Ninth Meeting of New Researchers in Statistics and Probability - PDF document

Ninth Meeting of New Researchers in Statistics and Probability University of Washington Seattle, WA August 1 - August 5, 2006 1

Welcome Hello and welcome to the Ninth IMS Conference for New Researchers. The purpose of the conference is to provide a comfortable setting for new researchers to share their research and make connections with their peers in an informal setting. The conference is kept relatively small so we hope you will get to meet most, if not all, of your fellow participants. We are very excited about our technical and social programs, and we hope you enjoy your time at the University of Washington. A Brief History of the NRC The first NRC was held in Berkeley, CA in 1993. Many of the characteristics of the conference were initialized at the first meeting – it was held prior to a much larger meeting (JSM), it had a smaller number of participants (49), and it had esteemed invited speakers (Iain Johnstone and Terry Speed). This first meeting was such a success that another one was planned for 1995, in Kingston, Ontario. This every-two-year pattern continued for the next 8 years, with NRCs being held in Laramie, Wyoming (1997), Baltimore (1999), Atlanta (2001), and Davis, California (2003). One important element of the NRC is the requirement that each participant present their work, in either a short presentation or a poster. Presentations, combined with the small conference size, allows there to be maximal intellectual and social interactions. However, by 2003, the conference had grown so much in popularity that it was impossible to accommodate all applicants, and still hold to the criterion that everyone presents their work; the decision was made to hold the conference every year (Toronto (2004)). Last year’s meeting was at Minnesota, and next year will be in Utah. The current plan is to hold NRC every year prior to JSM, assuming the demand for the conference continues. 2

Highlights of 2006 NRC The structure of the conference is built around short presentations by the participants. There will be 3-4 sessions per day of six 12 minute talks, with ample break and lunch times factored in. On Thursday evening, just before the conference dinner, we will have a posters session. Please interact with the poster presenters. We are pleased to have talks by our invited speakers (Roger Berger, Dalene Stangl, and Bruce Weir), and we present two panel sessions. The first panel session will contain members of funding agencies who will give advice on how to get funding for your research. The second panel will contain editors of relevant journals and will provide some tips and tricks to maximize the probability of getting papers accepted. Finally we are pleased to have Savas Dayanik give the “Tweedie New Researcher Invited Lecture” at this conference. Richard Tweedie played a significant role throughout his professional career in mentoring young colleagues at work and through professional society activities. Funds donated by his friends and family cover travel for the award winner. We hope you enjoy the meeting! Peter Craigmile , Chair of the IMS New Researchers Committee pfc@stat.ohio-state.edu Peter Hoff , Chair of the Ninth Meeting of New Researchers Conference hoff@stat.washington.edu Acknowledgments We thank the following institutions for their generous funding and support. • Institute of Mathematical Statistics (IMS) • National Institutes of Health (NIH) / National Cancer Institute (NCI) • National Science Foundation (NSF) • Office of Naval Research (ONR) • National Security Agency (NSA) • University of Washington • The Ohio State University 3

Schedule Tuesday Wednesday Thursday Friday Saturday Aug 1 Aug 2 Aug 3 Aug 4 Aug 5 8:45 Intro Intro Intro Intro Remarks Remarks Remarks Remarks 9:00 Session 1 Session 5 Session 8 Journal (6 speakers) (6 speakers) (5 speakers) Panel 10:15 Break Break Break Break 10:30 Session 2 Session 6 Session 9 Session 11 (6 speakers) (6 speakers) (5 speakers) (5 speakers) (10:30-11:30) (10:30-11:30) 11:45 Lunch Lunch Luncheon Roger Berger Dalene Stangl (11:30-12:30) (11:45-2:00) 2:00 Session 3 Grants Session 10 (5 speakers) Panel (6 speakers) 3:15 Break Break Break 3:30 Session 4 Session 7 Tweedie Speaker (6 speakers) (6 speakers) Savas Dayanik 6:00 Dinner Poster Dinner in Dorms Mixer in Dorms Evening Opening Pub Conf. Dinner Mixer Event Bruce Weir (7:00-10:00) (8:00-) (7:30-10:30) Notes The opening mixer will be held from 7:00 - 10:00pm on Tuesday in the Pompeii room in McMahon Hall. For the pub event, we will meet at the front entrance to McMahon Halls at 7:30pm on the Wednesday and go from there. All participant talks are 12 minutes long . 4

