Small Models are Beautiful: Efficient Estimators are Even More Beautiful

There is always a conflict between fidelity to nature and the simplicity of statistical models. This conflict becomes especially acute when data are noisy or uninformative, as is often the case for problems in resource management. In such situations, the most effective model must sacrifice fidelity in order to obtain efficient estimators. This point is illustrated for a simple linear model, which is solved completely. An application from fisheries management also illustrates the virtues of simple models. It also shows how equally simple models can differ sub stantially in their estimation efficiency.