New Statistical Methodology to Establish Construct Validity for Childhook Risk

描述（由申请人提供）：该导师研究科学家发展奖将支持候选人建立独立的研究事业，作为一名方法学家，开发尖端的统计方法，直接应用于酗酒和药物滥用预防。职业发展培训领域包括：1)发展精神病理学；2)酗酒和药物滥用预防研究和方法；3)现代计算技术；4)利用非参数回归对纵向数据进行建模。提出的研究计划的具体目的是：1)建立一个新的贝叶斯潜变量模型，从多个维度评估儿童酒精中毒的动态风险；2)建立一种新的半参数回归模型来验证儿童风险与青少年物质使用发展之间的关系；3)应用具体目标1和2中开发的方法，根据正在进行的一项针对酗酒高风险家庭的21年长期前瞻性研究收集的数据，检验假设。本研究开发的方法将有助于建立儿童风险的常用措施的结构效度。将制定两个统计方案并分发给预防科学家。基于经验数据推导出的纵向模型，将有助于根据儿童时期的危险指标预测新儿童青少年物质使用的发展轨迹。因此，它有可能提供关于儿童个体需要早期干预的时间和剂量的关键信息。

DESCRIPTION (provided by applicant): This Mentored Research Scientist Development Award will support the candidate in establishing an independent research career as a methodologist who develops cutting-edge statistical methodology with direct applications in alcoholism and substance abuse prevention. Career development training areas include: 1) Development psychopathology; 2) Alcoholism and substance abuse prevention research and methodology; 3) Modern computational techniques; and 4) Modeling longitudinal data using nonparametric regression. The specific aims of the proposed research plan are: 1) To develop a new Bayesian latent variable model for assessing dynamic childhood risk for alcoholism in multiple dimensions; 2) To develop a new semi-parametric regression model for validating the association between childhood risk and adolescent substance use development; and 3) To apply the methodologies developed in Specific Aims 1 and 2 to test hypotheses based on the data collected from an ongoing 21-year long prospective study of families at high risk for alcoholism. The methodology developed in this study will be useful in establishing construct validity of commonly used measures of childhood risk. Two statistical programs will be developed and distributed to prevention scientists. The derived longitudinal model based on the empirical data should be useful in projecting a new child's developmental trajectory of adolescent substance use based on his/her childhood risk measures. Thus, it has the potential to provide crucial information about the timing and dose an individual child needs for early intervention. PUBLIC HEALTH RELEVANCE: Alcohol use, alcohol problems and alcohol symptomatology are all developmental phenomena. The applicant's work would develop new methodology which allows for more accurate description of these developmental relationships than is allowed by currently available methods and takes account of the special statistical properties of the measures most commonly used by the field.