We are inviting applications from graduate students for the poster session. The number is limited, so please apply by November 15th.  Successful applicants will be notified by December 1st.  Please send CV, one letter of recommendation, title, and abstract to Florentina Bunea at bunea@stat.fsu.edu. We can provide support in the form of two nights lodging for shared accommodation. We expect the home institution to pay for the travel expenses and the registration fee. Participants: Lan Wang  Department of Statistics  Penn State University  “Nonparametric Analysis of Covariance”  Abstract: We consider the fully nonparametric model for analysis of covariance in Akritas, Arnold and Du (2000, Biometrika). Procedures for testing for no factor effects, while adjusting for the presence of the covariate, have been extended in Tsangari and Akritas (2001) to up to three covariates. The open problems that remain are a) handling more than three covariates, and b) testing for covariate effects and interactions. The talk will review the nonparametric model and test procedures, and explain their limitations. New test procedures will be introduced that have the potential of overcoming the difficulties. The new test procedures are inspired from recent advances in the asymptotic theory for ANOVA with large number of factor levels. Sara Taskinen, Hannu Oja, and Ronald H. Randles Department of Mathematics and Statistics  University of Jyvaskyla  Finland  Abstract:  New test statistics are proposed for testing whether two random vectors are independent. Two different approaches are discussed:  First, interdirection proportions (Randles, 1989) are used to estimate the cosines of Mahalanobis angles between centered observation vectors and between differences of observation vectors. Second, covariances  between multivariate affine equivariant signs and ranks are used. The test statistics arising from these two approaches appear to be asymptotically equivalent if each vector is elliptically symmetric.  Limiting Pitman efficiencies and simulations are used to compare the  tests to the classical Wilks' test. Jun Yan  Department of Statistics  University of Wisconsin-Madison "Functional Regression Models and Temporal Processes" Abstract:  We consider response and covariates which are temporal processes observed in overlapping intervals and can be viewed as functional data.  A functional generalized linear model is proposed for the mean of the response over time.  The process means are modeled marginally, without a Markov assumption. The framework is rich, extending standard parametric and semiparametric models in multistate survival analysis.Based on the abundance of cross-sectional data provided by the continuously observed temporal processes,we propose simple estimators of the time-varying coefficients using easy-to-implement moment methods.The functional estimators are shown to be uniformly consistent and to converge weakly to Gaussian processes. The estimation procedure does not require smoothing, unlike most approaches to varying-coefficient models. The nonparametric estimators are the basis for new tests of the covariate effects. They also enable estimation of submodels in which greater structure is imposed on the parameters, resulting in partly parametric models.  The methodology is useful in goodness-of-fit testing for the partly parametricmodels, and permits predictions involving estimated components from both the functional model and the submodels. The usefulness of the modeling strategy is illustrated in recurrent event simulations, where the proposed methodology is comparable in efficiency with existing methods when the true model coefficient is constant, and performs well estimating the time-varying covariate effects when it is not.  An analysis of the prevalence of the chronic graft versus host disease in bone marrow transplant patients is carried out, and the effects of treatment and gender are found to be time-varying. The methodology has been extended to multivariate processes.  An analysis of familial alcoholism data shows that the model is useful in modeling time-varying association structures as well as the mean structure.