Generalized behavioral framework for choice models of social influence: Behavioral and data concerns in travel behavior

Highlights

• Our proposed framework generalizes specifications of social influence choice models.

• We critically review social influence models in transportation using this framework.

• We find the processes of and motivations for social influence need to be emphasized.

• Social networks play a critical role in influence but often are an afterthought.

• We propose future research in influence mechanisms, networks, and decision rules.

Abstract

Over the past two decades, transportation has begun a shift from an individual focus to a social focus. Accordingly, discrete choice models have begun to integrate social context into its framework. Social influence, the process of having one’s behavior be affected by others, has been one approach to this integration. This paper provides a review and discussion of the incorporation of social influence into discrete choice models with specific application in travel behavior analysis. The discussion begins with a generalized framework to describe choice models of social influence. This framework focuses on the behavioral microfoundations of social influence and choice by separating the social influence mechanism from the source of its influence and by explicitly acknowledging the role of the social network in the model structure. This contrasts with prior work that focused on the measurement of contextual, endogenous, and correlated effects. Then, the state of the art in travel behavior research is reviewed using a taxonomy based on the generalized framework with research performed in sociology, social psychology, and social network analysis. The discussion then shifts to the importance of understanding the motivations for social influence, and the formation and structure of social networks are explored. Additionally, the challenges of collecting data for social influence studies are mentioned and the paper concludes with a look at the challenges in the field and areas for future research.

Introduction

Travel is an integral part of peoples’ lives which connects their residences and neighborhoods, work and economic opportunities, and geographical points of reference such as school, childcare, shopping, healthcare, and leisure. Increasingly, transportation researchers have become interested in the role of social interactions between people in a given individual’s travel behavior (Dugundji et al., 2008, Dugundji et al., 2011a, Dugundji et al., 2012). Borrowing from the field of economics (Durlauf and Ioannides, 2010), social interactions are defined as “direct interdependences in preferences, constraints, and beliefs of individuals, which impose a social structure on individual decisions” (p. 452).

Within travel behavior research, the literature on social interactions is becoming relatively well-established. But recently, there has been growing interest in decisions involving social influence.¹ Social influence deals with how an individual’s decision making process is altered by others’ actions, behavior, attitudes, and beliefs of others (and the individual’s perceptions of these). Of particular interest is the analysis of models in which the decisions of others are incorporated into discrete choice models. Since travel may involve different types of social influence from peers, family, neighbors, colleagues, and even society at large, incorporating these social effects into discrete choice models is non-trivial. These models are grounded in theories of individual choice of independent decision makers. Additionally, they are generally estimated on cross-sectional, choice-based data sources which make it difficult to identify social influence effects and their motivations. These motivations are important for understanding long-run behavior and for guiding organizations on appropriate intervention strategies to encourage behavioral change.

The incorporation of social networks, the types and timing of interactions, and how social networks and interactions interface in spatial dimensions are difficult to model and identify from current data sources. Social influence models use a wide variety of network structures, varying from cliques to sparse networks, and the connections made can be due to similarity in social standing and interests and spatial proximity. Individuals’ networks are also bounded by limitations in cognitive effort, time, and space. The spatial dimension of social networks is still an open research field and its use in transport models of social influence has been limited both in its actual application and its simplicity.

With an emphasis on behavioral and data issues, this paper aims to provide a behavioral framework for describing choice model approaches for decisions involving social influence. The paper begins with a quick example of how a simple hypothesis can be explained by various social and non-social factors. In Section 3, a generalized behavioral framework for choice models of social influence is introduced. Section 4 describes past research in travel behavior using this framework and describes the shortcomings in current models in the need to understand the motivations behind social influence. Sections 5 Social influence mechanism: Types, motivations, and tactics, 6 Social networks: Process and structure, 7 Data collection for social influence, networks, and sources describe the framework’s components of social network, social influence mechanism, and influence sources. Section 5 summarizes recent research on the types, motivations, and tactics of social influence. Section 6 describes the behavioral processes behind social network formation and common structural forms and Section 7 summarizes procedures for gathering social influence and social network data. The paper concludes with a summary and areas for future research.