Analysis of Paired Design Approach Experiments Using a MIXED Model Approach.

The purpose of this paper is to describe the paired design approach for analyzing field-scale water quality studies using a MIXED model approach using Statistical Analysis System, Inc. (SAS) software. The advantage of the paired design approach is that year-to-year climate and hydrologic variability are statistically controlled and that management effects can be attributed to the imposed treatment rather than experimental unit (field or watershed) differences. Data from two experiments in Minnesota were used to illustrate the paired design approach using the PROC MIXED procedure. The purpose of one experiment was to compare changes in combined runoff and subsurface drainage water quality as a result of active management of a vegetated open-ditch for treatment of agricultural runoff compared to normal practice of no management within a vegetated open-ditch. The purpose of the second experiment was to compare changes in subsurface drainage water quality as a result of converting conventional free subsurface drainage to controlled subsurface drainage. A comparison of SAS program code and output for analysis of variance methods (ANOVA) using PROC REG and PROC GLM will be compared to the likelihood method using PROC MIXED.