Dissertation Abstracts

ABSTRACT OP THE THESIS

Thünian Analysis at the Macro-Scale:
A Trend-Surface Analysis of the Spatial Organization of American Agriculture

by PETER 0. MULLER, Ph.D.

Thesis Director; Professor Arthur Getis

Methodological change in geographic research during recent years has made traditional approaches to agricultural locational analyses increasingly unacceptable to a discipline seeking to broaden its theoretical foundation. A dynamic and theoretically valid framework for understanding American agricultural spatial organization is badly needed to replace the currently used static and idiographically-derived regional classifications. A viable alternative to such approaches is a modern version of Von Thünen's location theory in which the logic of the original model is extended from the scale of a single city and its umland to the macro-scale of the Megalopolis conurbation and the conterminous United States. By using a time-distance radial measurement, it is possible to show that through the advancement of transportation technology in the last century and a half the 50-hour radius of the Isolated State model has widened to encompass the entire breadth of the United States.

A sub-continental portion of the northern United States in which empirical irregularities are minimized was chosen as the study area in which to search most efficiently for macro-Thünian patterns. In the absence of data on economic rent, net income per acre was chosen as the surrogate variable. Four related per-acre variables were also studied to expand the analysis: land value, gross income, production costs and median income. County-level data were obtained from the 1964 Census of Agriculture with the exception of median income information, which was taken from the 1960 Census of Population. Trend-surface mapping, a technique of spatial analysis capable of filtering out broad regional trends from local and chance variations, was chosen as the method for studying the map and statistical properties of the distribution of each variable. Limitations in the TRENDS computer program necessitated areal sampling in which 181 of the 1376 counties in the study area were selected as control points for data input.

The map analysis showed good agreement between the expected and observed spatial patterns for each variable. The degree of fit of the trend-surfaces was measured by the coefficient of determination (r2), and these results, particularly for the central net and median income variables, were high enough to accept the hypothesis of a macro-Thünian interpretation of agricultural organization in the study area. Judgments based upon the combined map and r2 results allowed a conclusion of the empirical verification of the macro-scale model.

It is hoped that these findings can help close a major gap in the literature of economic geography, shed new light on evidence collected at other scales of Thünian Analysis, and lead to additional investigations at the continental scale in other advanced economies.