Involvement in Early Head Start home visiting services: Demographic predictors and relations to child and parent outcomes

Abstract

One strand of home visiting research investigates efficacy while another investigates under what conditions programs achieve outcomes. The current study follows the latter approach. Using a within-program design in a sample of 11 home-based sites in the Early Head Start Research and Evaluation study, this study found that three components of home visits (quantity of involvement including number of home visits, duration in the program, length of visits and intensity of service; quality of engagement including global ratings of engagement by staff and ratings of engagement during each home visit; and the extent to which home visits were child focused) represented distinguishable aspects of home visit services. Demographic variables predicted components of involvement, and home visit involvement components were differentially related to outcomes at 36 months, after controlling for demographic/family factors and earlier functioning on the same measure. Only one quantity of involvement variable (duration) predicted improvements in home language and literacy environments at 36 months. Quality of involvement variables were negative predictors of maternal depressive symptoms at 36 months. Finally, the proportion of time during the visit devoted to child-focused activities predicted children's cognitive and language development scores, parent HOME scores, and parental support for language and learning when children were 36 months of age. Implications for home visiting programs and policies are discussed.

Introduction

One prevailing view of home visiting programs discounts their efficacy (Chaffin, 2004) while another view finds support for program effects under some conditions or for some subgroups. Related to this latter view, some have suggested that the important question may not be whether the programs work but under what conditions (Gomby, 1999; Sweet & Appelbaum, 2004). This approach suggests studying variations in the ways that parents are involved in the programs. However, for the most part, parent involvement variables tend not to be included in studies of home visiting programs (Gomby, Culross, & Behrman, 1999). In this study, we used data from 11 sites that participated in the Early Head Start Research and Evaluation Project, an experimental design study, to examine more closely the role of parent involvement in home visiting within the program group. The study operationalizes a conceptual model of home visiting for early childhood intervention programs provided by Korfmacher et al. (2005) and measures three components of parent involvement in home visiting services—quantity of involvement, quality of engagement, and visit content (the extent to which visits were child-focused). It examines specific variables that measure these components; examines demographic predictors of the variables; and relates these variables to child and family outcomes, both before and after controlling for key demographic predictors and earlier functioning on the child and family outcomes.

Home visiting is a service delivery mode that serves substantial numbers of children and families in the United States today. Between a quarter and a half million preschool-age children and their parents were enrolled during 2004 in five of the largest programs serving families of young children through home visits (Early Head Start; Healthy Families America, HIPPY, Parents as Teachers, and the Parent Child Home Program; Home Visiting Forum, 2004). If for no other reason than because home visiting serves so many children and families, it is important to better understand under what conditions it is an effective service strategy.

Some studies and reviews suggest that home visiting as a service delivery mechanism is generally not efficacious (Chaffin, 2004, Gomby, 1999) and that home visiting may not represent a viable early intervention option. However, a closer look at outcomes in home visiting programs shows favorable effects on a wide range of parent behaviors and, to some extent, on child outcomes. Effects on parents include improvements in maternal health and reduced subsequent pregnancies (Kitzman, Cole, Yoos, & Olds, 1997; Olds et al., 1998), increased parental reading to children (Johnson, Howell, & Molloy, 1993); greater reliance on non-violent discipline (Duggan et al., 1999; Heinicke, Fineman, Ponce, & Guthrie, 2001); greater verbal responsiveness and provision of stimulating activities (Black et al., 1994); increased sensitivity in interactions (Olds et al., 2002); greater emotional coherence and expressiveness (Olds et al., 2004); less instrusiveness and more support for autonomy (Heinicke et al., 2001); reduced domestic abuse (Duggan et al., 1999); in a population of clinically depressed mothers, reduced depressive symptoms (Gelfand, Teti, Seiner, & Jameson, 1996); decreased parental stress (Duggan et al., 1999); in a sample of substance-dependent mothers, reduced drug usage (Black et al., 1994); and less child maltreatment (Daro & Harding, 1999; Olds et al., 1997, Olds et al., 1998; Olds, Henderson, Kitzman, & Cole, 1995; Wagner & Clayton, 1999). In addition, the Early Head Start evaluation, the source of data for the current study, reported parents in home-based programs were more supportive in interactions with their children and had less parenting stress than control group parents when children were 36 months of age (Administration for Children and Families [ACF], 2002; Love et al., 2005).

