Abstract
Regression analysis has become a very popular method for analyzing existing plant and research data. The existence of correlations among the independent variables is recognized as a major pitfall in the proper interpretation of such data. This paper outlines the errors in interpretation that may result when specific degrees of correlation exist. The situation is examined for a two-variable case, fitting first and second degree models to data from sources of varying complexity. A computer was used to generate and analyze the data.
Additional information
Notes on contributors
John D. Hinchen
Mr. Hinchen is a Senior Research Specialist in Monsanto's Organic Division. He is a Fellow of ASQC.