RACE/ETHNICITY, EDUCATION & LATE LIFE COGNITIVE FUNCTION

在有关痴呆症或认知障碍的研究中，高龄、教育程度和种族/民族始终与认知能力差相关。已发表文献中的重要证据质疑了跨种族/民族和教育程度组的认知状态评估的结构有效性。对于观察到的不同种族/民族和教育程度组的认知状态与其他健康结局之间的关系差异，有一个非常合理的解释：这些组中认知功能评估的系统测量误差（项目偏倚）。本研究的目的是探讨这一假设进行项目反应理论（IRT）为基础的项目偏见评估的认知评估非常老年人。数据将从非常老的美国大陆居民的概率样本中获得。具体而言，（1）来自最年长者资产和健康动态（AHEAD）数据集的认知评估数据将用于将IRT模型拟合到来自非常年长的美国居民样本的修改的认知状态电话访谈响应，以及（2）将进行项目偏倚分析，以测试零假设，同时保持基本认知能力的恒定组差异，这些项目是（a）不受种族/族裔的偏见，和（B）不受教育程度的偏见。此外，（3）将根据AHEAD数据集中存在的丰富的社会人口统计学和社会经济学指标，对由于种族/民族教育程度水平导致的总体认知能力和系统测量误差的组间差异证据进行评价。拟议的分析代表了一个显着的进步，通过探索如何教育程度，年龄，社会经济地位和种族/民族影响认知评估项目水平的晚年认知功能的理解。这项工作的结果将（l）为开发文化公平的认知评估工具的持续努力提供信息，（2）促进对认知评估中种族/民族和教育影响的理解，（3）将有助于解释认知与其他健康状况指标之间的关系中的种族/民族差异以及认知下降和痴呆的可疑风险因素。

Advanced age, level of educational attainment and race/ethnicity are consistently associated with poor cognition in studies concerning dementing disorders or cognitive impairments. Important evidence in the published literature questions the construct validity of cognitive status assessment across race/ethnicity and educational attainment groups. There is a very plausible explanation for observed differences in the relationship of cognitive status and other health outcomes across race/ethnicity and educational attainment groups: systematic measurement error (item bias) in the assessment of cognitive functioning across these groups. The aim of this proposal is to investigate this hypothesis by conducting an item response theory (IRT) based item bias evaluation of cognitive assessments of very old persons. Data will be obtained from a probability sample of very old continental U.S. residents. Specifically, (l) cognitive assessment data from the Asset and Health Dynamics of the Oldest Old (AHEAD) data set will be used to fit an IRT model to modified Telephone Interview for Cognitive Status responses from a sample of very old U.S. residents, and (2) item bias analyses will be conducted testing null hypotheses that, while holding constant group differences in underlying cognitive ability, the items are (a) free from bias according to race/ethnicity, and (b) free from bias according to level of educational attainment. Further (3) evidence of group differences in overall cognitive ability and systematic measurement error due to level of race/ethnicity educational attainment will be evaluated conditional on the rich array of sociodemographic and socioeconomic indicators present in the AHEAD data set. The proposed analysis represents a significant advance in the understanding of late life cognitive functioning by exploring how educational attainment, age, socioeconomic status and race/ethnicity affect cognitive assessment at the item level. Results from this work will (l) inform continuing efforts to develop culture fair cognitive assessment instruments, (2) promote understanding of race/ethnicity and education effects in cognitive assessment, and (3) will help explain race/ethnicity differences in the relation of cognition to other health status measures and suspected risk factors for cognitive decline and dementia.