Air pollution is a worldwide challenge. Despite the serious environmental and health effects, pollution can go unnoticed until very dangerous levels are reached. Rita Abdel Sater, Mathieu Perona, Elise Huillery and Coralie Chevallier use a novel and safe method to make pollution more visible, and demonstrate that personalised feedback can change household behaviours.
The Great Smog of London: when pollution became impossible to ignore
On the morning of December 5th 1952, high-pressure weather conditions, along with residential and industrial fires created a lethal smog that covered the city of London. As the day progressed, the smog became thicker and by evening, it had become so dense that people couldn’t see their own feet. Smoke particles covered clothing, vehicles were abandoned in the street, and residents struggled to breathe. The Great Smog of London went on for 5 days, and is estimated to have killed 12,000 people.
Although concern had been growing about the negative health impact of air pollution for years, it took a devastating event to change public perception and pressure the Government to enact the Clean Air Act of 1956. The Great Smog became a turning point in the history of environmentalism.
London’s pollution would never be so visible again, but the problem did not stop in the 1950s, nor is it limited to London. Around the globe, pollution remains deadly, and mostly invisible. But how can we make it visible without recreating one of the worst environmental disasters?
In our study, we gave air quality micro-sensors to wood-burning households in Paris and showed that it helped make pollution more visible, leading to changes in perceptions and polluting behaviours.
Pollution from residential combustion is largely misunderstood by the public
From movie scenes to holiday postcards, fireplaces are symbols of warmth, wellbeing, nature, and togetherness. But the reality is less romantic. Even occasional residential wood-burning creates harmful pollution, and is a major health and environmental concern for authorities worldwide. In fact, residential wood-burning is one of the leading sources of fine particulate matter (PM2.5) pollution in the world. This is true even in Europe, where it only covers 3% of total energy needs.
Any combustion activity creates harmful pollution. In homes, PM2.5 levels can be up to five times higher than outdoors when wood fires, candles, or incense are used. A single incense stick emits four times more particulate matter than a cigarette, while scented candles can spike PM concentrations to 15 times the World Health Organisation’s recommended limits.
Is information alone enough to limit avoidable combustion behaviour?
Unfortunately, theory says: probably not. Even when households are informed, they may not change their behaviour due to a number of psychological biases. Wood, candles or incense burning is typically associated with positive feelings and considered natural, healthy and low-polluting. These positive feelings are widespread and well-entrenched, given that humans have burned wood for heating purposes for 350,000 years. This positive affect heuristic likely distorts the perception of health and environmental risks, generates disbelief and works as an obstacle to household behaviour change, as well as support for wood smoke control policies. Low risk salience is compounded by the invisibility of pollutants and the delayed nature of their health effects. Optimism bias can also play a role, leading people to underestimate their actual exposure and risk of suffering future health consequences relative to others.
These psychological biases contribute to creating a knowledge-action gap when it comes to burning activity. How to bridge that gap? Personalised information that increases the salience of one’s own pollution exposure may be key.
The intervention: air quality micro-sensors and personalised feedback on polluting activity
In collaboration with the French Behavioural Science unit as well as the local environmental authorities, we tested the effectiveness of two interventions aimed at raising households’ awareness of health risks associated with wood-burning and other indoor pollutants.
We ran a randomised controlled trial with 281 occasional wood-burning households in France. All households received screenless air quality micro-sensors that recorded levels of fine particulate matter every five minutes throughout the winter of 2020.
We divided the sample into three groups:
- A Control group received no intervention during the study.
- An Information group received weekly leaflets on indoor pollution sources, health risks, and good practices.
- An Information + Personalised Feedback group received the same leaflets plus a personalised weekly report. This included:
- A graph showing their own PM2.5 levels over the previous week
- A comparison to other similar households in the study
Figure 1: Excerpts from generic information leaflet covers (weeks 6 and 7) received by all households in groups 2 and 3
Source: Abdel Sater, R. et al. (2024). Notes: Upper panel says “All smoke kills. Didn’t you know?”, and lower panel says “Wood-burning emits more PM2.5 than traffic”.
