Tuesday, January 22, 2013

Cell phones track human migration

Here’s a question every scientist at some point asks themselves: does this data that I can easily and (relatively) inexpensively collect reasonably approximate the data that I would collect in an ideal world where I had bucket loads of money and an infinite amount of time? It may not be apparent from science news coverage, but a lot of science involves routinely checking that the methods we are using to investigate a question will actually result in an answer. So here’s a story about the unsung iceberg hiding beneath any good scientist’s crowning achievement (of which only a few ever make it into the press, anyway).

Like any other animal, humans move around. They commute from home to work, travel to see their grandparents, move to new cities, vacation in Tahiti... you get the idea. A lot of epidemiologists, sociologists, and airline companies would love to know why they do so.  But, before you can predict why people go where they do, there has to be good data that show where people actually go. Unlike wolves, humans are somewhat averse to GPS collars that track their every move, so the easiest data for researchers to access is national censuses that ask whether a person has changed residences within the last year.


But, do all of the permanent relocations that happen within any given year represent the day-to-day, or even week-to-week movements of people who don’t necessarily change homes? Census data might not be fine-scale enough to capture the kinds of movements that are important, for example, in the spread of disease. Luckily, people are willing carry little transmitters with them everywhere they go, periodically sending their locations throughout the day to the companies that provide these devices. (I’m talking about cellphones, all 6 billion of them.) Every time someone makes a call or sends a text, the nearest cellphone tower records their location and the location of the tower closest to the person they are calling. And cell phone companies have access to these records- the only thing standing in the way of this data and science is a little thing called privacy.


Cell records being difficult to access, graduate student and researcher Amy Wesolowski and colleagues decided to test whether freely available census data is able to approximate the movements shown by the fine-scale cell phone data. They were able to obtain cell phone data from Kenya during the year than matched the most recent national census and used this to calculate how many trips cell phone users made between different counties in Kenya. They then compared these trips to the relocations recorded in the census data. In the end, the movements recorded by the census were surprisingly similar to trips made by cell phone users, which suggests that, after some more validation, censuses could be useful for modeling movement over shorter time periods.


The yearly movements of humans in a moderately small African nation may not fall under the traditional realm of biogeography, but I present this research because all too often, the natural sciences categorize humans as an external force on whatever system is being studied. As a scientist, sometimes it is important to recognize that the natural laws we study also apply to us. Dispersal is the foundation of all biogeographic patterns and humans do it more than any other species on the globe.


You can find this article at:

ResearchBlogging.orgWesolowski, A., Buckee, C., Pindolia, D., Eagle, N., Smith, D., Garcia, A., & Tatem, A. (2013). The Use of Census Migration Data to Approximate Human Movement Patterns across Temporal Scales PLoS ONE, 8 (1) DOI: 10.1371/journal.pone.0052971

No comments:

Post a Comment