Inferring friendship network structure by using mobile phone data - PubMed (original) (raw)

Inferring friendship network structure by using mobile phone data

Nathan Eagle et al. Proc Natl Acad Sci U S A. 2009.

Abstract

Data collected from mobile phones have the potential to provide insight into the relational dynamics of individuals. This paper compares observational data from mobile phones with standard self-report survey data. We find that the information from these two data sources is overlapping but distinct. For example, self-reports of physical proximity deviate from mobile phone records depending on the recency and salience of the interactions. We also demonstrate that it is possible to accurately infer 95% of friendships based on the observational data alone, where friend dyads demonstrate distinctive temporal and spatial patterns in their physical proximity and calling patterns. These behavioral patterns, in turn, allow the prediction of individual-level outcomes such as job satisfaction.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

Fig. 1.

The effect of recency on self-report data. When subjects were asked to report their general proximity patterns, the survey responses were biased in favor of recent behavior. Although the correlation is lowest when only using observed behavior during the day before the survey, as we expand the observational time window to seven days, the correlation between the self-report and observed behavior increases. However, expanding the time window beyond seven days results in a decreasing correlation, leading us to conclude that subjects recall of information about their interactions begins to degrade after approximately one week.

Fig. 2.

Fig. 2.

Probability of proximity. Proximity probabilities at work and off campus for symmetric friend, asymmetric friend, and nonfriend dyads. Probability of proximity is calculated for each hour in the week and is generally much higher for friends than nonfriends. However, it is also apparent that asymmetric and symmetric friend dyads have different temporal and spatial patterns in proximity, with symmetric friends spending more time together off campus in the evenings.

Fig. 3.

Fig. 3.

Normalized extra-role histograms. The distributions of a pair of colleagues extra-role communication factor scores segmented by relationship. Ninety-five percent (21/22) of the symmetric friendships have extra-role scores above 5, whereas ninety-six percent (901/935) of symmetric nonfriends have extra-role scores below 5. The 28 asymmetric friends have more behavioral variance, drawing from behaviors characteristic of both nonfriends and friends.

Fig. 4.

Fig. 4.

A histogram of the extra-role distribution generated from behavioral data collected from September to January for two sets of dyads. The red bins represent the dyads that consistently confirmed they were not friends on both the January and May survey (n = 2153). The yellow bins represent the dyads that confirmed they were not friends on the January survey, but at least one individual named the other as a friend on the May survey (n = 32). Clearly these two sets of dyads come from distinct distributions; potential explanations for the yellow distribution could be survey error in January (i.e., the friendships existed, but were not reported in January), or that the dyads' behavior during the autumn was indicative of budding friendships that they only became aware of during the subsequent year.

Fig. 5.

Fig. 5.

Inferred, weighted friendship network vs. reported, discrete friendship network. Frame A shows the inferred friendship network with edge weights corresponding to the factor scores for factor 2, extra-role communication. Frame B shows the reported friendship network. Node colors highlight the two groups of colleagues, first-year business school students (brown) and individuals working together in the same building (red).

Comment in

Similar articles

Cited by

References

    1. Wasserman S, Faust K. Social Network Analysis: Methods and Applications. New York: Cambridge Univ Press; 1994.
    1. Kossinets G, Watts D. Empirical analysis of an evolving social network. Science. 2006;311:88–90. - PubMed
    1. Ebel H, Mielsch L, Bornholdt S. Scale-free topology of e-mail networks. Phys Rev E. 2002;66:35103. - PubMed
    1. Aiello W, Chung F, Lu L. Proceedings of the Thirty-Second Annual ACM Symposium on Theory of Computing. New York: Association for Computing Machinery; 2000. A random graph model for massive graphs; pp. 171–180.
    1. Onnela J, et al. Structure and tie strengths in mobile communication networks. Proc Natl Acad Sci. 2007;104:7332–7336. - PMC - PubMed

Publication types

MeSH terms

LinkOut - more resources