BEHAVIORALLY ANCHORED RATING SCALES (BARS)
The Behaviorally Anchored Rating Scales (BARS) method combines elements of the traditional rating scales and critical incidents methods. Using BARS, job behaviors from critical incidents — effective and ineffective behaviors are described more objectively. The method employs individuals who are familiar with a particular job to identify its major components. They then rank and validate specific behaviors for each of the components.
How to construct BARS?
Developing a BAR follows a general format which combines techniques employed in the critical incident method and weighted checklist rating scales. Emphasis is pinpointed on pooling the thinking of people who will use the scales as both evaluators and evaluees.
Collect Critical incidents:
People with knowledge of the job to be probed such as job holders and supervisors describe specific examples of effective and ineffective behavior related to job performance.
Identify Performance Dimensions:
The people assigned the task of developing the instrument cluster the incidents into a small set of key performance dimensions. Most performances fall generally between five and ten dimensions. Examples of performance dimensions include technical competence, relationships with customer handling or paperwork, and meeting day-to-day deadlines. While developing varying levels of performance for each dimension (anchors), specific examples of behavior should be used that is rated in terms of good, average or below average performance later.
Reclassification of Incidents:
Another group of participations who are knowledgeable about the job is instructed to retranslate or reclassify the critical incidents generated (in Step II) previously. They are given the definition of job dimension and told to assign each critical incident to the dimension that it best describes. At these stage incidents for which there is not 75% agreements are discarded as being too subjective.
Assigning Scale values to the Incidents:
Each incident is the rated on a one-to-nine scale with respect to how well it represents performance on the appropriate dimension. A rating of one represents in effective performance; the top scale value indicates very effective performance. The second group of participants usually assigns the scale values. Means and standard deviations are then calculated for the scale values assigned to each incident. Typically incidents that have standard deviation of 1.50 or less (on a 7-point scale) are retained.
Producing the Final Instrument:
About six or seven incidents for each performance dimension â€“ all having met both the retranslation and standard deviation criteria â€“ will be used as behavior anchors. The final BARS instrument consists of a series of vertical scales (one for each dimension) anchored (or measured) by the final incidents. Each incident is positioned on the scale according to its mean value