Content Analysis of Evaluation Comments as Evidence for Human Performance Appraisal by Text Mining

Abstract

Evaluation comments for job performance were mostly reported as evidence in human performance appraisal. If the comments reflected the contents of the behavioral anchors that indicate the desired job performance, it would be an endorsement that the human performance appraisal program were well operated. The coding rules were designed to extract the content of the behavioral anchors and critical word usage for performance levels. Subsequently, text mining analysis was conducted. The elements of behavioral anchors were extracted in about 20% of the cases. Positive words were more likely to appear in the high-rated group, especially when unaccompanied by negative forms. Word usage referring to the purpose of performance and expressing personal traits with intensifiers was more prevalent in the high-rated group. Word usage referring to 'unachieved' and the use of politeness expressions were more often observed in the low-rated group.