Pillbug Choice Chamber With Confounded Cues
Students place 10 pillbugs in a choice chamber with four quadrants: moist-light, moist-dark, dry-light, dry-dark. After 10 minutes they count 7 in moist-dark, 2 in dry-dark, 1 in moist-light, 0 in dry-light. The lamp over the apparatus warms the dark side by 2°C and condensation collects in the moist-dark quadrant. A student concludes that pillbugs prefer moisture and darkness equally and that the result is statistically significant because 7 of 10 is more than half. The class has a chi-square table but only one chamber and 30 minutes. The teacher asks whether the design can separate the two variables. No pillbugs were acclimated; one escaped and was returned. The chamber lid is clear plastic, and the wet paper towel touches the dark-side wall.
1. Evaluate the design and the statistical claim. Can the experiment separate moisture preference from darkness preference?
Read the worked answer
It cannot separate the two variables. The moist-dark quadrant is warmer by 2°C and has condensation, so pillbugs choosing it could be responding to moisture, darkness, temperature, or humidity. The wet paper towel touching the dark-side wall may also create a humidity gradient that is not independent of the moisture treatment. The escape and return of one pillbug adds disturbance, and the 10-minute acclimation period is short. The statistical claim is also weak: with four categories and 10 individuals, expected counts are 2.5 per quadrant, below the usual chi-square guideline of 5. A chi-square calculation would give about 11.6 with 3 degrees of freedom, which looks significant, but the low expected counts make the approximation unreliable. More importantly, significance would not rescue a confounded design. The student needs replication, longer acclimation, and independent control of light, moisture, and temperature.
2. Redesign the investigation so that moisture and light can be tested separately with the available equipment.
Read the worked answer
Run two separate choice experiments. In the first, keep light uniform and offer moist versus dry paper on opposite sides, with a barrier that prevents direct contact and with temperature monitored on both sides. In the second, keep moisture uniform and offer dark versus light covers, with the lamp positioned so it does not create a temperature difference. Randomize the side assigned to each treatment and rotate the chamber between trials to control for side bias. Acclimate pillbugs for 10 minutes before releasing them, use at least 30 individuals per trial if available, and record positions every minute for 10 minutes rather than only at the end. Treat each trial as a replicate rather than pooling all individuals as independent. Use an exact test for small samples. Report temperature and humidity for each quadrant so confounds are visible.