CORE--BIOMETRICS AND STATISTICAL DEVELOPMENT

University Of Colorado Denver, Aurora CO

Investigators

Linked publications & trials

Abstract

Description: This Core has two Specific Aims: 1) To provide routine statistical support for the Projects, which encompasses assistance with the design of studies, advice on database design and content, summarizing, plotting, and analyzing data, interpretation of results, and collaboration in the writing of manuscripts for publication and/or presentation; and 2) The development of advanced parametric and nonparametric analysis of variance (ANOVA) techniques in areas of benefit to Project objectives. Data analysis for the Projects in this grant often involves estimating and comparing linear and nonlinear models fit to data collected repeatedly over time or dose on the same animals. For situations where the usual normal theory is appropriate, the Core investigators and their students have been involved in the development of mixed model methodology in which BLUP (Best Linear Unbiased Prediction) is available as an intermediate computational step preceding their usage in estimating and comparing group mean response curves, or constructing confidence bands for these mean response curves. The investigators plan to explore the usefulness of these BLUPs, not only for predicting the responses of individual animals, but also for correlating variables of responsiveness. For example, the performance of a rat, as assessed by its average distance from the platform in a water maze probe trial, may be associated with its CREB level or CREB phosphorylation slate, or perhaps the total distance swam at day 8 in a water maze training trial. The results of such analyses have the promise of providing insight to the physiological mechanisms involved in learning and memory and their changes with age. The ability to correlate physiological variables with behavioral performance serves to tie together the efforts of the Projects, making the most of this multi-faceted investigation of a common question, and optimizing our ability to draw conclusions about the role of physiological factors in motor learning and aging. The Animal Core's centralized database will be vital towards this aim. For situations where the usual normal theory assumptions are in doubt, the Core investigators and their students have already been involved in the development of nonparametric analogues to linear mixed model methodology appropriate for the analysis of longitudinal data. The investigators now plan to extend these to the nonlinear case, with a new emphasis on BLUP estimation.