Discrete Time Survival Analysis Applied to Experimental Data
Time-to-event outcomes are common in agricultural sciences. For example, how long it takes until flowering is one of the critical research questions for plant scientists. Despite the popularity of survival analysis in medical studies for past decades, application in agricultural sciences has been less discussed. The main strength of this method is their ability to handle missing data over time, namely, right-censored data. Even well-designed experimental data may encounter drop-out, which can be ignored in methods such as analysis of variance (t-test) for comparing survival times for two (or more) groups. Survival models also tend to have greater statistical power to detect a significant treatment effect than methods for binary response such as logistic regression. The goal of this study is to review basic concepts of survival analysis, importantly discuss the benefit of this method when it comes to agricultural research applications. Examples vary such as time-to-damage of seed quality, time-to-clean of water, time-to-death of animals or plants. This paper demonstrates the advantage of survival analysis on experimental data through both simulation and real data studies.