The Balcony Sensor and the Asthma Claim
A sixth-form student in Msida wants to prove that traffic pollution is causing asthma among students at her school. She mounts a €120 low-cost PM2.5 sensor on her balcony, 2.5 m above ground and 15 m from a busy road. She records data for three days in February, including a weekend. She compares her readings with the official background station 400 m away and finds her sensor averages 55 µg/m³ while the official station averages 12 µg/m³. The sensor was not calibrated and humidity was above 80% on two days. Her data show a sharp rise between 07:30 and 08:30 on school days. She writes: ‘Traffic causes asthma; the school must close the road.’ She has no asthma data and does not know how many students have the condition.
1. Evaluate the extent to which the student’s evidence supports her conclusion.
Read the worked answer
The evidence is suggestive but does not support the causal claim. The student has a plausible hypothesis: a school-day morning spike near a road is consistent with traffic. But three days, one balcony and one uncalibrated sensor cannot establish either exposure or health effect.
The 55 µg/m³ average is not comparable with the official 12 µg/m³ background station because the instruments, siting and purposes differ. Humidity above 80% can bias some low-cost optical sensors. The absence of asthma data is decisive: even a perfect pollution measurement would not show that traffic causes asthma among students. The claim ‘the school must close the road’ adds a policy conclusion that the design cannot test.
This is a correlation-to-causation error, plus a measurement-validity problem. A fair conclusion would be: ‘There are preliminary indications of a morning pollution peak near the road; further calibrated monitoring and health data are needed before any causal or policy claim.’
2. Identify the main confounding variables and sources of error, and explain how each could affect the result.
Read the worked answer
Confounders include weather (wind, humidity, temperature inversions), traffic volume and composition, the weekend/weekday difference, and the sensor’s height and distance from the road.
Sources of error include lack of calibration, sensor drift, humidity interference, incorrect inlet placement, and comparison with a station that measures a different scale. The official station may be a regional background site, so the difference may reflect location rather than instrument error. The student’s balcony may be shielded or exposed in ways that do not represent the school playground. The short period may coincide with unusual events.
Each problem can push the reading up or down; without calibration and a control site, the direction and size of bias are unknown, so the 55 versus 12 comparison cannot be treated as a real exposure difference.
3. Design a stronger study that could fairly test the claim, including ethical considerations.
Read the worked answer
A stronger design would use calibrated monitors at three sites: the school playground, the roadside, and a background location, run for at least eight weeks across seasons, with simultaneous temperature and humidity logging. The student should record traffic counts and meteorology, and rotate or co-locate sensors to check reliability.
For health, she would need consent, parental permission and anonymised data-for example, school absence or nurse visits-not individual diagnoses. She should pre-register the hypothesis and analysis, report uncertainty, and avoid identifying students. Ethics also require not alarming participants and not implying that closing a road is the only remedy. The study should be reviewed by a teacher or ethics committee and should state that it tests association, not proof of causation.