Author Information

Ramon C. Littell
Sudeep Kundu

Abstract

Traditional agricultural research has been concerned largely with demonstrating that new products or new practices increase yield from plants or animals; i.e. that a change has occurred. Concepts of experimental design have been effectively employed in production-agriculture research planning to control extraneous variation and thereby reduce experimental error. Good data analysis practices have been employed to control type 1 error rate and to correctly compute errors of estimation. In recent years, increased emphasis has been placed on food safety and environmental impact of agricultural products. Studies of these issues are concerned with measuring small effects with required precision or establishing upper bounds on the effects. Statistical emphasis is on limiting the margin of error and the type 2 error rates. This paper discusses these concepts in the context of an environmental study of effect of phosphogypsum (PG) on radon flux. An experiment in progress revealed essentially no statistically significant effect of the phosphogypsum. Two statistical questions were then raised: 1) How large of an effect would have been detected in the study? and 2) How should a future study be conducted that would produce measurements of the effect with specified precision? A retrospective power analysis was performed to estimate the minimum detectable effect (MDE) in the existing study in response to question 1. In response to question 2, a new study was designed with required numbers of plots and measurements to meet precision and power objectives, using variance component estimates from existing data.

Creative Commons License


This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS
 
Apr 26th, 1:00 PM

PLANNING A SAFETY STUDY OF AN AGRICULTURAL PRODUCT: EFFECTS OF LAND APPLICATION OF PHOSPHOGYPSUM ON RADON FLUX

Traditional agricultural research has been concerned largely with demonstrating that new products or new practices increase yield from plants or animals; i.e. that a change has occurred. Concepts of experimental design have been effectively employed in production-agriculture research planning to control extraneous variation and thereby reduce experimental error. Good data analysis practices have been employed to control type 1 error rate and to correctly compute errors of estimation. In recent years, increased emphasis has been placed on food safety and environmental impact of agricultural products. Studies of these issues are concerned with measuring small effects with required precision or establishing upper bounds on the effects. Statistical emphasis is on limiting the margin of error and the type 2 error rates. This paper discusses these concepts in the context of an environmental study of effect of phosphogypsum (PG) on radon flux. An experiment in progress revealed essentially no statistically significant effect of the phosphogypsum. Two statistical questions were then raised: 1) How large of an effect would have been detected in the study? and 2) How should a future study be conducted that would produce measurements of the effect with specified precision? A retrospective power analysis was performed to estimate the minimum detectable effect (MDE) in the existing study in response to question 1. In response to question 2, a new study was designed with required numbers of plots and measurements to meet precision and power objectives, using variance component estimates from existing data.