During a human development crisis like Covid-19, we quickly become aware of what we don’t know. We create models, estimates and predictions, but the exact numbers, costs and outcomes, given the multitude of potential futures, is unknown. This is the heart of the problem: how can policy-makers gain new insights to improve their decision-making?
Starting in early April 2020, the Government of Sierra Leone imposed a three-day lockdown, during which people had to stay in their homes, followed by a two-week inter-district travel ban. Our team at the MIT Governance Lab (MIT GOV/LAB) and Civic Data Design Lab (CDDL) worked with the government to understand if Sierra Leoneans complied with the mobility restrictions imposed by the government to help slow the spread of Covid-19.
Answering this question requires a reliable source of mobility data, which Sierra Leone lacks. Without the necessary public data infrastructure, the government was not able to measure and analyze the real-time mobility of citizens with the confidence and accuracy needed to see if the lockdowns were successful.
To tackle this problem, we looked to a non-conventional source of mobility data: call detail records (CDRs). A CDR is automatically generated by the telecommunications equipment whenever a mobile phone makes or receives a call or sends or receives SMS messages. Research published in Nature Communications documents how CDRs, in addition to other mobility data, have been used to address a host of policy-relevant issues ranging from mapping migration patterns to understanding disease spread, with implications for guiding and evaluating public health responses during the Covid-19 pandemic.
My team at MIT, in partnership with the Government of Sierra Leone’s Directorate of Science, Technology and Innovation (DSTI), coordinated with a mobile telephone operator, Africell Sierra Leone, to obtain secure access to anonymized and de-identified CDR data. With this data, we were able to create a distance-based indicator which we used to measure daily movement in Sierra Leone before and after the government’s policies restricting movement. Below are some high-level findings responding to the policy question of whether the lockdowns had the intended effect.
Figure 1: Percentage drop in mobility before versus during the first lockdown
The map in Figure 1 shows the change in average distance traveled for inter-district trips before and after the first three-day lockdown April 5-8, 2020. Across all districts, this change is negative, which means that on average, citizens decreased their travel between districts. Districts in dark blue show where the largest changes occurred (up to 70% reduction in some districts, showing high compliance with the lockdown), and light blue shows the least amount of change in mobility (lower compliance).
Figure 2: Interactive chart of district mobility
The interactive chart in Figure 2 shows the daily seven-day moving averages of the average distance travelled (for inter-district trips) in each district per day (use the drop down menu to select your district of interest). The blue line represents the district you have selected and the black line represents the national average. You can see for most if not all districts there are demonstrable decreases in the average distance traveled both after the three-day lockdown was put into place and after the 14-day inter-district travel ban went into effect. Click through to see the performance of districts.
These two visualizations demonstrate that citizens across districts complied with the mobility restrictions imposed by the government. Our collaboration with the government during the early stages of the pandemic helped to support decision-making during the crisis. These visualizations and the measures we derived would not be possible without having access to a fine-grained, time-series source of mobility data, made possible by Africell. CDRs do have their limitations as analytical tools, including privacy concerns. But for many countries, they remain valid and effective tools in a policymaker’s decision-making toolkit.
Coming soon: a detailed guide on how to use mobility data and CDR during crises, and a book chapter in Urban Informatics and Future Cities with more detail on our research collaboration in Sierra Leone.