Logistic Regression as a Statistical Tool to Analyse Susceptibility to Crown Gall

The information about susceptibility to Agrobacterium spp, of grapevine cultivars and fruit tree rootstocks is necessary for an integrated control of crown gall disease. Inoculation of plants in susceptibility experiments provide data where the dependent variable is either dichotomous (presence or absence of tumour) or counts (number of tumours, from one to few tumours per plant). Consequently the data follows a binomial or Poisson distribution. The statistical analysis most often used is analysis of variance (ANOVA) which assumes a normal distribution of data and homogeneity of variances. However when the dependent variable has only two values (or counts) the assumption for hypothesis testing in ANOVA are violated because the distribution of errors is not normal and variances are not homogeneous. The alternative methods are generalised linear models like logistic regression (LRM) that assume the real distribution of data (1,2).