Research about the process and effects of family and community connections with schools is evolving and does not yet provide clear directions for practitioners. As discussed in Issues 1 and 2, there are unclear and overlapping definitions of the concept, its dimensions, and its measurements. The body of empirical work on family and community connections with schools should be strengthened in several respects to generate a solid research base for this field. Most critical is the development of more cohesive theoretical models and frameworks that can be used to develop and test hypotheses that can inform theory. The use of a variety of research methods and designs, appropriate for the types of questions that need to be explored, is also critical.

Current State of the Research

The body of research in family and community connections with schools is described as being at an early stage of development. Researchers are still trying to understand the overall patterns of these connections. Three specific characteristics that point to this early development are: the lack of linkages between research and theory, the limitations of methodology, and the disconnect between qualitative and quantitative research.

An early developmental stage

In their review of the literature on parent involvement in homework, Hoover-Dempsey, Battiato, Walker, Reed, DeJong & Jones (2001) explain that the current body of research consists primarily of descriptive accounts of what parents do when they are involved, what teachers or schools do to invite parent involvement, and what student outcomes are related to parent involvement. Others (Montemayor & Romero, 2000; Van Voorhis, 2000) point out that most parent involvement studies to date have been looking for family patterns and fixed characteristics, such as parent education, socioeconomic level, and relationships at home. Early studies have also often utilized only one measure to explain the construct, such as attendance at school events (Van Voorhis). These studies represent an effort to collect baseline information to understand current practice and its outcomes in family and community connections with schools.

Montemayor and Romero (2000) warn that focusing on family patterns has narrowed results to those characteristics that “good families” have that help children succeed academically. Schools then shape their programs around these results, such as by offering classes to teach “not so good parents” how to read to their children and help them with homework, in the ways that “good parents” do. From these studies researchers have portrayed families through deficit model lenses: some families are broken and need to be fixed. Usually, this pattern merely serves to reinforce racial, ethnic, and class biases (Montemayor & Romero).

Lack of theories and conceptual frameworks

The body of research in this field that has been developed over the last three decades has not been well connected to theory. One reason for this, explored in previous sections, is the lack of clear definitions and good ways of measuring outcomes. In addition, there have been few attempts to pull the research together into theoretical models and conceptual frameworks.

In our review, we found no theoretical models for community connections with schools. Some models have been developed that focus solely on parent involvement or that integrate family and community connections with schools. Kohl et al. (2000) have examined the strengths and weaknesses of several of these current models in the literature: Grolnick and Slowiaczek’s (1994, as cited in Kohl et al.) three dimensions of parent involvement, Eccles and Harold’s (1996) five dimensions of parent-initiated involvement, and Epstein’s (1995) six types of school-family-community partnerships (as described in section 1 of this document). Hoover-Dempsey and Sandler (1997) and Chrispeels (1992, as cited in Chrispeels & Rivero, 2000) have also presented models for parent involvement.

However, Epstein’s model is the only one that has undergone extensive review by the research community. Epstein’s model is based on a social organization perspective of overlapping influence, emphasizing that children are best supported when families and schools have shared goals and work collaboratively. This model includes the community as an important arena of child and adolescent learning. It views school, family, and community relations as dynamic, in that their overlapping spheres can be pushed together or pulled apart by important forces: background and practices of families, schools and communities; developmental characteristics of students; historical and policy contexts; and time (Epstein & Sanders, 2000; Simon, 2000). Families, schools, and communities are jointly responsible for and influential in children’s development.

The field has greatly benefited from Epstein’s model and most studies reviewed concur that the dimensions, or types of partnerships, are well-defined and provide useful guidelines for researching these connections. However, there are components to family and community involvement that are not part of the Epstein model. Kohl et al. (2000) for instance, point to the focus of Epstein’s model on teacher and school-initiated behaviors rather than parent-initiated involvement.

