The relative merits of single and dual substituent parameter scales for the interpretation of substituent-induced chemical shifts in styrene derivatives

Abstract

The relative abilities of single and dual substituent parameter scales to reproduce substituent effects on magnetic resonance parameters in 4-substituted styrenes are compared, using both σ p and the θ scale of Edlund and Wold andthe σ 1, σ R 0 scales of Taft. Various statistical criteria demonstrate that the σ 1, σ R 0 correlations are significantly more precise, even though the θ parameters were estimated from the chemical shifts of the compounds being tested. It is further shown that there are statistically significant variations in the relative proportions of field/inductive and resonance effects for the different parameters, demonstrating that no single substituent parameter scale could reproduce the experimental data. Finally, principal-component analysis is used to demonstrate that a dual substituent parameter scale is both necessary and sufficient to account for substituent-induced changes in magnetic resonance parameters in styrene derivatives. Problems associated with the use of pattern recognition methods for classifying structures are commented upon.