Direct effects on children are less robust, perhaps in part because home visiting programs often initially target parents in order to have long term effects on children (Sweet & Appelbaum, 2004). However, some direct effects on children have also been found. These include reduced doctor/hospital visits for accidents and injuries (Kitzman et al., 1997); increased immunizations and improved nutrition (Johnson et al., 1993); among low birthweight babies, better health outcomes (Brooks-Gunn et al., 1994); fewer behavior problems (Aronen & Kurkela, 1996; Butz, Lears, & O’Neil, 2001); better emotional functioning (Heinicke et al., 1999; Jacobson & Frye, 1991; Van den Boom, 1995), and more secure attachment relationships (Heinicke et al., 2001; Lieberman, Weston, & Pawl, 1991). While multiple evaluations of home-based programs have not found child language/cognitive effects (Baker, Piotrkowski, & Brooks-Gunn, 1999; Wasik, Ramey, Bryant, & Sparling, 1990), outcomes in these domains have been reported in subgroups or specialized samples in rigorous studies (Beckwith, 1988, Black et al., 1994, Olds et al., 2002). The Early Head Start evaluation reported children in home-based programs were significantly better at engaging their parents during videotaped play activities than control group children. Within home-based programs that fully implemented the comprehensive Head Start Program Performance Standards (achieved high standards for quality and quantity of services aimed at enhancing child development as well as family functioning), Early Head Start children showed greater cognitive and language ability when compared to control group children (ACF, 2002) with effect sizes in the .20–.25 range for these cognitive and language outcomes. While home visiting programs have demonstrated some positive child outcomes, especially among youngest children, some reviewers argue that center-based programs are more likely to demonstrate child effects (Benasich & Brooks-Gunn, 1996). However, interventions for children under three (Brooks-Gunn et al., 1994) have often included home visiting/family support components as well as center-based components, illustrating that delivery of home visiting and center-based services may not be as separate as sometimes appears to be the case.

A review of the literature (see Gomby et al., 1999) and a recent meta-analysis (Sweet & Appelbaum, 2004) suggest the operative question may not be whether home visiting is an effective service delivery strategy but under what conditions programs achieve effects. Enormous variability in the purposes, services and outcomes both across and within home visiting programs has been documented (Sweet & Appelbaum, 2004). The differences, as well as the similarities, have been framed by the National Home Visiting Forum, consisting of six major home visiting programs in the US today (Nurse Home Visiting Program, Parents as Teachers, HIPPY, the Parent–Child Home Program, Healthy Families America and Early Head Start; described in Weiss, 2004). The task of addressing variation is complex and includes the need to identify salient features or meaningful components of home visiting; the need to understand how involvement may vary for different population subgroups and the need to learn more about the importance of involvement in relation to outcomes. The current study addresses each of these dimensions of variability.

The home visiting field, as is true for some other service fields, has had difficulty operationalizing and conceptualizing meaningful components of services. Littell, Alexander, and Reynolds (2001) provide five examples of operationalizations of a “simple” dichotomous indicator of program “participation” (e.g., receipt of any service; receipt of any service beyond intake/assessment; completion of a service; receipt of services for a specified time period; and receipt of a specified number of services). Littell et al. (2001) also found that most research studies use only a single measure of parent involvement, and that these measures vary widely and tend to focus on either: (1) dichotomous measures of participation (did the client receive services or not?); (2) measures of service frequency or attendance; or (3) ratings by the provider of participants’ level of parent involvement. Measures in the first two categories reflect simple or more directly-measured quantity of services, while clinician ratings’ may be more likely to include subjective judgment on the part of the provider that may reflect the emotional investment of the client. The authors found few studies that conceptualize parent involvement as a multi-dimensional construct.

Kazdin, Holland, and Crowley (1997) and Kazdin and Wassell (1999) have suggested a multidimensional model of mental health service participation that captures aspects of both amount and valence of participation; measures of both of these dimensions may be critical to understanding the relationship of parent involvement to program outcomes.

