Minimal cost multifactor experiments for agricultural research involving hard-to-change factors

Abstract

Multifactor experiments are very common in agricultural research. Randomizing run orders in multifactor experiments often witness in large number of factorwise level changes which will increase the cost and time of the experiments. Minimal cost multifactor experiments are such experiments where the cost of the experiment is minimum which can be achieved by choosing a factorial run order where the total number of factor level change is minimum as cost of the experiment is directly proportional to the number of level changes of factors. Here, a method of constructing minimal cost 2-level multifactor experiments with minimum number of factorwise level changes has been proposed. As for a same factorial combination, there may exist more than one minimally changed factorial run order, an exhaustive search was also performed to obtain all possible minimally changed run order for two level multifactorial experiments with three factors. Due to restricted randomization, adaption of these run orders may witness the effect of systematic time trend. Hence, the usual method of analysis may not be a feasible solution due to lack of randomization. Here, the analytical procedure of experiments using minimal cost multifactorial run order has also been highlighted based on a real experimental data. The work has been carried out at ICAR-Indian Agricultural Statistics Research Institute, New Delhi during 2019-20. The data from the real experiment used for explaining the analysis procedure has been collected from Climate Change Facility of ICAR-Indian Agricultural Research Institute farm, New Delhi, India based on experiments conducted during 2014-15.

Keywords

• Exhaustive search
• Hard-to-change factors
• Minimum level changes
• Minimal cost
• Multifactor experiment
• Run orders