Principal component analysis of state level food system Indicators.

Abstract

The food system’s interconnectivity with almost every aspect of society makes accurately characterizing it very important. This same interconnectivity also makes the problem of accurately characterizing the food system very complex. Indicator sets that attempt to capture the holistic nature of the food system and are repeated across location and time to allow for comparisons and stability testing are inevitably very large. Using a large data set of state-level food system indicators collected for 1997, 2002, and 2007, this thesis explores the possibility of using Principal Component Analysis to develop summary measures for groups of indicators. The results show that it is possible to characterize the information presented by groups of individual indicators by component scores, although the process is very difficult. Through Principal Component Analysis and Partial Common Principal Component Analysis techniques, selected groups of indicators for each state over the three years are reduced in dimensionality and shown to be stable over time. This then allows for states to be compared nationally, regionally, and temporally on specific aspects of their food systems.