Measures
Self-description: The Self-Description Questionnaire-I (Marsh, 1990) is a self-report survey intended for use with children ages 8 through 12 years. The general self scale and physical appearance scale of the survey were used here. The general self scale relates to, "… perception as an affective, capable individual, proud of and satisfied with the way he is." (Marsh, 1990, p. 7). The physical appearance scale relates to, "… self-concept regarding his physical attractiveness as compared with others, and the perception of how others think he looks." (Marsh, 1990, p. 5). Responses range from 1 (False) to 5 (True). The factor structure was supported across 8 studies, and demonstrated independence between scales. Each scale has 8 items. Items for the general self scale include, "Overall I have a lot to be proud of," and "I can do things as well as most other people." Internal consistency was 0.81. Sample items for the physical appearance scale include, "I like the way I look," and "I have a good looking body." Internal consistency was 0.83. Although usual test-retest methods were considered inappropriate due to expected changes in self-perception over time, findings suggested systematic changes over 6 months that supported the scales' stability (Marsh, 1990).

Task self-efficacy: The physical self-concept scale of the Tennessee Self-Concept Scale: 2 Child Form (Fitts and Warren, 1996) is a self-report survey intended for use with children ages 7 through 14 years. The physical self-concept score relates to the construct (within self-efficacy theory) of task self-efficacy (Annesi, 2006), or "… the individual's view of his body, state of health, physical appearance, skills, and sexuality" (Fitts and Warren, 1996, p. 23). A single score is recorded from responses ranging from 1 (Always False) to 5 (Always True) on 12 items. Item clusters, however, include identity (eg, "My body is healthy"), satisfaction (eg, "I don't feel as well as I should"), and behavior (eg, "I'm not good at sports and games"). Factor analysis supported the physical self-concept scale items relative to the other 5 scales of the Tennessee Self-Concept Scale: 2. Internal consistency for the 9- to 12-year-old age group averaged 0.70, and test-retest reliability over 1 week was 0.71 (Fitts and Warren, 1996).

Self-regulatory efficacy: The Exercise Barriers Self-Efficacy Scale for Children (Annesi et al., 2005) is a self-report survey that assesses the construct (within self-efficacy theory) of self-regulatory efficacy, or the degree one believes he or she possesses the ability to overcome social, personal, and environmental barriers to participating in physical activity (Annesi, 2006). Construction of the 10-item survey, each item beginning with the stem, "I am sure I can exercise three or more days per week even if…", was based on previous research (Marcus et al., 1992), and adapted for the ages of 8 through 12 years. Responses range from 1 (Not at all confident) to 5 (Definitely confident). Items include, "I was nervous being around other people" (social barrier), "My body felt uncomfortable while exercising" (personal barrier), and "The weather was bad (very hot, rainy, very cold)" (environmental barrier). Internal consistency averaged 0.79, and test-retest reliability over 1 week was 0.77 (Annesi et al., 2005).

Mood: The tension and vigor scales of the Profile of Mood States - Short Form (McNair et al., 1992) are self-report surveys of five items each. Responses range from 0 (Not at all) to 4 (Extremely). Items for the tension scale include "nervous," "anxious," and "tense;" and for the vigor scale, "energetic," "active," and "lively." Internal consistency (adult samples) averaged 0.91 and 0.88, respectively. Test-retest reliability over 3 weeks was 0.70 and 0.65, respectively (McNair et al., 1992). Profile of Mood States scales were used with children starting at 9 years of age (Annesi, 2005; Berger et al., 1997). Internal consistency for the present sample was 0.74 (tension) and 0.79 (vigor).

Voluntary physical activity: Voluntary physical activity is defined as physical activity completed of one's own volition, outside of when it is mandated based on requirements and/or expectations from structured settings. For example, when a student completes a bout of physical activity within a PE class, it would not be considered as voluntary physical activity as it would if he chooses to ride his bicycle after school. Consistent with recent research (Annesi, 2006; 2007; Berger et al., 1997), a single-item scale was used to assess the number of days a participant voluntarily completed a moderate-to-vigorous ("made you breathe harder than usual") bout of physical activity or exercise over the previous week, excluding such physical activities completed during school (eg, during PE class) or after-school programs (eg, during a Youth Fit For Life session). The item was based on review of the extant physical activity recall research (Piera et al., 1997), and adapted from recent research with 12-year-olds from Canada (Tremblay et al., 2000). Test-retest reliability over 1 week was .79. Significant correlations between reported days of voluntary moderate-to-vigorous physical activity, and time to complete a 1-mile (1.61 km) run/walk (r = -0.39, p < 0.01) and distance covered in a 6-minute run/walk (r = 0.33, p < 0.01), supported the scale's validity. A significant correlation of 0.70, with no significant difference in means, was found when the present scale was administered during the summer, then winter (and counterbalanced), suggesting that responses were not biased by season. Possible responses ranged from 0 to 7 days. Although a single item scale is sometimes discouraged, appropriate reliability, validity (especially, for purposes of this research, predictive validity), and applicability was clearly demonstrated.

