Design an intervention that distinguishes competing mechanisms

Frame the environmental question

Two mechanisms can reproduce the same baseline observation. Matching one endpoint therefore does not uniquely identify the process that generated it. A useful investigation seeks conditions under which the candidate explanations predict different outcomes, while preserving controls that keep the comparison interpretable.

Build the measurement or system model

The original biomass packet contrasts a nutrient-limited candidate with a light-limited candidate. Both predict baseline 10. Under a nutrient reduction, their predictions are 6 and 10; under shading, 10 and 6. The supplied observed responses are 6 and 9 respectively. These are invented conditional predictions, not a real nutrient-addition experiment or a universal ecological response law.

Work at the right scale

Residual means observed minus predicted. Squaring residuals can summarize discrepancy on one common response scale, but a smaller sum is not automatically a probability of truth. Models can have different flexibility, measurement uncertainty or omitted mechanisms. Choosing and evaluating on the same packet can also exaggerate apparent predictive success.

Challenge the inference

The nutrient candidate fits the first intervention and is closer in the shading packet, but a combined-limitation model or unequal manipulation intensity might explain the residual. Plan a factorial comparison that varies both drivers with controls, independent units and measured manipulation strength. Evidence for a mechanism should connect exposure, intermediate process and response rather than stop at a biomass difference.

Design a defensible comparison

Environmental causal assessment integrates several lines of evidence and considers alternatives. A chemical concentration, a biological association and an intervention may support different parts of a causal chain. State which evidence was used to propose the model and reserve genuinely new conditions for evaluating a prediction. A public worked packet remains practice even after its label is changed.

Worked original packet

At nutrient reduction, observed 6 matches the nutrient candidate 6; the light candidate 10 has residual−4 and squared residual 16. Under shade, observed 9 gives residual−1 for nutrient prediction 10 and+3 for light prediction 6. Baseline agreement alone would not distinguish them.

Investigation

Compare all three packets under both candidates. Retain residuals, propose a third mechanism and specify a new intervention whose predicted outcomes would distinguish it before seeing data.

Changed-case transfer

A new independent packet has observed 7 and candidate predictions 8 and 5. Calculate residuals and squared residuals; state what is still missing before causal selection.

Feedback

Residuals−1 and+2; squares 1 and 4. The first prediction is closer on this point, but uncertainty, replicated controls, model-selection history and alternate mechanisms remain necessary.

Sources and evidence boundary

Original instructional prose, values and exercises. Factual context checked 30 October 2026: primary teaching or agency source. These are synthetic computational comparisons, not measured field data, verified mastery or authorization for physical intervention. Retain your prediction, assumptions and source of each observation.