The other models have not received enough attention and have not been widely tested. As a result, when researchers have studied the relationship between student achievement and family and community connections with schools, they have often done so without links to a theoretical framework or model. The findings are difficult to compare and interpret. In addition, they do not inform theory and do not lend themselves to building upon each other. The quantitative studies we found about the relationship of family or community connections with schools to student achievement have offered important initial information to help frame this issue; however, they do not yield understandings of how, why, and under what conditions these connections are linked to student achievement. After uncovering relationships and developing theories and models, further experimental evidence needs to be collected to test those theories and give evidence of direct links of involvement to student success. Experimental designs are necessary to ascribe direct impact on student achievement to specific family and community involvement practices.

Challenges of methodology

In their review of the research, Epstein and Sanders (2000) took a historical look at the field and saw evidence of improvement and development in the research. They reported that researchers across the country and across disciplines have employed many methodologies, including surveys, case studies, experimental and quasi-experimental designs, longitudinal data collections, field tests, program evaluations, and policy analyses. Studies have grown from focusing mainly on preschools to elementary, middle, and high schools, and from focusing on what parents do on their own to what schools, families, and communities do in partnership. Studies have expanded from small, local samples to national and purposive samples of students and families with diverse racial and cultural backgrounds in urban, rural, and suburban locations.
Other authors take a different stance that emphasizes the problematic nature of the research. Fan and Chen (1999) conducted a meta-analytic synthesis of the literature about the empirical relationship between parent involvement and student academic achievement. They found that the vast proportion of literature is qualitative, and very few studies are empirically based. They were only able to include 25 studies that lent themselves to a statistical meta-analysis (which they recognized as presenting a limitation to their own study). Baker and Soden (1997) summarize the methodological flaws found to date in parent involvement research into four areas: use of non-experimental design, lack of isolation of parent involvement effects, inconsistent definitions of parent involvement, and non-objective measures of parent involvement. They also discuss the importance of program evaluations in this field. While recognizing that program evaluation may be the most challenging form of applied education field research, they complain that these studies tend to be among the weaker studies in the field, plagued by the flaws described above.

In our own review, we observed that researchers have faced numerous methodological challenges, including choice of design, sampling, measurement, and maintaining internal/external validity. For instance, researchers often relied on measures of perceived parent involvement instead of actual involvement (Reynolds, 1992; Catsambis & Garland, 1997; and others using data from The National Education Longitudinal Study of 1988; on only one perspective, usually the schools’ (teachers’ or principals’) perception of parent involvement (Izzo et al., 1999; Carey, Lewis & Farris, 1998; Fantuzzo, Tighe & Childs, 2000); on self-report surveys and questionnaires (Gutman & McLoyd, 2000; Sanders, Epstein & Connors-Tadros, 1999); or on retrospective information, when surveys or interviews ask information about involvement activities in the past (Miedel & Reynolds, 1999). These data collection strategies tend to distort or bias the findings.

Another challenge, tied to the lack of theories and frameworks, is that there are very few large-scale data sets that are longitudinal and reflect the kinds of questions that researchers need to address as they conduct deeper and richer studies. The National Education Longitudinal Study of 1988 (NELS: 88) is the most comprehensive data set on parent involvement, including data from parents of more than 24,000 eighth-grade students across the country. However, this data set has numerous limitations that contribute to the weakness of the many studies that rely on it. The NELS:88 does not include information about the initiator of contact, the length of the contact, or the quality of involvement. Only one of the top six parent involvement activities ranked as important to urban and minority parents and students (Xu, 2001) is included in NELS:88 indicators for parent involvement. The data set is not highly generalizable beyond the middle school age group. Finally, the NELS:88 are non-experimental data and can only be used to determine associations between variables, not cause and effect (Simon, 2000). Epstein and Lee (1995) suggest that researchers look at other national surveys and collect focused data in local, state, and regional surveys or field studies to assess the effects of particular parent involvement practices over time.

