Putting the Kappa Statistic to Use

Abstract

Inter-rater assessments of agreement are an essential criterion in the subjective evaluation of product quality. When assessments among raters demonstrate evidence of a lack of agreement (partial or total), there is a need to identify the source of disagreement. The objective being the reduction or mitigation of the influence different raters have on the assessment and the achievement of consistency among raters. The less influence that raters have on the assessment, the more confident one is in making critical to quality decisions. However, situations do exist in which user perceptions can be unreliable (not repeatable) and demonstrate poor correlation with engineered specifications. Quality management teams must be aware of this. When such situations exist, it is advisable to revisit the voice of the process as a reliable function of specification. Copyright © 2011 John Wiley & Sons, Ltd.