rive at overall conclusions about youth development programs. Before identifying “best practice,” we need to know that there is “good practice,” otherwise such selectivity runs the risk of capitalizing on chance findings or inadvertently highlighting unusual programs.

Of the many thousands of youth development programs in the United States, a reviewer seems able to find at most several hundred evaluation efforts, of which a few dozen typically satisfy the methodological criteria set forth. Often the reviews conclude by identifying a small number of model programs (as judged by rigorous evaluations) that are then described in detail. Table 26.1 lists some of the frequently cited model programs and their design features. (More detailed descriptions of most of these programs as well as many others are available at http://www.casel.org/about sel/SELintro.php by following the link to programs/curricula). These programs are not the only ones that work, but the evidence for their effectivess is especially solid because it usually involved evaluation with random assignment, multiple outcome measures, and long-term follow-up.

Table 26.2 summarizes some of the more recent reviews of empirical studies of the effectiveness of youth development programs in reducing problems and/or promoting well-being. As can be seen, the outcome measures ranged from the positive to the negative, although individual reviews tended to focus on only a few classes of outcomes. Every review offered its own conclusions about what works, but we rely here on syntheses provided by Eccles and Gootman (2002), Nation et al. (2003), and Park and Peterson (2004). The bottom line is that youth development programs can promote the positive and reduce the negative. Each of the reviews was able to point to empirical evidence that at least some programs achieved one or more of their stated goals, as shown by demonstrable effects on the outcomes of interest. However, caution is introduced by the following disclaimers: (1) lack of random assignment in many program evaluations; (2) inconsistent measures across studies, especially of positive outcomes; and (3) in most cases, lack of long-term follow-up data (i.e., years after the program is done).

A common thread of programs that work is supportive relationships (between youth participants and group leaders, teachers, parents, and so on) and, not surprisingly, fidelity of implementation. One-shot programs do not work well, although it is unclear just how long a program needs to be before it begins to work. Structured programs are more effective, a feature that can be achieved by manualizing the program, or spelling out in detail just what one does to make the program a reality. Everyone concludes, although these may be articles of faith more than evidence-based facts, that programs need to take account of the multiple links among socializing agents and that community norms supporting the goals of a program must be in place (cf. Schinke & Matthieu, 2003).

The reader will note that most of the reviews have been of programs and not of more general institutions that might promote well-being among youth. The Johnson et al. (2003) review is an exception—the socializing system of religion has been consistently associated with desired outcomes, although this is a purely correlational conclusion and leaves unanswered the hypothetical question of what coerced participation in religious activities would produce. (History tells us that this would be a disaster if done on a large scale, but we also wonder about the effects of parents forcing unwilling offspring to attend church.) The Child Trends (2003) research briefs are another exception, because they survey the effects of not only specific programs but also more general institutions and socializing systems.

And what about communities? Epidemiological research tells us that problems are more likely to occur in some communities than others, but the studies are not fine-grained and in any event we know that problems co-occur. Not enough is known about the community set-tings that help youth thrive in all the ways that we have described, although extensive re-search with subjective well-being shows that demographic variables (a proxy for certain community-level variables given ethnic, socioeconomic, and educational stratification in the United States) are but weakly associated with life satisfaction. Recent studies of character strengths in adults similarly show few consistent demographic correlates except gender and,

among African-Americans, religiosity, a finding reported by many previous researchers (Peterson & Seligman, 2004).

There is agreement at least within the contemporary United States about the positive characteristics of youth. Labels will vary, but the features proposed by different groups overlap substantially. These are best regarded as a family of characteristics, each of which exists in degrees. Children and adolescents are not simply doing well or doing poorly, and accordingly, we need to take a broad and nuanced view of the goals of positive youth development.

There is agreement that indicators and indices of positive youth development must do more than ascertain the absence of disorder and distress. Much further work needs to be done to craft generally useful measures of positive constructs and to see that these are routinely used in evaluations of youth programs (Moore et al., 2004).

There is, of course, agreement that positive characteristics are valuable in their own right but, more importantly for the purpose of this volume, that positive characteristics can buffer against the development of the most common psychological disorders among youth (Pollard et al., 1999). Furthermore, we can encourage optimal development through youth programs, either those that already exist (e.g., Big Brothers and Big Sisters) or those explicitly designed by psychologists, prevention scientists, and youth development practitioners for this purpose (e.g., the Penn Resiliency Program). There is also agreement that the personal characteristics of group leaders are critical for the success of their programs, as is parental support.

From the existing program evaluation research, investigators have agreed that programs are apt to be most successful—increasing positive outcomes and reducing negative outcomes—if they have the following features: