Key points are not available for this paper at this time.
BACKGROUND: This study investigated associations among adolescents' self-esteem in 6 domains (peers, school, family, sports/athletics, body image and global self-worth) and risk behaviours related to substance use, bullying, suicidality and sexuality. METHOD: A multistage stratified sampling strategy was used to select a representative sample of 939 English-, Afrikaans- and Xhosa-speaking students in Grades 8 and 11 at public high schools in Cape Town, South Africa. Participants completed the multidimensional Self-Esteem Questionnaire (SEQ; DuBois, Felner, Brand, Phillips, & Lease, 1996) and a self-report questionnaire containing items about demographic characteristics and participation in a range of risk behaviours. It included questions about their use of tobacco, alcohol, cannabis, solvents and other substances, bullying, suicidal ideation and attempts, and risky sexual behaviour. Data was analysed using a series of logistic regression models, with the estimation of model parameters being done through generalised estimation equations. RESULTS: Scores on each self-esteem scale were significantly associated with at least one risk behaviour in male and female adolescents after controlling for the sampling strategy, grade and race. However, specific self-esteem domains were differentially related to particular risk behaviours. After taking the correlations between the self-esteem scales into account, low self-esteem in the family and school contexts and high self-esteem in the peer domain were significantly independently associated with multiple risk behaviours in adolescents of both sexes. Low body-image self-esteem and global self-worth were also uniquely associated with risk behaviours in girls, but not in boys. CONCLUSIONS: Overall, the findings suggest that interventions that aim to protect adolescents from engaging in risk behaviours by increasing their self-esteem are likely to be most effective and cost-efficient if they are aimed at the family and school domains.
Wild et al. (Fri,) studied this question.