Analysis of Plant Protection Studies with Excess Zeros Using Zero-Inflated and Negative Binomial Hurdle Models
In this study, the analysis of data with many zeros for plant protection area was carried out by using the models of Poisson Regression (PR), negative binomial (NB) regression, zero-inflated Poisson (ZIP) regression, zeroinflated negative binomial (ZINB) regression, and negative binomial hurdle (NBH) model. As zero-inflated observations are too much in the studies, done in the plant protection area; models considering zero-inflated observations are frequently required. Mites (Acari: Tetranychidae; Stigmaeidae), the basic material of this study, can reach to quite high amounts under convenient temperature (18-32°C temperature). The fact that deviance obtained from PR model together with Pearson Chi-square and deviance goodness of statistics came about quite higher than the value of (1) represented that there was an overdispersion in data set. In the selection of appropriate regression model, Akaiki information criteria and Bayesian information criteria were used. At the end of these information criteria, ZINB regression was chosen as the best model. In ZINB model, the effects of Zetzellia mali, temperature, and periods were significant on the total P. ulmi number (p<0.01), while applying insecticide was insignificant (p>0.05).
Key Words: Zero-inflated data; overdispersion; Negative binomial Hurdle model; Zero-inflated models.