Changes on each measure (ie, difference scores) were derived by subtracting scores at baseline from scores at week 12.

Procedure
Participants already registered for a 12-week segment of YMCA after-school care volunteered to enroll in the investigation. Based on location of enrollment either the Youth Fit For Life protocol (treatment group) or an equal amount of time for unstructured physical activity (control group) was incorporated. All other aspects of after-school care were the same (eg, consumption of a snack, completion of homework). Parents and legal guardians were provided information packets, and were required to sign an informed consent form indicating a desire for their child to participate, and sufficient health to avoid undue risks through participation. Information on how collected data were to be handled, including how all identifiers were to be destroyed, was also provided. The contact information of the principal investigator was given so than all concerns and questions could be promptly addressed.

Youth Fit For Life consisted of 3, 45-minute sessions per week, and was delivered in elementary school multi-purpose rooms by after-school counselors whose training in PE methods was generally limited to a 6-hour block of instruction provided just prior to starting with the treatment. Using after-school counselors to administer the protocol directly facilitated testing in a manner that is consistent with the most efficient dissemination of the intervention. Participant-to-counselor ratios averaged approximately 12:1, and were limited to 15:1.

The Youth Fit For Life curriculum was supported by a manual, videotape, and quality control processes from wellness staff members of sponsoring YMCAs (National Cancer Institute, 2008). Its 4 components were cardiovascular exercise (noncompetitive games and tasks designed to maximize moderate-to-vigorous physical activity time for 20 minutes, 3 days per week), resistance exercise (age-appropriate use of resistance bands for 20 minutes, 2 days per week), nutrition/health information (1 new theme each week; 5 to 7 minutes, 3 days per week), and behavioral skills training intended to increase the quality of structured physical activity and increase amounts of voluntary, moderate-to-vigorous physical activity completed outside of structured programs incorporating exercise. Behavioral skills training was adapted from an exercise behavior change protocol for adults, and was administered 1 day per week for 20 minutes. It consisted of an assortment of brief, interactive group lessons. Self-management and self-regulatory areas covered were (1) methods for goal setting (short- and long-term), (2) establishing a system for progress monitoring which related to goals set, (3) the use of facilitative self-talk (eg, thought-stopping, cognitive restructuring), (4) identifying appealing physical activity types, and (5) recruiting social supports (eg, family and peer support; participation in a team or group activities). An interactive workbook supported the behavioral skills training component, and was referred to frequently. In addition, in groups led by the after-school counselors, participants discussed their efforts to utilize the behavioral skills.

All surveys were administered to both treatment and control group participants in a private area at baseline and week 12. Ethics required that participants were not provided opportunities to compare and contrast their personal data, and that all identifiers were promptly removed and destroyed. Without affecting validity, children were given an explanation of how their participation may serve to help develop programs to improve the health of children. In research such as this, it is important that the interests of the child participants and their parents be balanced against the advancement of health promotion research.

Data analysis
Statistical significance was set α = 0.05, 2-tailed, throughout. As suggested by Jaccard, 1998, a modified Bonferroni correction procedure was incorporated to adjust α-values for multiple tests, where appropriate. Assessment of sex differences were initially made on all variables. Dependent t-tests were next calculated to assess within-group changes in the 4 self-appraisal and 2 mood variables, and days per week of voluntary physical activity, over 12 weeks. Contrasts of changes between groups were then made using independent t-tests.

