Comparing test cultivars using reliability functions of test-check differences from on-farm trials.

An approach to selection is proposed that is based on the probabilities that a test cultivar outperforms a check by more than an amount d in a future environment. The function that gives these probabilities for all possible values of d is called the reliability function. When d=0, the value of the reliability function is the probability that the test cultivar outperforms the check. The method is illustrated using data from on-farm maize (Zea mays L.) strip test trials grown cooperatively by Pioneer Hi-bred International and farmers. Results indicate that reliability functions are useful for evaluating how test cultivars perform relative to a check across a range of environments since the location, slope and shape of the reliability function may be used to describe a test cultivar's performance, similarity to the check and stability, and identify environments where the test cultivar has performance problems.