Conceptual models have helped to guide the measurement of program services in medical and social work fields, but there have been fewer conceptual frameworks advanced in the home visiting field. Recently, Korfmacher et al. (2005) identified two major dimensions of parent involvement in home visiting programs: the quantity of involvement and the quality of engagement in home visiting services. These two broad areas subsume other components, many of which are potentially measurable. Quantity of involvement variables include number of home visits, duration of enrollment in program services, and total hours of home visiting services received. Quality of engagement includes variables such as staff ratings of how engaged parents are in program services generally, interest and engagement of parents in particular home visiting activities, and the quality of relationships between staff and parents. Thus, the Korfmacher and colleagues model includes both relatively simple constructs such as duration of enrollment in services as well as more complex, dynamic conceptualizations related to the parents’ motivation and actions in partaking in program services. In our study, engagement is thought to reside in the parent, although it may be observed and promoted in the context of transactions between staff and parent. The Korfmacher model provides the basis for the current work, which explores these theoretical constructs empirically. The current work is also informed by Wagner, Spiker, Linn, Gerlach-Downie, and Hernandez (2003) whose conceptual approach characterizes both quantity and quality of service components and by McCurdy and Daro (2001), who developed a more general conceptual model of factors related to enrollment and retention of participants in family support programs.

Different measures of parent involvement reflect different perspectives of those involved in a program. Staff reports of client involvement in services, attitude towards services, and rates of compliance or attendance are common. Fewer studies have examined participation from the point of view of the service recipient (Littell et al., 2001). Further, there has been little systematic study of the relative merits of different modes of data collection, despite the fact that some methods may be considerably easier from a data collection standpoint for some types of programs or evaluations than others. For example, are global staff ratings of a parent's participation sufficient, or must each service event be documented? Are parents reliable reporters of the amount of services they have received? How many types of data must be collected to represent home visiting services? In the current study of multiple dimensions of home visiting, we explore a variety of data collection modes in a diverse sample.

What do we know about who participates in home visiting programs? Wide variability in home visiting involvement across population groups has been reported (Daro & Harding, 1999; McCurdy, Hurvis, & Clark, 1996; McGuigan, Katzev, & Pratt, 2003). Families more advantaged from the standpoint of income and education tend to be more involved in home visiting, while high risk families appear to be the most difficult to serve successfully (Larner, Halpern, & Harkavy, 1992; Wagner et al., 2003). Luker and Chalmers (1990), for example, found mothers with limited maternal support were more likely to withdraw from the program early. Lower rates of involvement have been found among mothers who experienced family conflict (Herzog, Cherniss, & Menzel, 1986), were substance abusing (Navaie-Waliser et al., 2000), or were anticipating a change of residence (National Committee to Prevent Child Abuse [NCPCA], 1996), and higher rates have been found among mothers whose infants displayed health risks (Olds & Kitzman, 1993).

Some demographic/family factors seem to be unrelated or inconsistently related to family retention in home visiting programs, e.g., race/ethnicity, teen parent status. Herzog et al. (1986) did not find ethnic differences, however, one study found higher recruitment and retention rates for Hispanic parents (Dumka, Garza, Roosa, & Stoezinger, 1997). Wagner et al. (2003) reported lower rates of recruitment for African American parents, lower rates of attendance but higher rates of follow-up among Hispanic parents and higher levels of most forms of involvement among white parents (Wagner et al., 2003). Findings by age of parent have been equivocal, with some studies showing more participation among younger parents (Herzog et al., 1986; Olds & Kitzman, 1993) and others showing less (Birkel & Repucci, 1983). In the current study, we will measure the relation of mother's marital, employment/schooling, educational, cash assistance, and teen parenting status to home visiting involvement. We will also examine relations to race/ethnicity, maternal verbal ability, whether the target child has a disability, and residential mobility. We will explore how each relates to multiple home visiting involvement components. This is similar to a strategy employed by Wagner et al., who reported differential relations between demographic characteristics and types of engagement in a home visiting program.

