Testing a hypothesis about urban heat and vegetation
Two suburbs in the same city are compared. Suburb A has 42% tree canopy cover, a population density of 1,200 people/km² and a mean summer maximum temperature of 31.2°C. Suburb B has 9% canopy cover, a density of 4,800 people/km² and a mean summer maximum of 34.8°C. A student concludes: 'More trees cause lower temperatures, so planting trees will fix urban heat.' Additional data: Suburb A is 6 km from the coast; Suburb B is 18 km inland and contains a large industrial estate and a major arterial road. Both suburbs have similar income levels and housing age.
1. Evaluate the strength of the student's causal claim, identifying confounding variables and the limits of the comparison.
Read the worked answer
The claim is plausible but overstated as causal. The comparison is a two-case correlation, not a controlled experiment, and at least three confounders undermine a simple causal reading:
- Distance from the coast. Suburb A is 6 km from the sea, B is 18 km inland. Coastal sea breezes cool A independently of trees. This alone could account for part of the 3.6°C difference.
- Density and built form. B's density (4,800 vs 1,200/km²) means more impervious surface, more waste heat, less ventilation - all raise temperature regardless of canopy.
- Industrial estate and arterial road. These generate anthropogenic heat and traffic emissions, again independent of vegetation.
What the data do support: a consistent association between low canopy, high density and higher temperature, which is consistent with the urban heat island mechanism. But with n = 2, no statistical control and multiple confounders, the design cannot isolate canopy as the cause.
Strengths: similar income and housing age reduce socioeconomic confounding, and the temperature difference is substantial (3.6°C).
Verdict: the evidence supports canopy as one contributing factor, not the sole cause, and does not justify the claim that planting trees alone 'will fix' urban heat.
2. Design an improved investigation to test the canopy–temperature relationship, and state what it could and could not establish.
Read the worked answer
Improved design:
- Sample: select 30–50 sites across the city stratified by canopy cover (e.g. five bands from <10% to >50%), rather than two suburbs.
- Control for confounders: record distance to coast, impervious surface percentage, building density, traffic volume and industrial land use for each site, and include them as covariates in analysis.
- Measurement: standardised temperature logging (same instrument type, same height, same time of day, multiple days including heatwave and non-heatwave conditions) to reduce measurement error.
- Timing: repeat across seasons and years to test consistency.
- Analysis: regression of temperature on canopy cover while controlling for the covariates; this estimates the canopy effect independent of distance to coast and density.
What it could establish: the strength and direction of the association between canopy and temperature after accounting for confounders, and an approximate temperature benefit per 10% canopy gain.
What it could not establish: definitive causation (observational data still cannot rule out all unmeasured confounders), the effect of future planting (existing mature canopy differs from saplings), or whether tree planting alone 'fixes' heat - since density, built form and anthropogenic heat also matter and may require separate interventions.