In summary, a number of studies have found that long-term exposure to a positive or negative experience has important implications children and families, and we hypothesize that, in many cases, the effects of long-term experiences may be stronger than the effects of short-term experiences. Nevertheless, the potential of a longitudinal perspective for social indicators is not widely recognized, and little research has systematically compared point-in-time measures with longitudinal indicators. Moreover, although we have been able to identify a broad and interesting set of possible new longitudinal indicators, the set of available measures is very small. Other longitudinal indicators that could theoretically be of substantial interest cannot be tabulated because the data needed to create a rich set of longitudinal indicators are not readily available. For example, most constructs would need to be measured identically over time, so that they could be cumulated. In some cases, age appropriate measures would need to be developed so that, even though identical questions were not asked in each survey, the same construct would be measured on a comparable metric, so that a cumulative measure could be created. This has been done, for example, with parent involvement questions in the NLSY97. Every two years, parents are asked four questions about their involvement with each of their children in age-appropriate ways that change as the child gets older. Where such data exist, exemplary longitudinal indicators could be developed.
Moreover, this review suggests that a number of longitudinal indicators can be created from data currently available on important domains of the social context of families. In fact, a number of measures can be created for families with young children as well as families with older children. Examples include parent involvement in children's schools, religious attendance, civic engagement, turbulence, father involvement in childrearing, and parent-child relationships.
For other constructs where data are not currently available appropriate items could be inserted into ongoing surveys. Examples include marital quality and participation in activities such as sports teams or service clubs. In addition, some measures of behavior (not attitudes, perceptions, or values, which could more readily be distorted over time) could be created with retrospective data, but the accuracy of such retrospective data needs to be examined empirically. Examples include residence with both biological parents since birth, stability in child-care arrangements, and residential mobility. The cost of adding items to existing surveys is modest, compared to developing and fielding new surveys. However, we acknowledge that virtually all existing surveys are cash-strapped and some are already quite lengthy. Each additional item in a survey increases the cost of its administration. And even when budgets allow for adding items to surveys, increasing the length further could threaten surveys' response rates, since respondents may drop out when surveys becomes too burdensome. Nevertheless, many surveys welcome additional financial supporters, and respondents generally enjoy talking about their families and children, so both sponsors and respondents may welcome some new questions.
Most important, several longitudinal measures could be created by analyzing existing data and thus could be included in indicator reports rather readily. For example, an indicator of a consistently positive parent-child relationship and an indicator of consistent attendance at religious services could be created from the NLSY97. Analyses are needed to confirm that these long-term variables are associated with positive outcomes for families and children; however, prior correlational research and studies using point-in-time measures provide a strong basis for hypothesizing such associations. Assuming that these expectations are confirmed in multivariate prospective analyses that use longitudinal variables to predict child and family outcomes over time, the indicator portfolio could be greatly enriched by the addition of longitudinal measures of the social context of families.