A team has studied how people react to smoke from wildfires using mobile phone location data

This research represents a very interesting contribution to the study of behavior in critical situations. The use of mobile geolocation data to measure actual behavior—rather than relying on surveys or self-reports—allows for a precise observation of how people adjust their response time to wildfire smoke. The sample is large (163,000 devices), and the design leverages the unpredictable nature of smoke plume movement as a source of quasi-experimental variation, which strengthens the causal validity of the findings.

Its most significant contribution is distinguishing between two mechanisms: the ability to adapt versus risk awareness. 

Let me explain: when we see that people with higher levels of education protect themselves more from smoke, we might think two things: either they have more options for staying home (flexible jobs, the ability to work remotely), or they are more aware that smoke is dangerous. To determine which of the two explanations is correct, the researchers used a clever approach: they also looked at what people do when it rains. Rain, like smoke, is something that’s clearly visible and encourages people to stay home. But unlike smoke, everyone knows that getting wet is uncomfortable—you don’t need studies to understand that. 

The result was revealing: when it rains, both people with college degrees and those without stay home equally. In contrast, when there is smoke, people with higher levels of education protect themselves much more than others. 

If the difference were due to having more flexible jobs, we would see the same pattern with rain. But that is not the case. Therefore, the most likely explanation is that people with higher levels of education are more aware of the health risks of smoke and therefore react more strongly. 

The most concerning finding is that people do not begin to take protective measures until smoke levels exceed 50 µg/m³, even though adverse health effects begin at much lower levels (10–25 µg/m³). This indicates a significant gap between the moment when smoke becomes noticeable or “obvious” and the moment when it is already harmful. Combined with the steep educational gradient, the message for public policy is clear: air quality alerts and awareness campaigns must be issued earlier, be more accessible, and be specifically designed to reach populations with lower educational levels, where risk awareness appears to be lower.