Inference and Modeling Aspects of Multiple Ranked Lists

• Activity Number: 281
• Type: Topic Contributed
• Date/Time: Tuesday, August 5, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Michael G. Schimek*+ and Peter Hall and Vendula Svendova
• Companies: Medical University of Graz and University of Melbourne and Medical University of Graz
• Keywords: Nonparametric Inference ; Multiple ranked lists ; Parameter estimation ; R package ; Simulation ; Top-k ranked lists

Abstract

In recent years there has been an increasing interest in the statistics of ranked lists. Typically, lists from high-tech (e.g. omics) comprise tens of thousands of items. However, only a comparably small subset of k top-ranked items is informative. These items are characterized by a strong overlap of their rank positions when the lists are ranked by different assessors. A central task is the identification of these top-k items (Hall and Schimek, 2012). Most recently, a strategy for inference in multiple ranked lists has been developed (R package TopKLists). We will outline these new findings and introduce a novel tool for the graphical representation of the obtained inference results. The rarely considered but practically relevant case of dependencies across lists will be covered too. Another quite demanding task is the modeling of long multiple lists. Conventional model-based approaches are not practicable because the number of rankings is rather small in those data we focus on compared to the lengths of such ranked lists. Finally, we outline a new simulation-based concept for the estimation of parameters shared by a number of independent assessors that have ranked the same items.