Covid19 Pandemic
Are Sri Lankans Careful about CoViD19?
A partial answer with the help of Facebook Mobility Data
Yesterday, I asked my twitter-sphere the following question:
I was surprised by this answer. I expected a “reduction” in alertness from September to October, given the relative calmness around CoViD cases.
Of course, this type of poll is, at best, a very rough estimate. At worst, it can bias by all sorts of factors.
Is there a better way to answer the alertness question? Perhaps, one with data?
Mobility as Proxy for Carefulness
One proxy for “alertness” might be travel or mobility. If people are “more alert”, they might be more likely to stay at home. Or restrict travel to short, essential trips.
But, how do we know if people stayed home?
Facebook Data as a Proxy for Mobility
An interesting dataset that partially answers this question are the mobility data sets released by Facebook.
These datasets estimate the proportion of people who stayed within their region, based on Facebook use.
For Sri Lanka, an anonymised data set is available at a Divisional Secretariat Division (DSD) granularity. Here’s an example DSD:
The Y-Axis records the % of people in the DSD (Colombo) who remained within an area approximately 600m by 600m. At “peak-lockdown” this was about 54% in the DSD. Before lockdown it was about 13%, and seems to have stabilzed at about 18% in September and October.
Some Results
Here are mobility graphs for a selection of DSDs.
Colombo and Suburbs
There seems to be little difference between September 1st and October 1st. In fact, after lockdown eased, mobility increased only slightly.
There is a small uptick in immobilituy in October, and is probably due to the Minuwangoda outbreak. It would be interesting to monitor this trend.
Galle, Kandy, and Jaffna
Except for a little more noise (possibly due to a smaller amount of data), Galle, Kandy, and Jaffna, are very similar to Colombo.
…
You can look at more data at https://observablehq.com/@nuuuwan/sl-lockdown-facebook-movement-data.
Concluding Caveats
As I already said, Mobility is, at best, a partial signal for carefulness. Also, Facebook’s estimates for location might not be completely accurate. Especially for sparsely populated areas. Hence, take these results with a pinch of salt.
