|
Activity Number:
|
427
- Contributed Poster Presentations:Government Statistics Section
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2018 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Government Statistics Section
|
|
Abstract #327091
|
|
|
Title:
|
Latent Class Analysis with a Calibrated Conditional Independence Assumption
|
|
Author(s):
|
Joseph Kang* and Tandin Dorji
|
|
Companies:
|
Centers for Disease Control & Prevention and Oak Ridge Institute for Science and Education (ORISE)
|
|
Keywords:
|
Latent Class Analysis;
Complex Survey Data;
Clustering;
Calibration;
Weights ;
Conditional independence assumption
|
|
Abstract:
|
Latent class analysis (LCA) is used to effectively cluster multivariate categorical data. LCA is a parametric model that requires the strong assumption of the local independence among manifest item variables within a latent class. This presentation will: 1) discuss conventional methods that handle the violation of the conditional independence assumption and 2) propose a novel way of calibrating the validity of the assumption. A practical example is clustering disease comorbidities using the National Health and Nutrition Examination Survey (NHANES), which is a complex survey data set. In particular, clustering of multivariate sexually transmitted diseases will be presented. This presentation will describe insightful results from the LCA-based assessment of disease clusters and its calibrated analysis.
|
Authors who are presenting talks have a * after their name.
