Rating Scales
Much of the attention in the field of performance appraisal has centered on the development and use of rating scales. I propose that this is also a source of many of the problems associated with performance appraisals. When critics such as Coens and Jenkins write about appraisal biases and the politics of performance appraisals, they are usually beginning with the assumption that this appraisal will be confined to a rating scale. The higher scores given to employees when they are being reviewed for salary increases or for promotion require some rating scale. Most other forms of performance appraisal, such as unstructured documentation, structured narration, structured documentation and intention-focused assessment, are much less likely to be biased in this manner, for they require some form of evidence to support the appraisal being made.
Rating scales are also subject to a wide variety of psychometric problems. There are leniency and severity errors, central tendency and range restriction errors, halo and horn errors, and recency errors. There are also the fundamental attribution errors, whereby we tend to attribute our own success to internal qualities and our personal failure to external variables beyond our control. Even more importantly, we reverse the attribution when asked to rate the performance of other people: “we tend to see others’ success as a product of luck and their failure as a reflection of their incompetence, laziness, or something within their control.”
This decidedly deficit-based model of human behavior produces not only distorted ratings but also inadequate or even inaccurate assessments by those doing the rating with regard to the reasons for effective or ineffective performance. While attribution errors can appear in any mode of performance appraisal, they are most likely to influence ambiguous rating scale processes, particularly if these processes are deficit oriented.
Coen and Jenkins would have us respond to these perplexing problems by abolishing performance appraisal. I would suggest instead that these problems are better addressed by moving beyond the exclusive use of rating scales when assessing the performance of another person. Ratings scales should be only one element in a multi-method performance appraisal system. When coupled with data from other sources, such as documented project outcomes and narrations, rating scale results can used in a constructive, appreciative manner.
Typically, rating scales are either tailor-made or generic. Tailor made rating scales are constructed in response to the specific interests or needs of the person being assessed. This type of rating scale is most commonly found in the assessment of high-level executives. More generic rating scales are more commonly found in the assessment of mid-level and lower-level employees. While rating scales for lower-level employees are usually directed toward specific technical skills and knowledge, as well as general items regarding ability to work with other people, the rating scales for mid-level employees often focus on both analytic and interpersonal skills.
