Parking lot statistics—a story in three parts

Part 1:

Here’s a 1993 article from the American Sociological Review in which church attendance was measured by the number of cars in the parking lot (link from here).

Part 2:

In 2005 or 2006, an economist who does statistics reportedly tries to run over a sociologist who does statistics in a parking lot (but see denial here).

Part 3:

Commenter Zhao Fang argues that a recent preprint that claims evidence of coronavirus cases in China in 2019 is flawed because of misleading parking lot photos (see image above).

The preprint in question, “Analysis of hospital traffic and search engine data in Wuhan China indicates early disease activity in the Fall of 2019,” has perhaps received extra attention and respect because the corresponding author is a professor at Harvard medical school (and also the “Chief Innovation Officer” (?) of Boston Children’s Hospital).

Here’s their graph summarizing the car-park data:

The lines jump around a lot so it’s not clear what to make of this. Much depends on the data quality. They say, “Images with tree cover, building shadow, construction and other factors that present difficulties in defining the contours were excluded since this could lead to over- or under-counting of the number of vehicles.” I guess they should just post all their images online and then other people can check their counts.

I really have no idea what to think on this one. On one hand, Fang’s parking lot image is pretty compelling. On the other hand, just cos a paper has an author from Harvard, we shouldn’t automatically assume the data are junk. They do some good work at Harvard too, right? We should be careful not to judge a paper too harshly just because of the institutional affiliations of its authors. For one thing, the other authors of the paper are from Boston University, which maybe has less of a reputation for data opportunism?

In all seriousness, I know nothing about Chinese hospital parking lots, and I know nothing about the progress of coronavirus in 2019. The point here is not that this work is wrong of that ABC News should not have featured it (in that report, it says that one of the authors of the study is “an ABC News contributor,” whatever that means).

My take-home point here is that the authors should post all their data. The ABS news report says that “the study has been submitted to the journal Nature Digital Medicine and is under peer review.” So this is easy. If Nature Digital Medicine (or any journal) is interested in publishing this paper, they should make it a condition that the authors post all their data, including the images they excluded from their analysis.

Was there a mini-epidemic of coronavirus in Wuhan in late 2019? This is an important question, important enough that all data should be shared.