PARTITIONING METHODS IN LATENT CLASS ANALYSIS,

Abstract

A method of obtaining estimates of the parameters of the latent class model based on first classifying individuals into 'latent classes' according to some rule is presented. Properties of the estimates and implications to the identifiability problem are discussed. An example of the use of the partitioning method, contrasting it with the determinantal method, is given. (Author)

Document Details

Document Type
Technical Report
Publication Date
Mar 06, 1959
Accession Number
AD0701794

Entities

People

  • Albert Madansky

Organizations

  • RAND Corporation

Tags

Readers

  • Linear Algebra
  • Neural Network Machine Learning.
  • Theoretical Analysis.