@inproceedings{Wyble2007_3,<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Abstract = {ASTM has published the practice E2214-02, Standard Practice for Specifying and Verifying the Performance of Color Measuring Instruments. The goal of this practice is to standardize the procedures and terminology used for instrument evaluation, and also to improve communication between users and manufacturers and between customers and suppliers. The experimental procedures and mathematical techniques recommended in E2214 have been analyzed and will be published in a pair of upcoming Color Research and Application articles1,2. Most of the contents of those articles has been presented to ISCC previously. Briefly, they reported the results of a long term instrument evaluation of twelve commercial spectrophotometers. These instruments were analyzed for short-, medium-, and long-term repeatability, as well as inter-instrument comparison and accuracy. The primary focus of E2214 is on a multivariate analysis rather than the traditional univariate approach (eg: average ΔE). While some of the E2214 multivariate techniques appear mathematically sound, other do not. This paper will address some mathematical shortfalls, provide a few cautions, and propose solutions that are mathematically rigorous and should result in more useful analyses for the user community.

The recommended multivariate analyses are extensions of typical hypothesis testing to three-dimensional data. One fundamental requirement for many of these tests is that  the underlying data be distributed multivariate-normally. Establishing normality can be accomplished using several procedures. Using the medium-term color difference results, we will report the results of applying some common methods of determining the normality of data. Medium-term results are hourly measurements, and should be representative of the drift user might experience during a single day of instrument use. Measurements are treated as 80 pooled colorimetric triplets. The normality evaluations  will be made on XYZ and CIELAB values as well as their differences and univariate color difference ΔE*ab.},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Address = {Kansas City, Missouri, United States},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Author = {David R. Wyble and Michael H.  Brill and Danny C.  Rich and Peter  Fortini},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Booktitle = {Inter-Society Color Council Annual Meeting},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Keywords = {astm},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Month = {April},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Organization = {ISCC},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Title = {Considerations and Shortfalls of Instrument Evaluation Using ASTM E2214},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Url = {http://www.iscc.org/ISCC2007/abstracts/T12IG1_Wyble.pdf},<BR>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;Year = {2007}