Disconnect between qualitative and quantitative research

Currently, the bulk of quantitative research focuses on the effects of parent and community connections, while the qualitative research focuses on processes and the “how to” of making connections. Most quantitative studies (Muller, 1998, and others using NELS:88 data) are testing details about impact without making connections to theory. Most qualitative studies (for example, Peña, 2000; Tapia, 2000) try to examine the factors or conditions that make these connections happen, but few are connecting this process-oriented information to results. Although many of these qualitative studies are using theoretical concepts as a point of departure for their research questions and design (for example, Mapp, 1999), few are attached to theory or models of family and community connections with schools. As a result, these two bodies of research do not inform each other well. Very few studies connect the information gleaned from quantitative and qualitative studies or are designed to cover the gamut of information that is needed to inform the implementation of effective practices.

Building a Stronger Research Base

Taken as a whole, the current body of research in family and community connection with schools has helped inform many facets of the field. However, as the previous section points out, there is a growing need for a stronger research base that can more clearly and definitively inform further research, theory, practice, and policy. In particular, this section highlights the two areas where more work would greatly help advance the research base: further theoretical development and a diverse and innovative approach to methodology.

Need for theoretical development

There is a critical need to take the body of research we have and build additional theoretical models and conceptual frameworks that can propel us into the next stage of research. Hoover-Dempsey et al. (2001) argue that research would benefit substantially from increasing use of theoretically-based predictions about involvement, saying:

Careful use of theory, the derivation of warranted hypotheses, and the design of studies enabling carefully crafted tests of hypotheses, promise considerable additions to our collective understanding, not only of what happens but also why it happens—e.g., through what mechanisms and under what conditions do specific elements of the parent involvement process influence critical student attributes and outcomes. (p. 10)

Family and community connection frameworks could test the relationship between different components, address the problem of unclear and overlapping definitions, and provide greater understanding of the predictors and impacts of connections. Researchers can begin to thoroughly examine a concept as they use theories and frameworks in a variety of settings, with a variety of samples ranging in population characteristics and size, thus isolating a variety of variables to understand the relationship between them. Achieving this breadth of scope is necessary to build a core body of research that can truly support practice.

Theorists and researchers have emphasized the need for a multidimensional conceptualization of parent involvement that accounts for the distinction between parent- and school-initiated parent involvement and relies on ratings by multiple reporters (Kohl et al., 2000; Ho Sui-Chu, 1997). Kohl et al. in particular believe that making this distinction may help explain some of the contradictory research findings that have associated parent involvement with both positive and negative outcomes. Further, a 1994 study by Kohl and colleagues (as cited in Kohl et al.) pointed to the quality of the parent-child relationship as being more strongly associated with child outcomes than the amount of parent contact. Based on this study, these researchers argue that the quality of the involvement is another important component of any theoretical framework or model of parent involvement.

However, none of the models found in the literature account for parent or teacher perceptions of the quality of involvement. Kohl et al. (2000) have developed a theoretical model and empirically validated it, yielding six reliable parent involvement dimensions. The first three relate to types and purpose of involvement: parent-teacher contact to facilitate monitoring of a child’s school progress and homework assistance, parent involvement in school activities, and parent involvement directly with a child at home to facilitate intellectual stimulation and school success. The second three aim at measuring the quality of parent involvement: the quality of the relationship between parent and teacher, the teacher’s perception of the parent’s value of education, and the parent’s satisfaction with the child’s school. This model has not yet been used in studies by other researchers.

There is also a need for hierarchical models for defining involvement and outcomes. It is important that these models consider the full range of definitions and outcomes of school, family, and community connections, as described in the previous sections of this document. Most current frameworks tend to emphasize programmatic or activity-oriented parent and community involvement. Relationship-building elements and other process-based aspects of family and community connections should be included (Hirota et al., 2000; Mapp, 1999). The variety of outcomes of these connections, including the impact they have on all stakeholders, should also be featured in the development of new theoretical models. Community organizers for school reform have also called for the development of indicators that reflect the outcomes and the process of their work (Lewis & Henderson, 1998).

