2009 Youden Memorial Address: Perspectives on Prediction Variance and Bias in Developing, Assessing, and Comparing Experimental Designs 37th W. J. Youden Memorial Address
Abstract: The vast majority of methods used in practice to develop, assess, and compare experimental designs focus on their variance properties. Because models only approximate true unknown relationships, they are subject to bias errors as well as variance errors. Literature from last 50 years has developed methods that consider bias (and variance) properties of designs. A resurgence in work/publications on this topic has occurred in the last several years. Still, methods that consider bias properties are seldom used in practice. We need to routinely use methods that consider variance and bias properties to develop, assess, & compare designs.
Keywords: Prediction Variance - Prediction Bias