Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Dr. Richard D. Goffin


The purpose of the present study was to examine the narrative feedback quality and content of comments from supervisors, peers, and subordinates in a multisource performance feedback context. Research on performance management interventions tends to focus on issues such as rater training, scale development, scale formats, and reducing test and rater bias. However, other components in performance management interventions have received little attention, including narrative feedback. Narrative feedback takes the form of written comments describing the ratee’s performance on different dimensions. The narrative feedback quality variables included favorability, specificity, goal content, and feedback length. Predictor variables of narrative feedback quality including rater familiarity, rater acquaintanceship time, and ratee position tenure were also investigated. The narrative feedback content variables included the amount of relative content, absolute content, task content and trait content.

The data were collected using a commercial multi-source feedback instrument which included numeric ratings and narrative feedback from the perspectives of the ratees’ supervisors, peers and subordinates. A sample of 200 ratees with manager or director in their title were selected. Each of the 8,967 comments were coded by four trained research assistants. The results indicated that supervisors provided the highest quality narrative feedback, peers and subordinates were comparable. Rater familiarity tended to be positively related to narrative feedback quality, and, interestingly, acquaintanceship time tended to be negatively related to narrative feedback quality, suggesting that acquaintanceship time should not be used as a proxy for familiarity. Ratee position tenure was negatively related to narrative feedback quality, however the relationship was smallest for peers suggesting the use of peer raters for longer-tenured ratees. The rating source comparisons of the narrative feedback content variables suggested that all sources used about the same amount of each content type, and that the relationships between the content variables and narrative feedback quality were comparable across rating sources. The overall results for relative, absolute, and trait feedback content suggested that they were related to positive description and included little actionable content. Task content had the largest positive relationships with narrative feedback quality, indicating that future rater training should focus on the provision of task content.