The project on indicators of the social context of families is organized around two complementary perspectives: the ecological model (Bronfenbrenner, 1979) and a lifecourse framework (Elder, 1998). The ecological model notes that families are nested within and are affected by a number of influences, which range from quite distal factors at the societal level (macrosystems) to factors at the community level (exosystems) and then to more proximal factors in the immediate neighborhood or family (microsystems) (Coatsworth, 2002). The ecological model has recently been updated to include not only a consideration of the characteristics of persons, processes, and contexts, but also the consistency of these over time (Bronfenbrenner and Morris, 1998).
Proceeding from an ecological perspective, longitudinal indicators might be developed to reflect the stability of multiple ecological contexts of family life, ranging from distal factors such as the community, labor force, and media to more proximal factors such as the neighborhood, extended family, religious institutions and schools, to very proximal factors such as the family's interaction patterns and parenting.
Overlaying the ecological perspective is a lifecourse perspective, which emphasizes how the lives of families and individuals evolve and change over time and with development. Thus, a family with several preschool children is in a quite different situation than a family with several adolescents who are finishing secondary school and moving toward independence.
Social contexts for which indicators could be developed
Based on these complementary conceptual frameworks, a number of longitudinal indicators could be developed across the social contexts of families, including family structure; family functioning; family, work, and child care; school involvement and civic engagement; religiosity; youth development; and social connections.
Family Structure. Research indicates that events such as marital disruption can have negative implications for children (Moore, Morrison, and Glei, 1995). Moreover, research indicates that status distinctions such as whether biological parents are married or cohabiting are related to children's development (Seltzer, 2000). However, duration can provide important additional information. For example, children who live with both of their biological parents for a longer proportion of childhood enjoy numerous economic, social and psychological advantages. A measure of the number of years that parents remain together from the time that their first child is born provides this information. Such a measure could assess either the number of years of marriage or years of co-residence. Measuring the actual number of years is most successful for families in the same or similar life cycle stage, such as families with teenagers. If couples from different life cycle stages are combined, a measure of the proportion of time that parents remain together might be preferred.
A measure of consistent father involvement represents another aspect of family structure. A measure of the number of years that a father either resides with his child(ren) or remains in regular contact with his child(ren) if he does not reside in the household picks up a different construct than the duration of marriage or co-residence. Given the large number of children who do not continuously live with their biological father, it would be useful to understand how many fathers never live with their children, how many always do, and how families are distributed between these poles. Alternatively, assessing the proportion of time that fathers live with their children would yield a measure that is comparable across families in different life cycle states.
Family Functioning. Numerous aspects of family functioning have been found to be related to the well-being of the adults and children in a family, including family routines, the quality of the parents' marital relationship, the quality of the parent-child relationship, monitoring and supervision of children, and family communication. From a lifecourse perspective, it is important that these aspects of family functioning be ongoing in age-appropriate ways over time. For example, maintaining family routines consistently over time, sustaining family communication over time, being aware of and monitoring children's activities and friends as they grow older, and maintaining positive parent-child relationships as children age represent important aspects of family functioning that are not "one-shot" efforts but ongoing commitments.
Family, Work and Child Care. Secure parental employment provides an important base for family life. The absolute level of income is to be distinguished from the stability and dependability of that income. Not only is poverty associated with poorer child outcomes (Duncan and Brooks-Gunn, 1997; McLoyd, 1998), but so too is inconsistent income. For example, fluctuations in income have been linked with out-of-wedlock childbearing (Wu, 1996), as well as with lower reading and math achievement and higher behavioral problems among elementary school-age children (Moore et al., 2002).
Stable parental employment status can also be important. Some research has found that instability in maternal work status (a change between more, fewer, or no hours at all) is negatively linked with children's achievement and behavior in school (Moorehouse, 1986, as cited in Bronfenbrenner and Morris, 1998). While the largest effects were for children of mothers moving into full employment, the study also found negative effects of reductions in work hours or leaving the work force.
Consistent, dependable child care arrangements represent another longitudinal construct that affects adults, children and the family as a whole. Poorer development among young children has been linked with having a large number of child care providers or frequent changes in providers (Hayes et al., 1990). When child care arrangements fall through, parents may miss work and children may be cared for in a patchwork set of arrangements or even left in self-care.
As part of the Project on Child Well-being at the State Level, the construct of "turbulence" was developed. Turbulence assesses multiple changes in various domains of a child's life, such as changes in family structure, residence, school or child care arrangements, or fluctuations in family income (Child Trends, 1999). It has proven to be difficult to measure this construct with cross-sectional data, but it could be measured with longitudinal data that track family changes over time.
School Involvement and Civic Engagement. Attendance at signal moments such as graduation and pageants represents a different level of family involvement in the school than ongoing attendance at parent-teacher association meetings, volunteering in the classroom, and regular meetings with teachers. Continuous involvement in school keeps parents in touch with a central socializing institution in their child's life. Rather than coding involvement as high, medium or low at a particular time, such involvement might be coded over a period of years. Also, given the tendency for many parents to become less involved when their children are adolescents (Zill and Nord, 1994), sustained family involvement through the adolescent years may foster stronger academic ambitions and better school performance.
