At company called unacast has used cellphone tracking data to produce a “Social Distancing Scorecard.” It breaks down the US by state and county to measure the distances that cellphones travel as a proxy for social distancing. It then grades areas from A to F based on the percentage decrease. Obviously this is a rough proxy social distancing, but it’s currently the best we have.
It’s interesting to note geographical disparities. Eyeballing it, urban areas are doing more social distancing than rural ones. The Mountain West is doing less. All of that tracks what we know about how different demographics and political orientations are responding to the virus. Interestingly, Alaska is doing much more than most other areas, with the exception of the far north.
My own Mecklenburg County (which includes the city of Charlotte, though the metropolitan area encompasses several other counties, and which just issued its stay-at-home order today) gets an ‘A,’ for a reduction of 43%, despite the County Health Director’s repeatedly telling media that people aren’t adhering to social distancing guidelines. Next-door union County gets a ‘B,’ but that’s for a 40% reduction. There are a couple of counties in the eastern part of the state that have seen an increase. The state as a whole gets a ‘B’ for a 36% reduction – probably in part because the governor has been quite proactive in pushing social distancing. Ohio is down 40%, presumably also because of a proactive governor. It wouldn’t be hard to use this data to track the effects of specific policy interventions.
If you go down to Florida, the Miami area (and the cities up the coast) are all down 50% or so. The Panhandle, where a lot of WOO HOO SPRING BREAK happens are doing a lot less distancing.
And so it goes. The Washington Post has a write-up that gets at some of the privacy and other issues. It should be noted that social media data is already useful in public health surveillance. To cite only one example, one study found that content analysis of Twitter was able to predict community cardiovascular disease mortality – a significant result, since the average age of Twitter users is quite low, and (as the authors put it) “the people tweeting are not the people dying,”
Recent Comments