In the latent class analysis, it is assumed that each of the observed indicators is caused by the unobserved indicator or latent variable. The patterns of interrelationships among the observed indicators are analyzed to inspect the underlying latent variable. The source of the relationships between the observed variables is assumed to be the latent variable. The interrelationships among the observed indicators can be thought as conditionally independent by controlling the determined latent variable. The analysis depends on the interpretations of the latent class probabilities, conditional probabilities and the odds ratios. In this study, the microdata “Information and Communication Technology (ICT) Usage Survey on Households and Individuals” obtained from the research of Turkish Statistical Institute in 2012 are used. Firstly, the profile of internet using in Turkey is determined by using the descriptive statistics and a set of mutually exclusive latent classes for the individuals’ internet usage activities in Turkey is identified by the latent class analysis.
Keywords: Categorical Data Analysis, Latent Class Analysis, Latent Class Probabilities, Conditional Probabilities