Nonparametric tests in plant disease epidemiology: characterizing disease associations

ABSTRACT Nonparametric tests are suited to many statistical applications, including experimental design, regression, and time series analysis, for example. Often these tests are thought of as alternatives to their parametric counterparts when certain assumptions about the underlying population are questionable. Although suited for this scenario, there are a number of nonparametric tests that fill unique niches in the analysis of data, for example, characterizing interspecific associations. Quantifying the degree of association between two or more pathogens or diseases at a defined spatial scale is essential to gain a thorough understanding of disease dynamics, generate testable hypothesis behind the mechanisms that cause association, and is often necessary in modeling applications. In this paper, nonparametric approaches to characterizing interspecific associations will be covered. Specifically, I will address the use of rank correlation coefficients and the development of a randomization procedure for testing the Jaccard index of association against a null model.