Models without data can be useful, data analysis without mechanistic models often is not

Ecological phenomena are often affected by stochasticity, nonlinearities, and population structure, but without mechanistic models, inferring the effects of these factors is extremely difficult.  Meanwhile, in contrast to data, models can often be understood completely and are not necessarily tied to a particular system.  Models constructed without formal comparisons to data are therefore likely to be more useful than data analyses carried out without models.   Finally, many problems in environmental management require quantitative predictions, which is likewise difficult without models.  I will illustrate these points with examples from my own efforts to understand disease spread and complex population dynamics.