Comparison of statistical methods for separating iron deficient from sufficient soils

Abstract

Two correlation data sets for the DTPA soil test and one for the NH₄HCO₃‐DTPA soil test were used to evaluate various soil test correlation methods. Visual, graphical, Mitscherlich, Chi‐square, optimum predictive value, and two‐ and three‐class analysis of variance methods were compared. Critical levels were identified for each method and transition or marginal zones were determined for those methods which facilitate their identification. Critical levels ranged from 3.5 to 5.0 ppm NH₄HCO₃‐DTPA extractable iron and from 3.5 to 4.8 ppm DTPA extractable iron. Generally, transitional zones were very similar for all methods with both soil tests. Based on these results the Chi‐square method, supported by visual and/or graphical methods, is recommended because of their simplicity and ability to identify transition or marginal zones.

Key words:

• Soil testing
• Soil test correlation
• Chi‐square
• Analysis of variance
• DTPA
• NH₄HCO₃‐DTPA