UNIT-CELL REFINEMENT FROM POWDER DIFFRACTION DATA - THE USE OF REGRESSION DIAGNOSTICS

We discuss the use of regression diagnostics combined with nonlinear l east-squares to refine cell parameters from powder diffraction data, p resenting a method which minimizes residuals in the experimentally-det ermined quantity (usually 2 theta(hkl) or energy, E(hkl)). Regression diagnostics, particularly deletion diagnostics, are invaluable in dete ction of outliers and influential data which could be deleterious to t he regressed results. The usual practice of simple inspection of calcu lated residuals alone often fails to detect the seriously deleterious outliers in a dataset, because bare residuals provide no information o n the leverage (sensitivity) of the datum concerned. The regression di agnostics which predict the change expected in each cell constant upon deletion of each observation (hkl reflection) are particularly valuab le in assessing the sensitivity of the calculated results to individua l reflections. A new computer program, implementing nonlinear regressi on methods and providing the diagnostic output, is described.