Nonparametric Regression Method for Agreement Measure Between an Ordinal Measurement and a Continuous Measurement
Document Type
Presentation Abstract
Presentation Date
2-29-2016
Abstract
In this talk, I will present a non-parametric regression framework for an agreement measure between an ordinal measurement and a continuous measurement given a covariate. Unlike in the usual non-parametric regression methods, agreement measure is not observed as a random variable for each subject, thus commonly used non-parametric methods such as kernel methods are not directly applicable. Adopting the idea of stratified sampling in the framework of kernel regression, I will present the proposed inferential procedures including regression function estimation, asymptotic properties, and hypothesis testing. Finite sample performance of the proposed method will be illustrated via simulation studies. I will also present an illustration of the proposed method to a recent posttraumatic stress disorder (PTSD) study which reveals an interesting impact of depression severity on the agreement measure between a self-reported symptom instrument score (continuous measure) and clinician diagnosis (ordinal measure) in PTSD patients.
Recommended Citation
Rahman, AKM Fazlur, "Nonparametric Regression Method for Agreement Measure Between an Ordinal Measurement and a Continuous Measurement" (2016). Colloquia of the Department of Mathematical Sciences. 502.
https://scholarworks.umt.edu/mathcolloquia/502
Additional Details
Monday, February 29, 2016 at 3:10 p.m. in Math 103
Refreshments at 4:00 p.m. in Math Lounge 109