A Zero-Inflated Ordinal Logistic Regression Model in Examining the Relationship Between Participation in Social Activities and Media Consumption

• Activity Number: 552
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Social Statistics Section
• Author(s): Jihyung Shin*+
• Companies: Korea Infomation Society Development Institute
• Keywords: Zero-inflated Data ; Korea Media Panel Survey ; Media Usage Behavior ; Social Activities ; Media consumption

Abstract

As the purpose of media usage is not limited to communications or information seeking behaviors but can be extended to leisure or cultural activities, certain types of social activities can be characterized by the substitution or complementary in relation to the media consumption. In examining the relationships between frequency on social activities and media consumption, the data showed excessive zeros on the number of participation in certain social activities. Such data with excessive zeros are often observed in ordinal categorical response variables, but ordinary models for ordered categorical variables are not appropriate to interpret the data within the context. Two staged zero inflated ordinal logistic regression models are proposed for ordered categorical variables with excessive zeros, and applied to social activities data from Korea media panel study data. The first stage of the model separates zero responses into non-participants and zero participants for survey periods, and the second stage of model explains the intensity of participation in relation to the media consumption. The application results are presented.