It is also important that new models consider local context, including geographic, socio-economic and cultural contexts. It is unlikely that one model can explain the interactions between all communities, families, and schools. The challenge for researchers and theorists is to create models that are well-informed by local realities and experiences and that are flexible enough to adapt to the local needs. Models have been developed that take into consideration particular cultural and geographic characteristics, but they have received very little attention in research. The Intercultural Development Research Association (IDRA), for example, has created a parent-centered model and process that focuses on parent leadership in a child’s education. It encompasses four types of parent involvement with schools: parents as teachers, parents as resources, parents as decision-makers, and parents as leaders and trainers (Montemayor & Romero, 2000). Each type values and acknowledges the assets that families from all cultural backgrounds, socioeconomic groups, and geographical areas bring to their children’s education.

Finally, additional conceptual models for family and community connections with schools should draw from theoretical advances, concepts, and ideas from other disciplines beyond education. Researchers have started to expand their thinking in particular studies with positive results, but have not yet incorporated these ideas into models than can be further tested by others. For example, researchers have used social capital theory (Coleman, 1988, as cited in Ho Sui-Chu, 1997) as a way to understand and study the strategies that are needed to integrate family and community involvement in the change process in urban public schools. One study (Shirley, 1997) has measured social capital by the amount of concentrated and stable adult supervision and tutoring of children at home and the density of social ties between students, parents, teachers, and the community. Another researcher (Ho Sui-chu, 1997) has proposed a conceptual framework that indicates the relationships between school factors, family factors, parent involvement, and students’ learning outcomes by using the construct of “capital” (economic, political and social). Social capital could prove to be a very useful concept in developing a model that emphasizes process- or relationship-focused forms of connections.

Urban ecology of schools is another rich theoretical area that holds promise for creating models that are flexible and account for local context. Researchers are pointing to the need to explore the greater urban context in which schools exist to fully understand the connections between changes in the urban environment and their effect on schools (Kantor & Brenzel, 1993, as cited in Bartelt, 1997). Recent work highlights empirical relationships between forces affecting the ways in which cities grow and decline and educational development (Bartelt; Yancey & Saporito, 1997). Researchers must recognize the close relationship between an economic situation, family structure, and educational participation. Situations such as families with single parents and parents with several jobs also need to be addressed to inform school, family, and community involvement practices.

Need for a diverse and innovative approach to methodology

Although conducting research in this field is extremely challenging, as noted by Baker and Soden (1997), the development of theories and conceptual frameworks will help overcome some of the challenges of methodology, as researchers will be able to develop more precise and well-informed research questions. At that point, researchers will then be able to select the most appropriate design, taking into account its inherent limitations to make adjustments appropriate to the particular study. Addressing other methodological issues, such as measures, samples, internal/external validity, and analysis, will also contribute to a better body of research.

Qualitative and quantitative methods inform different facets of the research. Qualitative studies paint a rich, local picture and lend insight to the process. However, by design they do not tell us if the trend extends beyond the observations. These results should inform theory and conceptual development as well as subsequent quantitative studies that would indicate if these are broader trends. Baker and Soden (1997) openly advocate for the use of the true experiment as the design that adequately overcomes all threats to internal validity in education research. The critical component of this design, random assignment to the control and experimental groups, rules out pre-test differences between groups, so that post-test differences can be attributed to the independent variable. However, Dryfoos (1998a) maintains that experimental designs with random assignment are not feasible in school settings, and that finding and maintaining control groups is arduous and expensive. For example, schools that are located in disadvantaged neighborhoods can have turnover of students as high as 50 percent in one year, making it difficult for a researcher to keep the random experimental and control groups intact. Schorr (1997) also discusses the difficulty of experimental designs using random assignment as the only sources of reliable knowledge. She argues that other designs, considered “flawed alternatives” to experimental design, may provide less certainty about the cause of observed effects, but do offer a broader range of information that may be more useful in making judgments about what really matters.

