Evaluation of self-diffusion data using weighted means statistics

Several CALPHAD-based assessments of the self-diffusion mobility of Ni in pure Ni and Al in pure Al have been published. In an effort to determine which assessment best represents the experimental data, weighted means statistics are used to determine consensus estimators for the data. The advantage of this statistical approach, which uses a Gaussian distribution to represent the experimental error associated with each study, is that the relative error between studies is evaluated and that all available information is used to determine the best estimates of the parameters common to all of the included studies. The results of these statistical analyses are compared with previous CALPHAD-based diffusion mobility assessments. The comparisons show that inclusion of more of the available experimental data does not necessarily yield a more statistically accurate representation of the data. More generally, this analysis demonstrates how weighted mean statistics can be used to evaluate data that is a function of multiple parameters.