVS group was lower (M=11.44, SE=0.81) than that of the RIO control group (M=13.23, SE=1.01), the difference was not statistically significant, F(1, 286)=1.93, p=.165.
Social Support
Social support was examined through participants’ ratings of informal social support, formal social support, and partner support.
Informal and formal support analyses. Both informal and formal social support were measured using their respective PNM measures of frequency of contact with supports as well as their respective PNM measures of dependability of support. In order to isolate at-risk individuals, participants were categorized into “low” informal support groups or “medium/high” informal supports for both frequency and dependability of support based on their scores. For both informal support and formal support measures, one-way ANOVAs and Pearson’s chi-squared analyses were used to test for differences between these groups on several demographic factors:
median household income of the neighborhood block group, mother’s age at the birth of TC, TC’s age at T4 intake, program status (HVS, RIO), mother’s race/ethnicity, mother’s preferred language, mother’s relationship status at T4, mother’s employment status at T4, and mother’s education level at T4 to understand the relation between these factors and T4 CES-D scores.
Statistically significant and trend-level group differences were then considered as possible control variables for analyses. No trend-level difference was controlled for in a final model. Any statistically significant group differences that also had meaningful conceptual implications were then controlled for in four 2x2 mixed-model ANCOVAs used respectively to probe for a main effect of each measure of T4 informal and formal social support on adolescent mothers’ T4 CES- D scores as well as for an interaction effect of each measure of T4 informal and formal support and program enrollment status on adolescent mothers’ T4 CES-D scores.
Frequency of contact with informal support findings. Findings concerning mothers’
frequency of contact with formal support were informed by descriptive statistics of the measure, tests of group differences between low and medium/high groups, and a mixed-model 2x2 ANCOVA.
Descriptive statistics and group differences. On average, participants had 112.82
(SD=51.84) contacts with informal supports out of a total possible 320 contacts. As displayed in Tables 3 and 4, the only significant group differences between the low and medium/high
frequency of contact with informal support groups existed based on TC’s age at T4 intake, F(1, 238)=6.04, p=.015, mother’s relationship status, χ2(1, n=241)=9.67, p=.022, and mother’s employment status, χ2(1, n=241)=26.82, p<.001. A trend-level difference also existed between the groups based on mother’s education level, χ2(1, n=238)=0.11, p=.106. Upon further
consideration, the conceptually small difference in TCs’ ages at T4 intake between low and
medium/high frequency of contact groups was not meaningful in the overall analysis, and, thus, was not controlled for in further analyses. Additionally, because the frequency of contact with informal support measure includes frequency of contact with a partner, it was expected and logical that mothers’ frequency of contact with informal supports would vary according to their relationship status, so relationship status was not controlled for in this analysis. However, mother’s employment status was different between the two frequency of contact groups and was related to T4 CES-D scores, thus, I controlled for mother’s employment status as a covariate in all further analyses involving comparisons between the low and medium/high frequency of contact with informal support groups. Because mother’s education level was only significant at the trend level and because it was significantly related to mother’s employment status, χ2(1, n=283)=15.68, p<.001, I did not choose to additionally control for mother’s education level.
ANCOVA. A 2x2 mixed-model ANCOVA was used to probe for the interaction of T4
frequency of contact with informal supports and program status on T4 CES-D scores. The ANCOVA included T1 CES-D scores and mother’s employment status as covariates. There was no significant main effect of frequency of contact with informal supports on T4 CES-D scores, F(1, 234)=2.61, p=.108. Although the difference was non-significant, p=.108, the statistical trend concurs with my hypothesis, suggesting that mothers who had low frequency of contact with informal supports had higher T4 CES-D scores (M=14.42, SE=1.26) than mothers who had medium/high frequency of contact with informal supports (M=11.93, SE=0.83). There was also not a significant main effect of program status on T4 CES-D scores, F(1,234)=3.13, p=.078.
However, again, p=.078 suggests the evidence of a statistical trend that that RIO participants had higher CES-D scores (M=14.46, SE=1.16) than HVS participants (M=11.89, SE=0.89).
Furthermore, contrary to my original hypothesis, there was not a significant interaction between
mothers’ frequency of contact with informal supports (low, med/high) and program status (HVS, RIO) on T4 CES-D scores, F(1,234)=0.01, p=.910. These results are displayed in Figure 2.
Dependability of informal support findings. Findings concerning mothers’
dependability of informal support were informed by descriptive statistics of the measure, tests of group differences between low and medium/high groups, and a mixed-model 2x2 ANCOVA.