To ease researchers’ ability to compare their findings with work of others, and to build upon existing knowledge in a systematic fashion, researchers will need to develop and validate common instruments for measuring parent involvement across a variety of settings. Some researchers argue that it would be better to use direct observation of parental behavior. New assessment tools such as the Family Involvement Questionnaire (FIQ) (Fantuzzo et al., 2000) are being constructed to study multiple parent involvement variables. The FIQ has met multiple construct validity criteria and yielded three stable dimensions of parent involvement: school-based involvement, home-based involvement, and home-school conferencing.
Other authors argue that because the process and impact of parent and community connections with schools is so complex, standard research methods and indicators for measuring the impact of connections are not able to capture information fully (Honig et al, 2001; Montemayor & Romero, 2000; Schorr, 1997). For instance, Honig et al. point out that most of the current research reduces community contexts to uni-dimensional indicators, such as poverty rates, violence, and the number of community organizations, and correlates these with other indicators of child/youth development and learning. This approach can be problematic because it focuses on the correlation of economic and other resources and relationships to outcomes, and not on what enables children and youth to take advantage of these resources and relationships in ways that may lead to favorable outcomes.

In other cases the challenge involves isolating the effects of a single factor from others. This becomes even more difficult when the effect of a factor is not necessarily directly related to a measurable outcome. Montemayor & Romero (2000) illustrate this challenge when they describe the difficulties of examining the role and measuring the effects of parent leadership in education, in the midst of other factors. Most would agree that parents ultimately make a critical difference in ensuring the quality of public education for their children by taking on leadership roles, but current research methodologies do not allow for the full measurement of the results of these actions.

Some researchers are beginning to make use of new advances in statistical methodology, tools, and technology that contribute to better quantitative research. Recent studies (Ho Sui-Chu, 1997; McWilliam et al., 1999) have used Hierarchical Linear Modeling, which allows researchers to look at multiple factors and outcomes simultaneously. Evaluation studies of the more formal strategies, programs, or initiatives in the field (for example, Desimone et al., 2000) have also been a source of data on both process and outcomes.
Schorr (1997) presents at least four attributes that new approaches in methodology should have: (a) they build on a strong theoretical and conceptual base, (b) they emphasize shared interests between researchers and subjects, (c) they employ multiple methods and perspectives, and (d) they offer both rigor and relevance. She comments that using theory as a starting point is in the finest tradition of social sciences, where it is important to “construct conceptual maps that link one thing to another” (Knapp, 1995, as cited in Schorr). When it comes to disentangling such complex forces as the effects of communities, families, or schools on children, parents, or school staff, the most powerful tools are not statistical but conceptual. Therefore, it is essential to ground both design and measurement in theory. Theory-based methodologies help us determine what is working in situations where statistical analysis alone cannot provide the needed answers. Combining outcome measures with an understanding of the process that produced the outcomes can shed light both on the extent of change and on how the change occurred (Schorr).

Summary

What practitioners and policymakers really want is a single breakthrough study that resoundingly and unequivocally provides both concrete evidence about the impact of family and community connections with schools and a recipe to make it happen. But this is not likely to happen. New developments in research design and methodology that better link quantitative and qualitative research, and more and improved conceptual models to use in the research, can move the field toward a stronger research base. Epstein and Sanders (2000) believe there is much to learn and that:

As research proceeds and improves, researchers must continue to ask deeper questions, employ better samples, collect useful data, create more fully specified measurement models and conduct more elegant analyses to more clearly identify the results of particular practices and partnerships. (p. 290)

They also say it is important to conduct research that improves education policies and school partnership practices. Studies are needed at all grade levels, in differently organized schools, in varied locations, and with students and families with diverse racial, cultural and linguistic backgrounds, as suggested by Epstein and Sanders (2000). Baker and Soden (1997) outline several advances that would support some of the improvements needed in the research: funding allocations to applied educational research and program evaluations must increase, a new level of partnership must be forged between practitioners and researchers to enable the use of experimental procedures in service settings, and program staff concerns related to random assignment and potentially intrusive data collection procedures must be addressed.