Biased Experimental Fineness and Maturity Results

Abstract

In an earlier paper, we developed models and performed computer simulations to understand the variability in coefficients of determination (R2) between fineness and maturity, micronaire and fineness, and micronaire and maturity of cotton. We subsequently concentrated on derivation and testing of several diagnostic models to enhance the R2 and provide information about the analytical quality (accuracy) of the results. We then introduced modeling of biased fineness and maturity results. Error functions were derived based on micronaire values, specifically, Lord's micronaire model. This paper demonstrates testing of a key diagnostic model on two different sample sets of 21 cottons. The results from one sample set — analyzed on the fineness and maturity tester — fit the model. Results from the other sample set — analyzed on both the advanced fiber information system (AFIS) A-2 and AFISPRO — demonstrate a lack of fit to the diagnostic model. This lack of fit is due to bias in the AFIS fineness and maturity measurements compared to the more traditional Lord's micronaire model. As a consequence of the bias, the dynamic range of the AFIS raw data for both fineness and maturity is very narrow. Results are confirmed by image analysis.