Studies linking variability in services in early childhood programs to outcomes tend to concentrate on service dosage globally (Hill, Brooks-Gunn, & Waldfogel, 2003; Ramey et al., 1992). For example, these studies investigate whether (or not) families are involved in the program but a few go to the next level, as does the current study, to examine specific components of home visiting as predictors of outcomes. An exception to the general literature, Sweet and Appelbaum's meta analysis of home visiting (2004) demonstrated that cognitive outcomes varied as a function of specific components of involvement, including whether programs had professional staff, by number of home visits and number of hours of home visits, while parenting outcomes and parenting education were not affected by these inputs. Also, Powell and Grantham-McGregor (1989) reported that urban Jamaican 16–30-month-old children visited weekly in their homes had better outcomes on the Griffiths Mental Development Scale than children visited less often and than control group children. Thus, there is some support for a link between number and intensity of home visits and child cognitive/developmental outcomes but links between other varying features of home visiting (e.g., engagement of parent, child-focused activity) and child outcomes and between most component measures and parenting outcomes are basically unexplored or unsubstantiated. We explore some these links in the current study using different measures of involvement.

A first step toward greater understanding of how involvement components affect outcomes is to identify and collect a variety of different process measures and empirically examine their relations to each other and to outcomes. In the current study, seven measures of parent involvement are measured. Some of the measures are from parent report and others are from staff report. Following the Korfmacher et al. (2005) framework, the measures include indicators of quantity of involvement, such as number of home visits, duration in the program, the extent to which families received the services prescribed by the program, and length of visits, as well as measures of more subjective aspects of parent involvement that relate to the quality of engagement in the services, such as staff ratings of global engagement across the family's time in the program and rating of engagement during specific visits. Finally, because Early Head Start is a child development program and seeks to have positive effects on children's development, we include a measure of the content of the visit, specifically, the extent to which the visits focused on activities directly related to the child. Thus, the first study question is: Are there multiple and distinguishable components of home visiting services, related to quantity, quality and content? We hypothesize that quantity measures will be correlated with each other and not with other measures, that quality measures will be correlated with one another but not with other measures and that the content measure will be independent of other measures. Altogether we propose that these three areas will represent distinct components of home visiting service provision.

A second part of the current study explores how these parent involvement components are related to family characteristics. It is generally understood from the home visiting literature that families are differentially involved in home visiting programs. Moreover, parent characteristics may predict different types of involvement. Some families may stay involved in the program over a long period of time but may not ever seriously engage in intensive services. Families with other characteristics may be very involved for a short period of time. Factors such as maternal age, education, occupation, welfare involvement, verbal ability, race/ethnicity as well as child disability and mobility might all be expected to affect parent involvement in a home visiting program. Impacts on service receipt were found to vary to some extent according to family demographic factors in the national Early Head Start overall study of impacts (e.g., for some subgroups the difference between the program and control groups in percent of families who received weekly home visits was greater than for others) (ACF, 2002). Here, we further explore relations between demographic/family factors and parent involvement in home visiting. Thus, a second purpose of the study will be to address the question: Do demographic/family characteristics predict components of home visiting services? We hypothesize that maternal risk factors (teen parenting, no high school, welfare receipt, not being married, not being in school or working, housing mobility, low levels of verbal ability) will be associated with reduced involvement in home visiting as has been found in other studies (Duggan et al., 1999, Herzog et al., 1986, Larner et al., 1992, NCPCA, 1996), and that Hispanic parents and parents of children with disabilities will be more engaged in services than others (Dumka et al., 1997).

Finally, we examine whether different parent involvement components are themselves differentially related to child and family outcomes. Understanding which aspects of involvement are most importantly related to child and family outcomes will enable programs to fine-tune program services to achieve desired outcomes (Guralnick, 1997). Specifically, the third research question is: Are child and family outcomes predicted by home visiting components, before and after controlling for demographic factors and earlier functioning on the outcome measure? We hypothesize that before controlling for demographic factors, many service variables may be related to outcomes and when demographic variables, as well as earlier functioning, are controlled, quantity and quality of service may still relate to some outcomes. Based on the work of Korfmacher et al. (2005), we hypothesize that quality of engagement will be particularly related to parent–child interaction and mental health. We also propose that child-focused activity will affect child outcomes based on findings showing the closer the intervention to the child, the greater the likelihood of achieving child outcomes (Benasich & Brooks-Gunn, 1996). However, because few studies similar to the current one have been conducted, it is necessary to regard the current work as largely exploratory.

By answering these questions using a variety of involvement measures, we hope to be able to make recommendations about the kinds of involvement indicators that might be of greatest use to programs or researchers. The current study does not attempt to control for all program features that could affect child or parent outcomes because the number of programs in the study is relatively small; there may be other within-program, program-level, community-level or regional features besides those measured in this study that could contribute to outcomes.