STATISTICAL EVALUATION OF SORPTION ISOTHERM DATA

The statistical structure of 2 sets of soil sorption isotherm data was examined. Ni sorption isotherms with Puye soil differed depending upon whether 0.01 M CaSO4 or 0.0067 M CaCl2 was the background electrolyte. Isotherms with 21 replicates at each of 6 initial Ni concentrations were prepared for each background matrix. The data for each matrix could be described by both linear (R2 < 0.90) and linearized Freundlich (R2 > 0.98) isotherm equations. The variance structure and the normality of both untransformed (linear) and log-transformed (Freundlich) data sets were evaluated using a comparative table of variances and normal probability plots. Only the log-transformed data satisfied the homogeneity of variance and normality assumptions required for application of standard parametric statistics. Statistical comparisons using the log-transformed data indicated that significant differences between Cl- and SO42- isotherms could be explained on the basis of differences in free Ni2+ activity in the 2 systems. The importance of knowing the statistical structure of the data prior to making valid objective comparisons among isotherms is indicated.