Similar contrasts in levels of involvement can occur with civic engagement, with some adults voting only in presidential elections, for example, and other adults voting consistently in off-year elections as well. Similarly, sustained volunteering or community work over a period of years is quite different than participating in a one-time event, such as a morning walkathon (desirable as that may be). Obviously, even occasional involvement is good for the community, but sustained involvement has more potential to influence family processes and overall life style. That is, sustained engagement in civic life and community activities is not just a positive contribution to the community but represents a role model for children. However, the value of sustained engagement outside the family might vary depending upon the level of activity and the age of children in the home. We are not aware of research that explores this issue, but we speculate that, when children are young, modest levels of family engagement outside the home might be reasonable, while, when children are older, greater levels of engagement and activities that involve the family as a whole might be very positive.
Religiosity. Occasional attendance on special occasions and holidays represents a very different level of religious involvement than fidelity over time. Families that attend services or classes together, say grace or blessing or read religious texts together, watch religiously-oriented television or videos together, and engage together in activities sponsored by a religious organization over a period of years demonstrate an interest and commitment. Some research suggests that parental importance of religion translates into greater religious involvement by their children (Gunnoe and Moore, 2002), and numerous studies link family religious involvement with lower levels of risky behavior on the part of children (Bridges and Moore, 2002).
Youth Connections. Consistent engagement in positive activities during the high school years has been found to predict better outcomes in early adulthood (Zaff et al., 2003). Many activities require an ongoing investment before proficiency is achieved. For example, playing a musical instrument, singing, playing a sport, debate and dance all require ongoing training, practice and performance. Hence, a longitudinal measure of sustained involvement in some kind of youth development activity should theoretically represent a stronger measure than involvement at a moment in time.
Social Connections. Inherent to the notion of being connected to friends, neighbors, and social institutions is the presumption that this relationship is sustained over time. Indeed, social capital is conceptualized as a sense of trust and mutuality that is built up over time and sustained by means of ongoing interaction (Coleman, 1988). Similarly, friends and neighbors on whom families can rely are generally those who have built up the relationship over time. Of course, in disorganized and violent neighborhoods, such potentially harmful associations may be avoided (Mekos, forthcoming). Moreover, even in a stable neighborhood or voluntary association, there is turnover, so it cannot be assumed that social connections are absolutely static. In addition, long-term residence or membership does not necessarily mean that social connections exist. Rather, time may be a necessary but not a sufficient circumstance to create the conditions in which such social connections may develop. One aspect of this construct that has been assessed, though typically only at a point in time, is the presence of social support, which is generally associated with better outcomes for individuals (Sampson, 1991), though we have not yet identified studies of family-level outcomes.
Long-term residence in a low-income neighborhood or in a high-crime neighborhood would be valuable indicators of a difficult context for family life. On the positive side, long-term residence in a moderate to higher-income community with low residential mobility might suggest less distress and greater social connections (Ross, Reynolds, and Geis, 2000).
Appropriate time period
In addition to considering how various social contexts of families might lend themselves to longitudinal measures, researchers must also consider the appropriate time period over which these indicators should be assessed. In the absence of an empirical literature, such choices must be driven by theory and common sense. In general, the notion of a longitudinal indicator implies that it is measured over at least two or three years or longer, if possible. Many survey questions currently ask about activities in the past year, so this rather limited version of a longitudinal indicator is available now for some constructs. (For example, attendance at religious services in the National Survey of America's Families is assessed over the past year.) However, cumulating such a variable across two or three years or longer would be more informative. Long-term poverty and welfare receipt, for example, can be measured across four years in Survey of Income and Program Participation (SIPP) and even longer in the Panel Study of Income Dynamics (PSID) (U.S. Department of Health and Human Services, 2003). Lifetime measures represent a special case of longitudinal variables. One example would be the proportion of a child's life that his or her biological parents were married.
In general, it seems that measures should be assessed for at least two or three years and up to four or five years, or even longer when appropriate and feasible. Longer time frames are helpful for statuses that can be reported retrospectively, e.g., using an event history calendar. However, people's ability to recall distant events accurately argues against going back more than a decade, except for highly salient events, such as births and marriages. If data are taken from multiple waves of a longitudinal survey, the value of a longer time frame has to be balanced against the cost of collecting data for many years and the attrition that is likely to occur over a period of years and the extent to which the sample would be biased by that attrition. Fortunately, a number of longitudinal surveys (such as the Panel Study of Income Dynamics) have collected annual data on work, income, family structure, and other topics for many years and could readily support construction of longitudinal indicators.
Types of measures that could be assessed longitudinally
Our discussion of potential longitudinal indicators for each of the social contexts of families that we have considered for this project indicates that many cross-sectional indicators have a corresponding longitudinal indicator that provides a useful perspective on the stability or duration of that indicator across families. Nevertheless, not every construct lends itself to longitudinal measurement or is of sufficient importance to warrant an ongoing investment in data collection. Families can experience discrete events occurring at one point in time that change the course of their lives. Such discrete experiences are not appropriately assessed using longitudinal indicators, and can instead be assessed in a cross-sectional survey by asking families if they have ever experienced a particular event. In contrast, the strength of longitudinal indicators lies in their ability to assess the consistency of family characteristics, interactions, attitudes, values, or behaviors over time.
To decide which constructs warrant development of longitudinal indicators, we suggest identifying several potential measures in each of the major ecological domains and focusing on those constructs found to be related to family, adult and child well-being, and self-sufficiency. Examples of such constructs are provided in the next section.