Date of Award

August 2018

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Business Administration

First Advisor

Fatemeh Mariam Zahedi

Committee Members

Sanjoy Ghose, Derek Nazareth, Huimin Zhao

Keywords

dynamic sequential data extraction and integration, health-related lifestyle behavior, location-based information disclosure, online social networks, social influence

Abstract

More than half of U.S. adults suffer from one or more chronic diseases, which account for 86% of total U.S. healthcare costs. Major contributors to chronic diseases are unhealthy lifestyle behaviors, which include lack of physical activity, poor nutrition, tobacco use, and drinking too much alcohol. A reduction in the prevalence of health-risk behaviors could improve individuals’ longevity and quality of life and may halt the exponential growth of healthcare costs. Prior studies in the field have acknowledged that a comprehensive understanding of health behaviors requires the examination of individual’ behaviors in supra-dyadic social networks. In recent years, the growth of online social networks and popularity of location-based services have opened new research opportunities for observational studies on individuals’ healthy and unhealthy lifestyle behaviors. The goal of this three-essay dissertation is to examine the effect of various social factors, shared images, and communities of interest on healthy and unhealthy lifestyle behaviors of individuals. This dissertation makes novel contributions in terms of theoretical implications, data collection and analysis methods, and policy implications for promoting healthy lifestyle behaviors and inhibiting unhealthy behaviors.

Essay 1 draws on a synthesis of social cognitive and social network theories to conceptualize a causal model for healthy and unhealthy behaviors. To test the conceptualized model, we developed a new method—dynamic sequential data extraction and integration—to collect and integrate data over time from Twitter and Foursquare. The captured dataset was then combined with relevant data from the U.S. Census Bureau. The final dataset has more than 32,000 individuals from all states in the United States. Using this dataset, we derived variables to measure healthy and unhealthy lifestyle behaviors and metrics for factors representing individuals’ social support, social influence, and homophily, as well as the socioeconomic status of the communities where they live. To capture the impacts of social factors, we collected individuals’ behaviors in two separate time periods. We used zero-inflated negative binomial regression method for data analysis. The results of this study uncover factors that have significant impacts on healthy and unhealthy lifestyle behaviors.

Essay 2 focuses on embedded images in self-disclosed posts related to healthy and unhealthy lifestyle behaviors. While online photo-sharing has become widely popular, and neuroscience has reported the influence of images in brain activities, to our knowledge, there is no published research on the impacts of shared photos on health-related lifestyle behaviors. This study addresses this gap and examines the moderating role of shared images and the direct impacts of their contents. We relied on social learning and multimodality theories to argue that images can attract individuals’ attention and enhance the process of observational learning in online social networks. We developed a novel method for image analysis that involves the extraction, processing, dimensionality reduction, and categorization of images. The results show that the presence of photos in self-disclosed unhealthy lifestyle behaviors positively moderates friends’ social influence. Moreover, the results indicate that the contents of shared photos influence individuals’ health-related behaviors.

Essay 3 focuses on the role of personal interests in individuals’ health-related lifestyle behaviors. Prior studies have demonstrated that health promotional programs can benefit from targeting individuals based on their interests. Specifically, prior studies have emphasized the role of interests as a factor influencing behaviors. However, current literature suffers from two major gaps. First, there is no systematic and comprehensive approach to capture individuals’ interests in online social networks. Second, to our knowledge, the role of interests in individuals’ healthy and unhealthy lifestyle behaviors as disclosed online has not been investigated. To address these gaps, we examine the role of individuals’ interests in their health-related behaviors. The theoretical foundation of this study is a synthesis of homophily and self-determination theories. We developed a novel method—the homophily-based interest detection method—that involves network simplification, network clustering, cluster labeling, and interest metrics. This method was applied to social networks of individuals in Essay 1 to measure individuals’ interests. The results show that health-related interests are associated with individuals’ healthy and unhealthy lifestyle behaviors. Our findings indicate that other forms of interest, such as music taste and political views, also play a role. Moreover, our results show that belonging to healthy (unhealthy) communities of interest has an inhibitive role that prevents postings of unhealthy (healthy) behaviors.

Share

COinS