Proper Statistical Methods for Analyzing Substrate Mixtures

Wednesday, July 24, 2013: 2:05 PM
Desert Salon 13-14 (Desert Springs J.W Marriott Resort )
James Altland , USDA–ARS, MWA ATRU, Wooster, OH
Substrates are most often described as mixtures of various components in proportions that sum to 1 (for example 0.7 bark : 0.15 peat : 0.15 sand). When conducting a statistical comparison of substrates with varying component proportions, parameter estimates in a traditional regression analysis cannot be uniquely determined. This seminar will discuss a simple and practical approach to proper statistical analysis of substrate mixtures. We will also discuss how mixture analysis differs from factorial analysis in substrate experiments. Examples will be provided along with sample SAS code.