Wednesday morning Session 1: Mixed and Random effects models, Robust methods for Repeated Mea- sures and Finite Populations 9:00-10:15am, Location: Miller 301. Chair: Peter Hoff. Impact of the random effect distribution on inference for mean in linear mixed models Joshua Rushton: Cornell University, USA Functional Mixed-Effects Models for Periodic Data Li Qin : Fred Hutchison Cancer Research Center, USA Mixed-Effects, Posterior Means and Penalized-Least Squares Yolanda Munoz Maldonado : University of Texas-Houston, USA A Random Effects Four-Part Model for Longitudinal Medical Costs Lei Liu : University of Virgina, USA Rank-Based Analyses of Repeated Measure Designs Under Exchangeable Errors John Kloke : Pomona College, USA Robust model-based predictor of the finite population total Yan Li : University of Maryland, USA Session 2: Statistical Genetics, Microarray and Proteomic Analysis 10:30-11:45am, Location: Miller 301. Chair : Shuangge Ma. Estimation of Gene by Exposure Interactions in Case-Parent Triad Studies Tracy Bergemann : University of Minnesota, USA Building gene trees from SNPs data: an ancestral mixture models approach Shu-Chuan Chen : Arizona State University, USA MCMC Linkage Analysis for Two Genes and a Polygenic Component on General Pedi- grees Yun Ju Sung : University of Washington, USA Uncertainty in clustering posterior distributions of gene expression levels using MCMC samples Tanzy Love : Carnegie Mellon University, USA Boosting nearest shrunken centroid classifier for microarray data Baolin Wu : University of Minnesota, USAg Feature identification quantitation and sample size: a comparative analysis of workflows for LC-MS based proteomics Olga Vitek : Institute for Systems Biology, USA 5

Wednesday afternoon Session 3: Dimension Reduction, Principal components, Central Subspace Regres- sion 2:00-3:15pm, Location: Miller 301. Chair: Jeongyoun Ahn. Penalized Likelihood Principal Component Rotation Trevor Park : University of Florida, USA Penalized Spline Models for Functional Principal Component Analysis Fang Yao : Colorado State University, USA Constrained Dimension Reduction Based on CANCOR Jianhui Zhou : University of Virgina, USA Optimal sufficient dimension reduction for the conditional mean in multiple-response regression Jae Keun Yoo : University of Louisville, USA Model Based Approaches for Simultaneous Dimension Reduction and Clustering Xiaodong Lin : University of Cincinnati, USA Using intra-slice information for improved estimation of the central subspace in regres- sion Liqiang Ni : University of Central Florida, USA Session 4: Higher order asymptotics, Kurtosis, Probability models on simplexes, Methods for Censored data 3:30-4:45pm, Location: Miller 301. Chair: Yanyuan Ma. Higher-Order Asymptotic Normality Of Approximations To The Modified Signed Like- lihood Ratio Statistic For Regular Models Heping He : The Australian National University, Australia Conditional Properties of a Parametric Bootstrap Russell Zaretzki : University of Tennessee, USA Kurtosis: New Theoretical Results and Inference Issues Anna Maria Fiori : Universita degli Studi di Milano - Bicocca, Italy A class of probability measures on the simplex, with emphasis on the Dirichlet distri- bution Zach Dietz : Tulane University, USA Estimation of truncated and censored regression models Maria Karlsson : Umeaa Universitet, Sweden Estimation of Wood Fibre Length Distributions from Censored Data through an EM Algorithm Ingrid Svensson : Umeaa Universitet, Sweden 6