VIEW ARTICLE

Letter to the Editor

Cluster Sampling for Disease Incidence Data. G. Hughes, L. V. Madden, and G. P. Munkvold. Phytopathology 86:132-137. Accepted for publication 6 October 1995. Copyright 1996 The American Phytopathological Society. DOI: 10.1094/Phyto-86-132.

The existence of epidemiological data presupposes some form of sampling. Guidelines for sampling procedures taken from the epidemiological literature have often been put to use by plant pathologists. Such sources make it clear that different procedures are appropriate under different circumstances. In this letter, we discuss the problem of sampling for disease incidence data (collected on the basis of scoring plants, or plant units, as either 'healthy' or 'diseased'), when the objective is to estimate mean disease incidence (the proportion of plants, or plant untis, diseased) with a prespecified degree of reliability. We show that some well known methods based on unrestricted random sampling may not be appropriate without modification. An important aspect of this problem is the spatial pattern of diseased plants. Methods of characterizing pattern that can be incorporated into formulae for sample size determination are discussed.