STATISTICAL ERRORS IN X-RAY-ABSORPTION FINE-STRUCTURE DATA-ANALYSIS

General criteria for performing a rigorous statistical error determination in x-ray absorption fine-structure data analysis are formulated. The equivalence of the residual function increment approach (lambda: R({lambda}) < R(min) + chi(p,0.95)(2)) With more rigorous Monte Carlo procedures is demonstrated in the case of the analysis of the Br-2 spectrum. The possible existence of nonlinear effects in parameter space is discussed and a specific example referring to the S-C=O molecule case is illustrated. Practical suggestions are given to improve data collection procedures and to select appropriate fitting parameters in the analysis.