The benefits to come from producing the wealth of multi-level behavioral, genotypic, and imaging data in this PPG depend on the quality of the analyses of these data. The Statistical Analysis (SA) Core will form a center of expertise and support in Statistics and data analysis for the PPG, with a mission of enhancing scientific progress by ensuring optimal use of data. As optimal results in data analysis are achieved through a combination of expertise in data analysis with expertise in relevant subject matter, the SA Core has been designed to bring statistical experts from the faculty of the Yale Statistics Department together with select data- oriented personnel from Projects 1-5 who are intimately familiar with the science. This collaboration will develop and apply the most powerful and revealing analytic methods, maintaining rigorous statistical validity without artificially limiting the sophistication of the methodology. The individual Projects will perform analyses to achieve their specific aims and the SA Core will assist as needed. The SA Core will also be responsible for performing analyses that use data from more than one project to address integrative aims that span across cohorts, ages, and data types. Several integrative aims are detailed in the SA Core description. For example, we will use longitudinal fMRI connectivity measurements made in the prenatal and neonatal periods from Project 1 and Project 4 to predict autism severity and also to predict response to a novel social value training protocol proposed in Project 5. We will adapt brain connectivity networks from Project 2 derived as maximally associated with sustained attention in school-age children to study neonates in project 1 who are too young for selective attention to be measured effectively. And we will relate neurobiological characteristics of induced pluripotent stem cells from Project 3 to fMRI network connectivity and measures of attention and autism in school age from Project 2. After establishing baseline analyses through careful applications of classical methods such as ANOVA, regression, and mixed-effects models, the SA Core will seek to apply more modern, sophisticated developments in statistical theory and methodology to develop improved analyses. As detailed in the SA Core description, members were chosen to have distinct and complementary expertise in statistical areas expected to be of importance to the projects and integrative aims, such as high dimensional statistics (which applies whenever many variables are measured for each participant as in this PPG), dimension reduction, statistical computing, longitudinal analysis, and network models (which will apply to both brain connectivity networks and genetic networks in this PPG). A collateral benefit of the SA Core is the recruitment of Dr. Zhou and Dr. Negahban, scholars working at the highest levels of statistical research, to the field of autism research. We also anticipate increased participation by graduate students from the Yale Statistics Department, which is currently entering a vibrant period of expansion, in the autism research of the Yale Child Study Center. 1
