1. When Intention Precedes Action: Does Adolescent Self- Rated Risk Evaluation Predict Deleterious Decision Making? By: Lynn A. Agre, MPH Ph.D. Candidate, Rutgers University, School of Social Work American Public Health Association November, 2006

2. Introduction The premise for this query into adolescent self-rated risk as a cofactor in determining the likelihood to engage health risk behaviors, such as substance use and earlier sexual behavior, arises out of the self-rated health literature, a five-point Likert scale demonstrated to be a strong predictor for mortality among adults (Idler and Benjamini, 1997). The self-rated risk scale used in the Young Adult portion of the National Longitudinal Survey on Youth in the 1998, 2000 and 2002 waves evaluates how discerning adolescents are in their planfullness and proclivity toward sensation seeking.

3. Research Questions It is postulated that those adolescents who identify as more risk prone versus risk adverse are more likely to engage in alcohol and drug use, in addition to sexual behavior, particularly in early adolescence. Further, it is also hypothesized that youth who originate from households of mothers with higher educational attainment, as a proxy for social support, will be less likely to engage in health risk behaviors, such as substance use and early onset of sexual behavior (Rosenbaum and Kandel 1990).

4. Influence of Social Support on Prosocial Behavior Social support can be included under the general rubric of social environmental factors as an essential component in buffering certain patterns of behavior and how these affect adolescent health risk decision making. The process of social support is contingent upon the participation of another person in a reciprocal relationship where some benefit is exchanged between the person experiencing the illness episode or crisis and the other who is not. House (1981) defines four different types of social support: emotional, appraisal, informational and instrumental.

5. Influence of Maternal Education as Appraisal and Informational Support Emotional support refers to emotional concern, love and empathy received from those in the domain. Appraisal support entails deriving information relevant to self-evaluation. Informational support pertains to seeking knowledge about the situation. Instrumental support involves help with daily activities. Maternal support in this context is measured as social support, since maternal education has been demonstrated to delay the initiation of health risk behaviors in adolescence (Mensch and Kandel, 1992).

7. Theoretical Framework The framework selected for assessing the multiple interacting environments is the Bronfenbrenner ecological approach (1979), comprised of the individual, family, and extra-familial level (Small and Luster, 1994) contained in the ecosystem (Ginther, Haveman, and Wolfe, 2000). In this study, the effect on health risk behavior is based upon an adolescent’s self-rated perception of risk in conjunction with the influence of multiple environments. Bronfenbrenner’s ecological paradigm considers role expectations of the individual in different environments in contrast to the internal-external locus of control model simply viewing impulse control as total reliance on inhibition of self (Rotter, 1966).

9. Internal-External Locus of Control (Rotter, 1966) The internal-external locus of control model does not take into account how multiple environments and the influence of the behavioral exchange within those environments can temper the individual’s capacity to engage in behavior detrimental to physical and mental well-being. Therefore, the Bronfenbrenner model views the individual as both the decision maker and the operator, neither placing the blame on the self, nor viewing another as blameworthy. Rather, the change in behavior is dependent upon the fluid process of one environment influencing another. Thus, various forms of social support in these different contexts of the ecosystem, i.e.micro-, meso- and exo- systems—act as buffers when disruptions occur in each of these domains.

11. Model: Presaged Intention Predicts Outcome The predictor variables included in the analysis are maternal age, maternal education, adolescent age, gender and race. The psychosocial indexes encompass: (1) the seven-item short form of the CESD (Derogatis, 1977) on which respondents indicate on a four ‑ point scale how often they have experienced symptoms of depression during the past week; (2) the self-mastery scale consisting of select measures originally developed by Rosenberg and Pearlin (1978); (3) self-esteem scale as a composite of ten variables about perception of control over life problems and capacity to solve these problems; and (4) parenting scale assessing the adolescent’s perception of how well parents agree on household rules. The individual measures for each potential scale have been summed to create one single variable.

13. Scale Construction The higher the score on each of the indexes the better the mastery, the self-esteem and the quality of parenting as perceived by the adolescent. In contrast, however, the higher the score on the short form of the CESD, the worse the depressive symptoms according to the adolescent. The neighborhood index evaluates the quality of the neighborhood environment from the respondent’s point of view, using the dichotomous yes/no format. Finally, the risk behavior scale contains six reverse- coded items where a higher score means greater willingness to engage in risk behavior i.e.: (i) often does things without thinking; (ii) planning takes the fun out of things; (iii) uses self ‑ control to keep out of trouble; (iv) enjoys taking risks; (v) enjoys new/exciting experiences; and (vi) feels life without danger is dull.

14. Bivariate Methods Methods for testing group distinction between high and low risk takers, include t-tests to examine mean differences between the sociodemographic control measures and the psychosocial scales, all of which are significant. Mantel-Haenszel Chi-Square is also employed to test for independence between psychosocial well being states and high and low risk propensity, and a variety of health behaviors including age at first alcohol, tobacco and marijuana use and age at initiation of sexual behavior. This analysis of association between two binary variables is applied to ascertain if proportion of one group is different from another. The low and high risk groups are statistically distinct from each other at the.05 significance level on all the psychosocial scales and detrimental health behavior variables.

15. Multivariate Results In the multivariate model, higher adolescents scores on the self-rated risk index are significantly correlated with other psychosocial well-being measures such as greater depressive symptoms, lower self-esteem, but a higher sense of mastery. The regression analysis demonstrates that alcohol and drug use affects adolescents likelihood to rate herself/himself as risk prone and increase the likelihood of initiation of sexual intercourse at a younger age. Male adolescents are more likely to engage in risk behavior at an earlier age than females, particularly African American males. Neighborhood quality does not increase likelihood of an adolescent to perceive herself/himself as risk prone but does increase the initiation of sexual activity at a younger age.

16. Discussion Adolescents who report more depressive symptoms also perceives themselves as higher risk takers. Risk is studied as a matrix of concomitant exposure to adverse social conditions, the youth’s judgment of that milieu and her/his internalization of his response and sensitivities to those environs.(Link and Phelan, 1997). Self-assessment of risk during adolescence could emerge as a potential predictor in determining later-life health trajectories.

17. Regression Analysis - Dependent Variable: Age when first had sex 1998

18. Regression Analysis - Dependent Variable: Age when first had sex 1998 (Continued)

19. Regression Analysis - Dependent Variable: Risk Scale 1998

20. Regression Analysis - Dependent Variable: RISK2_1998 (Continued)

21. Regression Analysis - Dependent Variable: # of people had sex with in last 12 months 1998 (Continued)

22. - Regression Analysis - Dependent Variable: # of people had sex with in last 12 months 1998 (Continued)

23. - Regression Analysis - Dependent Variable: On Average how often R drank in the past 12 months.

24. - Regression Analysis - Dependent Variable: On Average how often R drank in the past 12 months (Continued)

25. - Regression Analysis - Dependent Variable: Age when first began to drink alcohol once a month or more 1998

26. - Regression Analysis - Dependent Variable: Age when first began to drink alcohol once a month or more 1998 (Continued)

27. - Regression Analysis - Dependent Variable: How often in past 30 days smoked cigarettes? 1998

28. - Regression Analysis - Dependent Variable: How often in past 30 days smoked cigarettes? (Continued)

29. - Regression Analysis - Dependent Variable: Age of respondent when first used marijuana? 1998

30. - Regression Analysis - Dependent Variable: Age of respondent when first used marijuana? (Cont.)

31. Behavior & Health Risk Decision Making