Parra Gilbert R, DuBois David L, Sher Kenneth J
Department of Psychology, University of Memphis, 202 Psychology Building, 3693 Norriswood Avenue, Memphis, TN 38152-6400, USA.
J Clin Child Adolesc Psychol. 2006 Sep;35(3):386-402. doi: 10.1207/s15374424jccp3503_4.
Latent variable mixture modeling was used to identify subgroups of adolescents with distinct profiles of risk factors from individual, family, peer, and broader contextual domains. Data were drawn from the National Longitudinal Study of Adolescent Health. Four-class models provided the most theoretically meaningful solutions for both 7th (n = 907; 48% boys) and 11th (n = 1039; 51% boys) graders. The 4-class solution for 7th graders included low risk (LR; 66%), socioeconomic disadvantage (SD; 19%), peer high risk (PHR; 9%), and family high risk (FHR; 6%) groups. Similarly, the 4-class model for 11th graders included LR (32%), SD (43%), high risk (HR; 21%), and FHR (4%) groups. Subgroup membership predicted reported levels of depressive symptoms and conduct problems both concurrently and over time. Strengths and potential limitations of using latent variable mixture modeling to investigate risk profiles for adolescent psychopathology are discussed.
潜在变量混合模型被用于识别来自个体、家庭、同伴及更广泛背景领域的具有不同风险因素特征的青少年亚组。数据取自青少年健康全国纵向研究。对于七年级学生（n = 907；48%为男生）和十一年级学生（n = 1039；51%为男生），四类模型提供了最具理论意义的解决方案。七年级学生的四类解决方案包括低风险（LR；66%）、社会经济劣势（SD；19%）、同伴高风险（PHR；9%）和家庭高风险（FHR；6%）组。同样，十一年级学生的四类模型包括LR（32%）、SD（43%）、高风险（HR；21%）和FHR（4%）组。亚组成员身份同时预测了报告的抑郁症状水平和行为问题，且在一段时间内持续有效。本文讨论了使用潜在变量混合模型来研究青少年心理病理学风险特征的优势和潜在局限性。