Descriptive statistics and group differences. Participants’ average rating of the
dependability of the informal supports with whom they were in contact was 1.65 (SD=0.78), with the lowest possible rating being zero and the highest possible rating being four. As displayed in Tables 3 and 5, the only significant group differences between the low and medium/high dependability of informal support groups existed based on mother’s employment status, χ2(1, n=221) =8.86, p=.003, and mother’s education χ2(1, n=238)=5.52, p=.019. A trend-level
difference also existed between the groups based on mother’s race/ethnicity, χ2(1, n=245)=6.15, p=.104. Because mother’s education level is related to mother’s employment status, and because there was a more significant difference between mothers when they were grouped by
employment rather than education, I controlled only for mother’s employment status as a covariate in all further analyses involving comparisons between the low and medium/high dependability of informal support groups. I also did not control for mother’s race/ethnicity because the groups were different from each other only at trend-level; additionally, the findings suggested that the White mothers had the greatest proportion of participants who rated their dependability on informal support as medium/high, which contradicts the consistent findings of previous research (Almeida, Molnar, Kawachi, & Subramanian, 2009; Maton et al., 1996).
Therefore, it is highly unlikely that difference in dependability of informal support across racial
groups is a meaningful difference that could be responsible for the variance in the T4 CES-D scores of these groups.
ANCOVA. A 2x2 mixed-model ANCOVA was used to probe for the interaction of
mothers’ T4 ratings of the dependability of their informal supports and participants’ program status on T4 CES-D scores. The ANCOVA included T1 CES-D scores and mother’s
employment status as covariates. There was not a significant main effect of dependability of informal support on T4 CES-D scores, F(1, 239)=3.51, p=.062. However, the statistical trend (p=.062) partially corroborates my hypothesis, suggesting that mothers with low dependability of informal support had higher T4 CES-D scores (M=13.93, SE=1.02) than mothers with
medium/high dependability of informal support (M=11.41, SE=0.86). The analysis also revealed that there was no significant main effect of program status on mothers’ T4 CES-D scores, F(1,239)=1.92, p=.168. Lastly, contrary to my hypothesis, there was not a significant interaction between dependability of informal supports (low, med/high) and program status (HVS, RIO) on T4 CES-D scores, F(1,239)=2.44, p=.120. However, p=.120 suggests the existence of a
statistical trend that, among mothers with low dependability of support, the difference in CES-D scores between HVS mothers (M=14.04, SE=1.23) and RIO mothers (M=13.81, SE=1.62) was less than the difference in CES-D scores between HVS mothers (M=9.45, SE=1.13) and RIO mothers (M=13.36, SE=1.30) among mothers with medium/high dependability of support. These results are displayed in Figure 3.
Frequency of contact with formal support findings. Findings concerning mothers’
frequency of contact with formal support were informed by descriptive statistics of the measure, tests of group differences between low and medium/high groups, and a mixed-model 2x2 ANCOVA.
Descriptive statistics and group differences. On average, participants had 22.17
(SD=18.85) monthly contacts with formal social supports out of a total possible 300 contacts. As displayed in Tables 3 and 6, the only significant group differences between the low and
medium/high frequency of contact with formal support groups existed based on mother’s employment status, χ2(1, n=249)=15.21, p<.001. Trend-level differences also existed between the groups based on TC’s age at T4 intake, F(1, 246)=2.12, p=.147, mother’s T4 relationship status, χ2(1, n=249)=15.21, p=.149., and median household income of block group, F(1,
247)=2.47, p=.117. I did not control for differences between low and medium/high frequency of contact with formal support groups based on TC’s age at T4 intake, mother’s T4 relationship status, or median household income of block group because each of these differences was significant only at the trend-level. Because mother’s employment status was significantly different across groups, I controlled only for mother’s employment status as a covariate in all further analyses involving comparisons between the low and medium/high frequency of contact with formal support groups.
ANCOVA. A 2x2 mixed-model ANCOVA was used to probe for the interaction of T4
frequency of contact with formal supports and program status on T4 CES-D scores. The
ANCOVA included T1 CES-D scores and mother’s employment status as covariates. Counter to my predictions, there was not a significant main effect of frequency of contact with formal supports on T4 CES-D scores, F(1, 242)=0.37, p=.543. There was also no significant main effect of program status on T4 CES-D scores, F(1,242)=2.31, p=.130. However, p=.130 suggests the existence of a statistical trend that RIO participants had higher CES-D scores (M=13.63, SE=1.02) than HVS participants (M=11.63, SE=0.83). Furthermore, contrary to my hypothesis, the interaction between frequency of contact with informal supports (low, med/high) and
program status (HVS, RIO) on T4 CES-D scores was non-significant, F(1,242)=1.41, p=.236.
Figure 4 displays a graph of these results.
