Analysis of the bulb weight of garlic with a focus on sample selection bias

Abstract

To perform analyses to estimate the bulb weight of garlic, many studies have used actual survey data on garlic growth, which are constructed by periodically examining growth information. Since some individual plants may wither or necrotize during growth, sample selection bias may occur if only surviving individuals are used to estimate the bulb weight of garlic without consideration of the characteristics of individuals that drop out. In this study, the Heckit model was used to estimate the final bulb weight of garlic, with correction for non-randomly selected garlic individuals. As a result of this estimation, it was shown that the presence of garlic damage had a significant effect on survival in the selection model. Some growth variables, such as the leaf sheath and the number of leaves, were directly proportional to the final bulb weight of garlic in the outcome model, whereas variables such as planting density and the number of days of rainfall were inversely proportional to the final bulb weight.