A manufacturer tested two sensors over six months. All units of Sensor A were tested during the first three months, and all units of Sensor B during the last three. The calibration machine gradually drifted upward over the six months, but researchers attributed the higher readings for Sensor B entirely to sensor design.