Dependability of formal support findings. Findings concerning mothers’ dependability of formal support were informed by descriptive statistics of the measure, tests of group
differences between low and medium/high groups, and a mixed-model 2x2 ANCOVA.
Descriptive statistics and group differences. Participants’ average rating of the
dependability of their formal supports with whom they were in contact was 1.68 (SD=1.11), with the lowest possible rating being zero and the highest possible rating being four. As displayed in Tables 3 and 7, the low and medium/high dependability of formal support groups were not significantly different based on any of the tested variables. A trend-level difference existed between the two groups based on mother’s race/ethnicity, χ2(1, n=247)=7.31, p=.063; however, I did not control for mother’s race/ethnicity because the difference was significant only at the trend-level.
ANCOVA. A 2x2 mixed-model ANCOVA was used to probe for the interaction of
mothers’ ratings of the dependability of their formal supports and participants’ program status on T4 CES-D scores. The ANCOVA included T1 CES-D scores as a covariate. Counter to my expectations, there was not a significant main effect of dependability of formal support on T4 CES-D scores, F(1, 242)=0.07, p=.799. The analysis also revealed that there was not a
significant main effect of program status on T4 CES-D scores, F(1,242)=2.39, p=.123. However, p=.123 suggests the existence of a statistical trend that RIO participants had higher CES-D scores (M=13.54, SE=1.04) than HVS participants (M=11.49, SE=0.83). Lastly, contrary to my original hypothesis, there was not a significant interaction between dependability of formal
supports (low, med/high) and program status (HVS, RIO) on T4 CES-D scores, F(1,242)=0.71, p=.401. Figure 5 displays a graph of these results.
Partner support analyses. Partner support was examined using the QRI support scale, a continuous measure. Because partner support was evaluated using a continuous measure,
Pearson’s correlation tests and Pearson’s chi-square tests were used to test for variables that might be related to partner support. Pearson’s correlation tests were conducted to test for
significant associations between QRI scores and median household income of the neighborhood block group, mother’s age at the birth of TC, TC’s age at T4 intake. Strong correlations would indicate concern that these measures were not entirely independent of each other. Pearson’s chi- squared analyses were used to test for differences in QRI scores between groups based on participants’ program status (HVS, RIO), race/ethnicity, preferred language, mother’s T4 relationship status, mother’s employment status at T4, and mother’s education level at T4 to understand the relation between these factors and T4 CES-D scores. Three nested multiple regression analyses were then conducted to examine the predictive relations between T4 CES-D scores and T1 CES-D scores, QRI scores, program status (HVS, RIO), and the interaction of QRI scores and program status. When necessary, statistically significant and trend-level group
differences were controlled for by entering additional variables into the model.
Descriptive statistics and group differences. Mothers’ average rating of the support they
receive from their partner was 2.69 (SD=0.51), with the lowest possible rating being one and the highest possible rating being four. As displayed in Table 8, there were not any significant associations between QRI scores and the tested variables, and, as displayed in Table 9, there were not any significant group differences in QRI scores based on the tested variables.
Regression analyses. I conducted three nested multiple regression analyses to examine
the predictive relations between T4 CES-D scores and T1 CES-D scores, QRI scores, program status (HVS, RIO), and the interaction of QRI scores and program status. Table 10 summarizes and compares the results of all three models. First, I ran a linear regression which established that T1 CES-D scores statistically significantly predicted T4 CES-D scores, F(1, 285)=29.00,
p<.001, and that T1 CES-D scores accounted for approximately 9.2% of the explained variability in T4 CES-D scores. Next, I conducted a multiple regression analysis to test the main effects of QRI scores and program status on T4 CES-D scores. The multiple regression model statistically significantly predicted T4 CES-D scores, F(3, 127)=14.04, p<.001, explaining approximately 25% of the outcome variance. In partial support of my hypothesis, QRI scores added statistically to the prediction at a trend level, p<.15 (see Figure 6). Next, I conducted the third multiple regression analysis to test the effect of an interaction of QRI scores and program status on T4 CES-D scores. This multiple regression model significantly predicted T4 CES-D scores, F(4, 126)=10.48, p<.001, but did not explain additional outcome variance over and above the second model. Contrary to my predictions, the added interaction term did not significantly predict T4 CES-D scores.
Social-Emotional Experiences in Childhood
Mothers’ social-emotional experiences in childhood were examined through participants’
ratings of the quality of their childhood relationships with their parents and by checking agency data for substantiated reports of maltreatment in mothers’ pasts.
Mothers’ childhood relationships with parents analyses. The quality of mothers’
childhood relationships with their parents was examined using the PBI care subscale. Pearson’s correlation tests were conducted to test for significant associations between PBI scores and the