Keep Your Friends Close and Your Facebook Friends Closer: A Multiplex Network Approach to the Analysis of Offline and Online Social Ties

Desislava Hristova, Mirco Musolesi, Cecilia Mascolo

Researchers from the University of Cambridge and the University of Birmingham have investigated the intersection of digital and physical social lives using a "multiplex network" approach. Their study examines whether online social circles, specifically Facebook, act as echo chambers or if they expose users to more diverse perspectives than their offline interactions.

The Multiplex Graph Model

Social scientists have long observed homophily, the tendency for similarity to foster friendship. However, quantifying the "strength" of a social tie across different communication channels is challenging. To address this, the authors employ a Multiplex Graph Model. Rather than treating all interactions as a single link, they represent the social network as an ensemble of distinct layers: physical proximity (detected via Bluetooth), phone calls, and SMS logs.

Multiplex Graph Model Representation

This model allows for the calculation of a Multiplex Weight ($mw_{ij}$), representing the fraction of communication channels utilized between two individuals. A weight of 1.0 indicates that a pair interacts across all three layers (Calls $\cap$ SMS $\cap$ Proximity), signifying a stronger social tie.

Dataset and Interaction Patterns

The study utilizes the MIT Social Evolution dataset, which tracks 74 students in a dormitory. The data includes sensor-based logs (Bluetooth, calls, SMS) and survey data regarding political orientation, music interests, and health habits.

Degree Distributions across Communication Layers

The degree distributions across these layers reveal varying levels of interaction frequency. The researchers found that multiplexity is a powerful predictor of intimacy; there is a 75% probability that a pair interacting in the intersection of all three layers is a "close friendship."

The Diversity Paradox

A key finding of the research is the "Diversity Paradox." While social media is often feared to be an echo chamber, the data suggests that students' Facebook circles were actually more diverse in terms of political opinions and music tastes than their immediate offline circles. Conversely, for "situational" factors like residential floor or graduation year, Facebook networks were less diverse, suggesting social media is primarily used to maintain connections with immediate peers.

Key Results and Implications

  • Tie Strength Prediction: The multiplex model successfully identified 90% of reported Facebook friendships and 75% of close friendships.
  • Similarity Correlation: A significant positive relationship exists between communication multiplexity and profile similarity (Spearman rank correlations between 0.73 and 0.81).
  • Diversity Gains: Over 50% of students possessed more politically varied networks on Facebook than in their physical offline interactions.

Limitations and Future Work

The study's primary limitation is its small, relatively homogeneous sample size of 74 MIT students. Additionally, the research focuses on network structure rather than content consumption; algorithmic filtering on Facebook might still limit exposure to the diversity present within the network. Future research could apply this multiplex approach to larger datasets and incorporate content-layer analysis to determine if network diversity translates to diverse information consumption.

Summary updated

BibTeX
@inproceedings{hristova2014keep,
  author = {Desislava Hristova and Mirco Musolesi and Cecilia Mascolo},
  title = {{Keep Your Friends Close and Your Facebook Friends Closer: A Multiplex Network Approach to the Analysis of Offline and Online Social Ties}},
  booktitle = {ICWSM},
  year = {2014},
}
Figure