N. Eagle, Y. de Montjoye, and L. Bettencourt (2009), "Community Computing: Comparisons between Rural and Urban Societies using Mobile Phone Data", IEEE Social Computing, 144-150.

Abstract of the paper

We present a comparative analysis of the behavioral dynamics of rural and urban societies using four years of mobile phone data from all 1.4M subscribers within a small country. We use information from communication logs and top up denominations to characterize attributes such as socioeconomic status and region. We show that rural and urban communities differ dramatically not only in terms of personal network topologies, but also in terms of inferred behavioral characteristics such as travel. We confirm the hypothesis for behavioral adaptation, demonstrating that individuals change their patterns of communication to increase the similarity with their new social environment. To our knowledge, this is the first comprehensive comparison between regional groups of this size.

Summary

This paper presents the first quantitative comparison of urban and rural communities based on a complete mobile phone graph of an entire nation. The authors begin with an overview of related work, detailing both the study of rural and urban societies within the sociology and social psychology literature, as well as previous studies involving mobile phone data. They then provide a detailed description of the data and list both the individual and social network attributes associated with urban and rural communities. Individuals who have moved between urban and rural communities are identified and they measure how their different attributes change in response to their environment. Finally, they made a conclusion with a discussion on the potential of this type of data for a wide range of additional questions.

Related Works

The Effects of Cities on Personal Networks

• Many of the foundational concepts of sociology, social psychology and economics have originated from the observation that the transition of populations from rural areas to urban centers results in both behavioral and socioeconomic changes, such as researches based on observations of urbanization.
• There are also some qualitative observations imply specific measurable consequences for the structure of social networks and to their geographic variation, such as publications of Claude S. Fischer.

Behavioral Studies using Mobile Phone Data

• The recent analysis of data from mobile phone service providers have led researchers to increased insight into human movement patterns. While some researchers take issue with labeling these insights as ’universal laws of human movement’, it is clear that through the analysis of cellular tower location data from hundreds of thousands of people, it is possible to finally quantify some of the more fundamental rules of human motion, such as works in "Understanding individual human mobility patterns"

Data Description

The data used in this paper consists of four years of call data records (CDR) for every mobile phone subscriber within a small country. They do not have access to phone numbers, but rather unique IDs that provide no personally identifiable information. Besides the standard information within CDR including voice and text-message communication and location estimates based on cellular towers, they also have access to additional subscriber data, including pre-paid scratch card denominations, air-time sharing, product adoption data, and phone model. The data is similar to that in Reality Mining project at MIT Media Lab.

Methodology

Segmenting Regions

• They divide the 1.4 million subscribers into three categories based on geography: those living in the country’s capital (600k), the other 11 urban towns (500k), and rural areas (300k). The regional ties based on calling volumes.

Identifying Individuals

• To establish an individual’s regional label and weight, we identify the region where the individual spent the majority of time based on the cellular tower data for each week. For example, an individual who pends 3 weeks in the capital and then spends a week split between the capital and the rural area would have the regional label of ’Capital’ with a weight of 0.875. The individual is then associated with the most probable region, and subsequently these weights are no longer used in this analysis.

Results

The different results are obtained through One-way ANOVA.

Individual Behavior Attribute

• Socioeconomic Status via Calling Card Denominations
• Travel: Distances between Cellular Towers
• Product Adoption: Air time Sharing

Social Network Attribute

• Frequency and Volume
• Degree and Average Volume per Degree
• Alter Attributes

Behavioral Persistence

They find results in support of the behavioral adaptation hypothesis. Individuals migrating to the city subsequently increase their call frequencies matching the behavior expected for a typical urban dweller.

Discussion

This paper represents an initial analysis of how mobile phone data can provide comparative insight into human and social behavior in urban and rural communities. The authors have tested, confirmed and quantified classical hypothesis in sociology, social psychology and economics that urbanization leads to increased communication, and present a methodology for inferring socioeconomic status based on airtime top-up denominations. They have also confirmed hypothesis for behavioral adaptation of individuals based on changes in their patterns of communication to increase the similarity with their new social environment. Such results will advance the conceptual framework in the social sciences and economics and may result in new approaches to public policy.

Related Publications

• N. Eagle, A. Pentland, and D. Lazer (2009), "Inferring Social Network Structure using Mobile Phone Data", Proceedings of the National Academy of Sciences (PNAS) Vol 106(36), pp. 15274-15278.
• N. Eagle (2008), "Behavioral Inference Across Cultures: Using Telephones as a Cultural Lens", IEEE Intelligent Systems 23:4, 62-64.
• N. Eagle and A. Pentland (2006), "Reality Mining: Sensing Complex Social Systems", Personal and Ubiquitous Computing, Vol 10, #4, 255-268.