Testing a Stream pH Claim with a Confounded Design
Students test a stream above and below a road-salt storage pile. Before a storm, pH at three upstream sites is 7.2, 7.1, and 7.3; at three downstream sites it is 7.0, 6.9, and 7.1. After 2 cm of rain, upstream pH is 6.6, 6.5, and 6.7; downstream is 5.9, 5.8, and 6.0. They conclude that the salt pile acidifies the stream and that rain is the cause. They used a handheld pH meter calibrated once at the start with pH 7 buffer. No rain pH was measured. The downstream site is 30 m below a culvert that drains a parking lot. The students sampled on one day and did not duplicate any measurement.
1. Evaluate the conclusion. Which parts are supported, which are not, and what confounders matter?
Read the worked answer
The data show two things. First, pH fell at both upstream and downstream sites after rain, so a regional rain-related effect is plausible. Second, the upstream-downstream difference increased from about 0.2 pH units before rain to about 0.7 pH units after rain, which is consistent with an additional downstream source. Because pH is logarithmic, a 0.7-unit difference is about five times the hydrogen-ion concentration. But the conclusion that the salt pile is the cause is not established. The culvert drains a parking lot, which can deliver its own runoff, salt, or acidic water. No rain pH was measured, so the students cannot separate rain acidity from road-salt effects. A single calibration at pH 7 does not establish accuracy at pH 6, where electrode drift and temperature effects are larger. With three non-duplicated samples and one storm, the result is suggestive but not conclusive.
2. Design a stronger study. Specify the controls, measurements, and calibration checks you would require, and explain how each addresses a weakness in the original design.
Read the worked answer
A stronger design would sample upstream, the culvert outfall, and downstream before, during, and after several storms, plus a nearby control stream not affected by the salt pile. Measure pH, specific conductance, temperature, and discharge at each site, because conductance helps distinguish salt loading from general dilution. Collect rain directly and sample the parking-lot runoff at the culvert. Calibrate the meter with at least pH 4 and pH 7 buffers before and after each sampling trip, and record the calibration slope; this tests accuracy in the pH range actually observed and detects drift. Duplicate one sample per site and analyze duplicates to estimate measurement precision. Repeat over multiple storms so that a single unusual event does not drive the conclusion. This design separates rain effect (control stream and rain pH), salt-pile effect (upstream versus downstream after accounting for culvert input), and measurement artifact (calibration and duplicates).