Skip to Main Content

An official website of the United States government

Time-Varying Measurement Error Models

Principal Investigator
Name
Raymond Carroll
Degrees
Ph.D.
Institution
Texas A&M University
Position Title
Distinguished Professor of Statistics
Email
About this CDAS Project
Study
IDATA (Learn more about this study)
Project ID
IDATA-10
Initial CDAS Request Approval
Mar 20, 2017
Title
Time-Varying Measurement Error Models
Summary
I have a long history of working on measurement error problems with researchers at NCI (named under collaborators). I am working with them on a project of time-varying measurement error models, which attempts to understand the measurement error properties on the instruments in iData. To date I have no access to that data, and I would like to do alternative analyses based on notions of functional data analysis (where the data are functions measured with error) and longitudinal data, while taking into account the multivariate nature of the iData (different devices, questionnaires and and web-based 24 hour recalls, plus doubly-labelled water).
Aims

(a) Fit models to the iData observations that account for correlation over time, and correlation among instruments and measurement error in those instruments.
(b) Understand the variability in the physical activity accelerometers and the different patterns of physical activity they measure. Different patterns do arise in such data because of the different times people are active, and it is a challenge to create metrics that can be used to summarize such activity for future use in cohort studies.
(c) Write papers, primarily in Statistics journals, on the new models and the numerical results.

In collaboration with others

Collaborators

(1) Victor Kipnis, Doug Midthune, Kevin Dodd, NCI-DCP (time varying)
(2) John Staudenmayer, University of Massachusetts, Amherst
(3) Ya Su (postdoc), Tianying Wang (graduate student), Alex Asher (graduate student), Eli Kravitz (graduate student), Texas A&M University