Living-laboratory for computational social science
The “Friends and Family” living laboratory study is the first Social fMRI experiment to be conducted. It is one the most comprehensive mobile phone experiment performed in academia at its time. The dataset generated from the study one of the the largest and richest dataset ever collected on a residential community, totaling in over 1 million hours of continuous data collection from mobile phones, as well as auxiliary information from additional sources. The study was conducted over a period of 15 months between March 2010 and June 2011. The study was conducted with members of a young-family residential living community adjacent to MIT.
The study touched on many aspects of life, from social dynamics to health to purchasing behavior to community organization. The two high-level topics that unify these varied aspects are: (a) how people make decisions, especially the social aspects involved in decision making, and (b) how we can empower people to make better decisions using personal and social tools.
There was a many different research questions that we would like to try and answer at large. They can be unified under two high-level themes of investigation: First is the attempt to understand how people make decisions, with the emphasis on the social aspects involved in decision making. The second theme is investigating ways to empower people to make better decisions using personal and social tools. These two themes could be applied in a broad range of contexts that touch on many aspects of life - from health and wellness to a better understanding of social dynamics; from purchasing behavior to community organization. A key objective is to do all of this in the real world, as people live their everyday life, and not under artificial laboratory conditions.
All members of the measured community were couples or families, and at least one of the members is affiliated with the university. Over the course of the study, 140 individuals were part of the subject pool. Due to the significant length of the study, some subjects had graduated while others joined the residence community, and this was a planned part of the study. At its peak, the study had close to 130 simultaneously active participants over a period of 7 months during the 2010-2011 academic year.
The study included a broad range of methodological tools, including observations, automated sensing and data collection, regular questioners, interviews, sociometric surveys, and targeted intervention experiments. The main data collection and intervention tool was a mobile-phone-centric social and behavioral sensing system deployed to the study participants. The dataset includes continuous collection of over 25 phone-based signals - including location, accelerometery, Bluetooth-based device proximity, communication activities, installed applications, currently running applications, multimedia and file system information, and additional data generated by our experimental applications. Each of the signals was sampled at a resolution appropriate to its characteristics. In addition, we collected financial information through receipts and credit card statements, performed logging of Facebook-based socialization activities, daily polling of mood, stress, sleep, productivity, and socialization, as well as other health and wellness related information, standard psychological scales like personality tests, and many other types of manually entered data by the participants.
The data enable us to construct multiple network modalities of the community - such as the phone communication network, physical face-to-face encounters network, online social network, self-reported network, and more.