Statistical approaches to environmental monitoring: Did we teach the wrong things?

Two statistical topics that are frequently 'missed' at the level of an applied statistics course or workshop are presented here, along with examples of their use in natural resource monitoring. The first topic deals with use of correlated observations in calculations of variance estimates for a regional mean, required sample size determination, and confidence intervals. Even if correlation values are known only approximately, one can still obtain information on the effect of correlation on the aforementioned computations. The second topic concerns the use of Bayesian techniques in hypothesis testing. Although many authors have advocated its use, Bayesian analysis can be quite complex. Some Bayesian techniques are translated into graphical form that can be readily used by the reader. Both topics represent areas of statistical expertise that can be explored at a very theoretical level; however, there is enough applied material that merits translation for use by those in natural resource monitoring.