Schrager Sheree M, Goldbach Jeremy T, Mamey Mary Rose
Office of Research and Sponsored Programs, California State University, Northridge, Northridge, CA, United States.
Division of Hospital Medicine, Children's Hospital Los Angeles, Los Angeles, Los Angeles, CA, United States.
Front Psychol. 2018 Mar 15;9:319. doi: 10.3389/fpsyg.2018.00319. eCollection 2018.
Although construct measurement is critical to explanatory research and intervention efforts, rigorous measure development remains a notable challenge. For example, though the primary theoretical model for understanding health disparities among sexual minority (e.g., lesbian, gay, bisexual) adolescents is minority stress theory, nearly all published studies of this population rely on minority stress measures with poor psychometric properties and development procedures. In response, we developed the Sexual Minority Adolescent Stress Inventory (SMASI) with = 346 diverse adolescents ages 14-17, using a comprehensive approach to de novo measure development designed to produce a measure with desirable psychometric properties. After exploratory factor analysis on 102 candidate items informed by a modified Delphi process, we applied item response theory techniques to the remaining 72 items. Discrimination and difficulty parameters and item characteristic curves were estimated overall, within each of 12 initially derived factors, and across demographic subgroups. Two items were removed for excessive discrimination and three were removed following reliability analysis. The measure demonstrated configural and scalar invariance for gender and age; a three-item factor was excluded for demonstrating substantial differences by sexual identity and race/ethnicity. The final 64-item measure comprised 11 subscales and demonstrated excellent overall (α = 0.98), subscale (α range 0.75-0.96), and test-retest (scale > 0.99; subscale range 0.89-0.99) reliabilities. Subscales represented a mix of proximal and distal stressors, including domains of internalized homonegativity, identity management, intersectionality, and negative expectancies (proximal) and social marginalization, family rejection, homonegative climate, homonegative communication, negative disclosure experiences, religion, and work domains (distal). Thus, the SMASI development process illustrates a method to incorporate information from multiple sources, including item response theory models, to guide item selection in building a psychometrically sound measure. We posit that similar methods can be used to improve construct measurement across all areas of psychological research, particularly in areas where a strong theoretical framework exists but existing measures are limited.
尽管构念测量对于解释性研究和干预工作至关重要，但严格的测量工具开发仍然是一个显著的挑战。例如，虽然理解性少数群体（如女同性恋、男同性恋、双性恋）青少年健康差异的主要理论模型是少数群体压力理论，但几乎所有已发表的关于该人群的研究都依赖于心理测量特性和开发程序较差的少数群体压力测量工具。作为回应，我们对346名年龄在14至17岁的不同青少年开发了性少数青少年压力量表（SMASI），采用了一种全面的方法进行全新测量工具开发，旨在生成具有理想心理测量特性的测量工具。在对经过改进的德尔菲法提供信息的102个候选项目进行探索性因素分析后，我们将项目反应理论技术应用于其余的72个项目。对总体、最初得出的12个因素中的每一个以及不同人口亚组的歧视和难度参数以及项目特征曲线进行了估计。由于歧视性过高，两个项目被删除，经过可靠性分析后又删除了三个项目。该测量工具在性别和年龄方面表现出构型和标量不变性；一个三项因素因在性取向和种族/民族方面表现出显著差异而被排除。最终的64个项目的测量工具包括11个分量表，总体信度（α = 0.98）、分量表信度（α范围为从0.75至0.96）和重测信度（量表>0.99；分量表范围为从0.89至0.99）都非常出色。分量表代表了近端和远端压力源的混合，包括内化的同性恋消极态度、身份管理、交叉性和消极预期（近端）以及社会边缘化、家庭排斥、同性恋消极氛围、同性恋消极沟通、消极披露经历、宗教和工作领域（远端）。因此，SMASI的开发过程说明了一种整合来自多个来源信息的方法，包括项目反应理论模型，以指导构建心理测量健全的测量工具时的项目选择。我们认为，类似的方法可用于改进所有心理学研究领域的构念测量，特别是在存在强大理论框架但现有测量工具有限的领域。
