Developing a General Method for Analyzing Psychomotor Vigilance Task (PVT) Data: Modeling Sleep Inertia in Children

Abstract

Some data analysis applications may violate the assumptions of standard (e.g., linear model) frameworks. In such cases, a common solution is to develop a more suitable system-specific model, and from it derive the statistical tools necessary to conduct the desired data analysis. In this thesis, we propose a method to model Psychomotor Vigilance Task (PVT) data which is a test to measure alertness and is commonly used in psychology. In particular, we focus on applying these methods to data from studies of measuring sleep inertia in children. We will start by looking at relevant methodological and application specific background. We will then look at basic statistical analysis such as mean, median, and standard error and plot some examples to get a sense of the data. Then we will introduce a flexible model that better represents sleep inertia. We describe a maximum likelihood based estimation procedure and analyze parameter identifiability. Lastly, we use the likelihood ratio test to compare nested models.