The Statistics Core is responsible for the coordination of scientific activity among the various projects and cores, and for budgetary planning, data base management, and statistical services. In addition, will promote the development of new analytic procedures tailored to the special problems that danger of capitalizing on chance when many different measures are taken with relatively small populations. Our solutions to this problem include (a)constraints on the level of analysis to which a strong hypothesis applies, (b) methods for aggregating multiple variables into a single robust variable, (c) constraints on the order in which hypotheses are tested, and/or constraints on the way that individual variables are used within a single complex model, and (d) a clear and fruitful separation between hypothesis testing (confirmatory analysis) and hypothesis generation (exploratory analysis). We also outline plans for integrative analyses across the populations under study within the Multicenter, including the construction of models that can be used for systematic cross- population comparisons.
Showing the most recent 10 out of 166 publications