Regression analyses were next conducted for the treatment group only, to assess relations between changes in the self-appraisal and mood variables, and voluntary physical activity changes. Linear bivariate correlations were calculated between changes in the self-appraisal and mood variables, both controlling for baseline scores and not controlled, and voluntary physical activity changes. Thus, both bivariate and partial correlation coefficients are reported. Based on relationships indirectly suggested within social cognitive and self-efficacy theory, a series of linear multiple regression equations, with simultaneous entry of changes in the measures of self-appraisal, mood, and both self-appraisal and mood, were then conducted to assess their explained variance on voluntary physical activity changes. Although use of change scores has sometimes been questioned, it has been considered advantageous when, as here, it is driven by a priory theory (Fitzmaurice, 2001). In addition, tests for skewness and kurtosis suggested that the score changes over 12 weeks were appropriate for use in parametric tests for these data. Baranowski and colleagues indicated the importance of accounting for the dynamic nature of changes in health behaviors (Baranowski et al., 1998). Thus, consistent with previous research (Annesi, 2006), actual changes in scores were incorporated into the multiple regression analyses rather than controlling for baseline values. As previously suggested (Fitzmaurice, 2001; Williams and Zimmerman, 1996), this enabled analyses of changes while retaining the naturally occurring array of actual (rather than statistically adjusted) baseline scores, and their changes. The sample size had sufficient experimental power to detect a medium effect size at a .90 level (Cohen, 1988).

Preliminary analyses: For the treatment group, days of voluntary physical activity per week at baseline were significantly higher for boys (M = 2.68, SD = 2.21) than for girls (M = 1.95, SD = 1.93), t₁₄₄ = 2.07, p = 0.04. For the control group, voluntary physical activity at baseline was also significantly higher for boys (M = 2.77, SD = 2.19) than for girls (M = 1.97, SD = 2.03) at baseline, t121 = 2.07, p = 0.04. Boys also had significantly higher scores at baseline on physical appearance for both the treatment, t144 = 2.07, p = 0.04, and control, t121 = 1.98, p = 0.05, groups. There were, however, no significant differences by sex in score changes from baseline to week 12, on any variable under study, for either the treatment or control groups (all p-values >0.10). Floor and ceiling effects were not a problem. Therefore, consistent with previous research (Annesi, 2004b; 2005; 2006), data were aggregated by sex for further analyses.

Changes in psychological and physical activity factors: For the treatment group, dependent t-tests indicated significant within-group score changes over 12 weeks on the self-appraisal factors of general self, physical appearance, physical self-concept, and exercise barriers self-efficacy (all d-values = 0.14 to 0.23); and the mood factors of tension and vigor (d-values = 0.42 and 0.26, respectively) (see Table 1). Frequency of voluntary moderate-to-vigorous physical activity significantly increased by a mean of 1.05 days per week (SD =1.63) over 12 weeks, with a moderate effect size (d = 0.58). No significant within-group score changes were found for the control group on any variable (see Table 1). Contrasts of score changes indicated that significantly greater improvements were made on each of the self-appraisal, mood, and physical activity measures in the treatment group (see Table 2).

Relation of changes in psychological factors and physical activity: Significant linear bivariate correlations (all dfs =144) were found between score changes over 12 weeks in the self-appraisal factors of general self, physical appearance, physical self-concept, and exercise barriers self-efficacy, and changes in days per week of voluntary physical activity (r-values = 0.29, 0.37, 0.38, and 0.34, respectively; all p-values <0.001). Significant bivariate correlations were also found between score changes in the mood factors of tension (r = -0.20, p = 0.02) and vigor (r = 0.31, p < 0.001), and changes in days per week of voluntary physical activity. When baseline scores on each self-appraisal and mood factor were controlled, similar strength partial correlation coefficients also suggested significant relationships between score changes in general self (r_12.3 = 0.28, p = 0.001), physical appearance (r_12.3 = 0.34, p < 0.001), physical self-concept (r_12.3 = 0.36, p < 0.001), exercise barriers self-efficacy (r_12.3 = 0.34, p < 0.001), tension (r_12.3 = -0.19, p = 0.02), and vigor (r_12.3 = 0.22, p = 0.01), and changes in days of voluntary physical activity per week.

When the four measures of changes in self-appraisal were simultaneously entered into a multiple regression equation, a significant portion of the variance in changes in days of voluntary, moderate-to-vigorous physical activity per week was accounted for (see Table 3, Model 1). When the 2 measures of changes in mood were entered into a separate regression equation, a significant portion of the variance in changes in days of voluntary physical activity per week was explained (see Table 3, Model 2). When changes on all self-appraisal and mood measures were simultaneously entered into a third regression equation, a greater portion of the adjusted variance in changes in days of voluntary, moderate-to-vigorous physical activity per week was accounted for, than for either of the two previous models (see Table 3, Model 3).

