The Analysis of Ordered Response Data and Model Diagnostics

Speaker:Dr. Ivy Liu, Victoria University of Wellington, New Zealand

Time: 10:00 am - 11:00 am, Jan. 07(Mon.), 2013

Venue: E303B

Abstract:

Measurement data with ordinal categories occur frequently: e.g. in medicine a continuous clinical response is often categorised into an ordered set of subtypes based on histological or morphological terms, and in a questionnaire Likert scale responses might be 'better', 'unchanged' or 'worse'. Such data are easy to collect, and are often appealing to human respondents who may be happier to rank their responses than provide absolute scores. However, researchers often use wrong models by assigning scores to the ordinal outcomes. This talk has two parts. First, I will review methodologies used for analyzing ordered categorical response variables. The second part will discuss some graphical and numerical methods for checking the adequacy of models, including using a process that converges in distribution to a Brownian motion.