The Library CO2 Sensor and the 'Fresh Air' Claim
A school installs a low-cost CO2 sensor in the library to monitor air quality. The manufacturer claims the sensor is accurate to ±50 ppm and updates every 30 seconds. Students record readings every 10 minutes for a week and find that CO2 rises from about 600 ppm at 8:30 am to over 1800 ppm by 2 pm, then falls after school. The sensor is mounted near a window that is sometimes opened. A student claims the data prove the library has 'dangerous air' and that opening the window 'fixes' the problem. The school wants to use the data to decide whether to install mechanical ventilation.
1. Evaluate the reliability and validity of the sensor data for the decision the school wants to make.
Read the worked answer
Reliability (consistency): The sensor updates every 30 seconds but readings are logged only every 10 minutes, so short peaks and troughs are missed. A single sensor gives no check on repeatability; without a second sensor or a co-located reference instrument, we cannot tell whether the pattern is real or an artefact of one device. The stated ±50 ppm accuracy is not the same as precision, and low-cost sensors often drift over weeks, especially for CO2.
Validity (does it measure what we need?): CO2 is a reasonable proxy for occupancy and ventilation, but the sensor is near a window that is sometimes opened, so readings depend on an uncontrolled variable. The 1800 ppm peak could reflect occupancy, window state, or sensor drift. The claim of 'dangerous air' is not supported: CO2 at these levels is mainly an indicator of poor ventilation, not a direct toxicity threshold, and the data do not measure other pollutants.
For the decision: the data show a consistent daily pattern, which is useful, but they are not sufficient to justify a capital decision without calibration against a reference instrument, a second sensor, and logging at shorter intervals.
2. Describe how the students could calibrate the sensor and design a fairer test of whether opening the window reduces CO2.
Read the worked answer
Calibration: Place the sensor alongside a calibrated reference CO2 meter in the same location for at least 24 hours, including a period of known fresh air (e.g. outdoors or a well-ventilated room). Compare readings at several concentrations (e.g. 400, 800, 1500 ppm) and calculate the offset and any scale error. Repeat after the study to check drift. Record temperature and humidity, because low-cost sensors are sensitive to both.
Fairer test of the window:
- Choose matched days with similar occupancy and weather, or alternate window-open and window-closed conditions within the same day.
- Fix the sensor position away from the window and direct sunlight.
- Log at 1-minute intervals so peaks are captured.
- Record occupancy counts, window state and outdoor CO2/temperature at the same time.
- Compare the rate of CO2 rise and the peak value under each condition, not just a single afternoon reading.
This turns an uncontrolled observation into a comparison with a control condition, so any difference can be attributed to the window rather than to occupancy or weather.