Findings from this preliminary investigation suggested that the Youth Fit For Life protocol was associated with a significant, moderate increase in days of voluntary, moderate-to-vigorous physical activity completed per week, and significant, small improvements in measures of self-appraisal and mood, in the present sample of African American preadolescents. There were no significant changes associated with the control condition, which reserved time for physical activity but did not administer a structured physical activity curriculum. This suggested that, as presented in hypothesis 1 and hypothesis 2, the Youth Fit For Life protocol induced significantly improved feelings of self-efficacy, self-image, and mood; and increases in freely chosen physical activity in the present sample of African American preadolescents - a group that may be especially at risk for overweight.

A primary focus of the investigation was the relationship of changes in theoretically proposed psychosocial correlates of physical activity, and changes in weekly moderate-to-vigorous physical activity sessions completed outside of programmed settings. Only the treatment group was included in these analyses because of the restricted ranges in change scores in the control group (that failed to demonstrate a significant improvement on any variable assessed). The significant linear bivariate correlations of changes on the self-appraisal factors and changes in physical activity suggest that increases in both general and physical self-description, and increases in both task and self-regulatory self-efficacy, were associated with increased voluntary physical activity. This confirmed hypothesis 3, principles of social cognitive and self-efficacy theory, and some, but not all, research on correlates of physical activity in youth (Sallis et al., 2000). It is likely that, as intended, the Youth Fit For Life curriculum and administration method served to improve both feelings of one's physical self and competence with physical tasks; and those improvements facilitated more freely chosen physical activity.

The significant linear bivariate correlations of changes in tension and vigor scores, and changes in physical activity sessions completed, suggested relationships similar to those found with adults (Annesi, 2004a; Institute of Medicine, 2007). It is likely that, as in adults, physical activity participation is associated with improved mood over time for preadolescents, and improved mood serves to reinforce physical activity behaviors (Annesi, 2004a). Regression analyses suggested that simultaneously accounting for changes in each of the self-appraisal and mood factors increased the accuracy of prediction of changes in days of voluntary physical activity compared to either of the 2 types of predictor variables alone. Simultaneously accounting for changes in each of the self-appraisal and mood factors accounted for a substantial 36% of the variance in changes in number of days of moderate-to-vigorous physical activity per week, over 12 weeks. This confirmed hypothesis 4. It is likely that the improved feelings of the self, and improvements in mood, associated with the Youth Fit For Life protocol, served an additive function in regard to inducing freely chosen physical activity in the participants. Inspection of the independent, significant strength of 5 of the 6 factors in multiple regression model 3 ( -values in Table 3) supports this. This will be important knowledge for future developers of treatments that seek to induce increased out-of-treatment physical activity.

Replication is required before generalization of findings is possible, and should include a longer timeframe, account for physiological factors at baseline (eg, BMI), evaluate effects of sex more thoroughly, and directly assess intervention components for their specific effect. For example, a pilot version of Youth Fit For Life was associated with significant improvements in exercise barriers self-efficacy for girls but not for boys, which precipitated modifications in supporting materials for the behavioral skills component. Adjustments based on proposed sex differences in learning styles (Pomerantz et al., 2002) were incorporated into the treatment that were subsequently associated with equalized effects. Effects specific to the African American sample presently tested should also be analyzed. As the present preliminary research is extended into intervention design, curriculum modifications directed at changes on relevant psychological factors should also be accounted for to maximize impact. Effect decomposition, as was completed within this investigation, allows research to be used to purposefully direct curriculum development.

Although experimental confounds consistent with field research were certainly possible here (eg, expectation effects, social support effects), and should be considered limitations, this study served to simultaneously account for extant theoretical, intervention, and application research on psychosocial correlates of physical activity in a group with demonstrated need. Factors consistent with increased activity in African American youth were suggested, and implications for intervention were extended. Future related field research should be careful, however, to maximize integrity of experimental controls, while attempting to maintain high external validity.

• Social cognitive theory offers a framework for understanding correlates of physical activity in youth.
• This study suggests that it is possible for a convenient physical activity intervention, led by after-school care counselors with minimal training, to improve participants' self-perceptions, mood, and voluntary physical activity.
• Improvements in self-perceptions and mood appeared to be significantly associated with increased free-time physical activity in African American 8- to 12-year-olds.
• This study's findings may lead to a better understanding of physical activity promotion in youth, and foster improvements in physical activity curricula.