Multivariate Analysis: Greater Insight in Complex Systems.

Many scientific investigators measure and collect multiple response variables in an effort to understand the more complex nature of the system being studied. Multivariate statistical methods encompass the simultaneous analysis of all random variables measured on each experimental or sampling unit.  Many agronomic research systems studied are, by their very nature, multivariate; however, most analyses reported are univariate (analysis of one response at a time). 

Agronomic research examples will be used to address correlation among multiple measured variables, and the relative importance of variables in differentiation of patterns.  Topics presented will include data exploration methods, dimension reduction techniques, and partitioning methods, as well as methods for multivariate hypothesis testing.  With these methods, researchers can develop understanding of the latent variables which are not directly measurable.   The information signal identified from a data set using these methods can potentially simplify the variables needed to capture the variability required to understand a complex system.   

Many familiar univariate statistical methods have multivariate generalizations that allow researchers to take greater advantage of the richness and complexity in their data quite easily; other methods are unique to the multivariate dimension analysis.