The Three-Plot Mulch Trial
A community garden volunteer wants to know whether shredded paper mulch keeps soil moister than straw mulch or bare soil. She marks three 2 m by 3 m plots along a fence. Plot 1, near a maple tree, gets shredded paper; Plot 2, in full sun, gets straw; Plot 3, partly shaded by a shed, is left bare. Each plot receives 4 L of water every Monday for four weeks. Every Friday at 3 p.m. she measures soil moisture at one spot 10 cm deep. Results: bare soil averaged 14% moisture, straw 19%, paper 22%. She concludes paper mulch is best. A neighbor points out that the plots also differ in shade and that one sample per plot may not represent the plot. The volunteer asks what would make the comparison fair and what the current data can and cannot support.
1. Identify the main confounding variable and explain how it could produce the observed pattern even if mulch type had no effect.
Read the worked answer
The main confound is shade/sunlight, because the paper-mulch plot is near the maple tree while the straw plot is in full sun and the bare plot is partly shaded. Shade reduces evaporation and can keep soil cooler, so the paper plot's 22% moisture could reflect its location rather than the paper itself. The straw plot in full sun had 19%, so straw may be doing something, but we cannot separate mulch effects from shade.
The bare plot being lowest at 14% does not rescue the conclusion: it was only partly shaded and had no mulch, so its low moisture is consistent with no mulch but also with other differences such as soil texture, slope, or root uptake. With one plot per treatment, treatment is completely confounded with location. The data are consistent with paper helping, but they do not demonstrate it.
2. Explain why one moisture reading per plot per week limits the conclusion. What sampling change would improve it?
Read the worked answer
One reading per plot per week is a single sample of a variable surface. Soil moisture can vary over a few meters because of roots, stones, slope, shade, and watering unevenness. A single spot may be unusually wet or dry, and with no repeated samples there is no way to estimate that variability or to distinguish a treatment effect from chance. Four weekly readings at the same spot are repeated in time but not independent in space, and they may all miss a wet or dry zone.
A better approach: take multiple random readings per plot each week-say 8–10 locations-at the same depth, average them, and report the range or standard deviation. Use a calibrated moisture meter or take soil samples consistently. More readings per plot reduce measurement noise; they do not fix the confounding by shade, which requires changing the layout or blocking.
3. Design a fairer four-week comparison using the same three mulch treatments. Include layout, replication, and at least two controlled variables.
Read the worked answer
Fairer design:
- Use one uniform garden area with similar soil, slope, and sunlight, or divide the area into blocks by shade level and include each mulch treatment in every block.
- Randomly assign treatments to subplots within each block, rather than letting location determine treatment. Use at least four replicated subplots per treatment per block.
- Standardize mulch depth (e.g., 5 cm), application date, and watering amount/time. Keep the bare control un-mulched but otherwise treated the same.
- Measure moisture at multiple random points per subplot at the same depth and time of day, and have the measurer not know which treatment is which where possible.
- Record temperature and rainfall, since they affect evaporation.
This design separates mulch type from shade and gives an estimate of variability. The current three-plot setup cannot do that.