This is a Mentored Quantitative Research Career Development Award Application to gain comprehensive understanding of clinical and biological correlates of Alzheimer's disease (AD) and to assess the statistical applications and the quality of longitudinal statistical methodologies in AD publications over time. The training goals, which will be met through an active mentored research project as well as formal course work, seminars, workshops, and participation in ADRC clinical/cognitive assessments, are threefold: (1) to enhance my knowledge in the area of AD research in general, (2) to receive training in the collection, management, analysis of AD research data, and (3) to understand and assess which, and how quantitative methodologies have been used in AD studies.
The specific aims of the research project are: (1) to characterize the trend over time on the use of statistical designs and methodologies in AD research, the way statistical analyses were reported, and the types of statistical analyses were reported, and the types of statistical packages used in AD publications, (2) to develop an instrument to assess the quality of statistical methodologies in published longitudinal AD studies, (3) to identify less-than-optimal uses of statistics in published longitudinal AD studies, (4) to assess the role of biostatisticians/epidemiologists in published longitudinal AD Studies, and (5) to assess whether attenuation or absence of learning effects over time from psychometric tests differentiates individuals with preclinical AD from nondemented individuals without AD neuropathology, and to compare longitudinal statistical methodologies and demonstrate the consequences of less-than-optimal statistics in the analyses. Data from756 original research articles since 1984 will be reviewed to address Specific Aim 1 - 4, which will provide not only the first ever systematic and critical evaluation of statistics in AD publications, but also guidance on how the application of statistics can be improved in the future.
Specific Aim 5 will not only test an important hypothesis about the predictors of preclinical AD from WU ADRC database bust also provide a comparison of longitudinal statistical methods and demonstrate the inferential consequences when less-than-optimal statistics are used in the test of the hypothesis. These investigations will yield substantive contributions related to the statistical methodologies of AD research and help the 'translational' process disseminating advanced longitudinal statistical methodologies to research in AD Field.
