Activity Number:
|
70
- Utilizing High-Dimensional and Complex Data in Personalized Medicine
|
Type:
|
Contributed
|
Date/Time:
|
Sunday, July 30, 2017 : 4:00 PM to 5:50 PM
|
Sponsor:
|
Mental Health Statistics Section
|
Abstract #323837
|
View Presentation
|
Title:
|
Assessing Latent Dimensionality of a Symptom Battery in a Sample with Many Subjects Exhibiting No Symptoms
|
Author(s):
|
William Christensen* and Melanie M Wall and Irini Moustaki
|
Companies:
|
Brigham Young University and Columbia University and London School of Economics
|
Keywords:
|
factor analysis ;
latent variable model ;
zero-inflated ;
dimensionality
|
Abstract:
|
Common methods for determining the number of latent dimensions underlying a symptom battery or item set include eigenvalue analysis and examination of fit statistics for factor analysis models with varying number of factors. Given a set of dichotomous items, we will demonstrate that these empirical assessments of dimensionality are likely to misestimate the number of dimensions when the data represent a mixed sample of pathological and healthy subjects. We demonstrate the problem is exacerbated when there is a preponderance of individuals in the sample with no symptoms. Several dimension-assessment tools are considered and simulated data experiments are conducted to demonstrate their relative utility when dealing with a mixed-status sample. Recommended approaches are illustrated using an example assessing the dimensionality of a social anxiety disorder symptom battery.
|
Authors who are presenting talks have a * after their name.