What we found: Personalised information works in changing behaviour – especially where it matters most
The improvement was both measurable and sustained:
- Households that only received generic information did not significantly reduce their pollution levels.
- Households that received personalised feedback reduced their PM2.5 concentrations by an average of 20%.
- Among the most polluted households, the reduction reached 40%.These households also saw a 50% reduction in the number of days that PM2.5 levels exceeded the WHO’s daily exposure guideline.
These effects began in the third week and remained stable even after the feedback stopped, highlighting the potential for long-term impact.
Figure 2: Average treatment effects on indoor daily PM2.5 levels, by week since the first message
Source: Abdel Sater, R. et al. (2024). Notes: The vertical lines represent the start (week 0) and the end (week 9) of the intervention. Confidence intervals are computed at the 95% confidence level.
Mechanisms matter: while information increased awareness, only personalised information changed perception of own air quality, reducing optimism bias
Both interventions increased awareness. Households understood the health risks of wood-burning and smoking better, and many reported intentions to reduce their use in the future.
However, only those who received personalised information updated their perception of their own indoor air quality. In other words, while general information changed minds, only personalised feedback changed salience and self-perceptions – and with that, behaviour.
While our findings show clear promise, several limitations should be considered. All participants knew the study focused on indoor air quality, which may have influenced their behaviour; if this occurred equally across groups, our estimates remain valid; if not, the true treatment effect may be under- or overestimated. Our sample was not fully representative – it was wealthier, more educated, and more concerned about pollution than the general population – potentially limiting external validity. Additionally, the personalised feedback combined two elements (a graph of household pollution levels and a peer comparison), making it difficult to isolate which aspect drove the behaviour change.
These results are particularly useful for policymakers aiming to reduce household pollution. For behaviours influenced by deep-rooted cognitive biases, even well-designed information campaigns are not enough to drive meaningful change. What works is making pollution visible and personally relevant – something that real-time feedback can achieve. With air quality sensors becoming increasingly available and affordable, short-term lending schemes could offer a cost-effective solution: behaviour change was observed as early as the third week, suggesting that even brief exposure to personalised pollution data can prompt households to adapt their habits.
There was a time, not so long ago, when cigarette smoke filled nearly every public space – from office buildings and cafés to hospitals and even patient rooms. Today, the thought of someone smoking indoors, especially around children, is almost unthinkable. But reaching that point took decades of scientific evidence, public health advocacy, and cultural shift. Our research suggests that behavioural interventions drawing on real-time information, personalised feedback, and affordable tech innovations, might bring change more swiftly.
For more details, read the full study here.
Rita Abdel Sater is an Impact Evaluation Officer at the French Development Agency (AFD), where she leads evaluations in health, social protection, and gender equality. She holds a PhD in Behavioral Economics from École normale supérieure and Sciences Po Paris. Her research work focuses on how poverty affects cognitive mechanisms and the design of evidence-based environmental public policy.
Mathieu Perona is the executive Director of the Well-Being Observatory at Cepremap, where he fosters the use of subjective well-being metrics and cognitive science to design public policy. He holds a PhD from the Paris School of Economics.
Coralie Chevallier is a cognitive scientist at École normale supérieure – PSL, where she co-leads the Evolution and Social Cognition group in the Jean-Nicod Institute. Her research examines how environmental harshness and motivational factors shape social behavior, with applications to education, health, and environmental policy. She holds a PhD in cognitive science and applies her work to real-world challenges in public decision-making.
Elise Huillery is a Full Professor of Economics at Université Paris Dauphine – PSL and Visiting Professor at ENS – PSL. Her research combines experimental and historical methods to study inequality, education, and development, with a focus on understanding and fighting inequity in education. She is affiliated with J-PAL, LIEPP, and EUDN.