What is an Assessment Center?
The Assessment Center Method, Applications, and Technologies
SECTION 2: Validity and Fairness
- Section 1: How an Assessment Center Works
- Section 2: Validity and Fairness
- Section 3: Adoption of the Assessment Center Method Outside the United States
- Section 4: Selection and Placement of Empowered Personnel
- Section 5: New Simulations, Tests, and Methods
The assessment center method, in its modern form, came into existence as a result of the AT&T Management Progress Study (Bray, Campbell, & Grant, 1974). In this study, which began in the late 1950s, individuals entering management positions in Bell Telephone operating companies were assessed and, from then on, their careers were followed. The study was unusual in that it was pure research. Neither the individuals assessed nor their bosses were given information about their performance in the center. Nor was this information in any way allowed to affect participants’ careers. Participants were assessed soon after they entered management as new college recruits or after they were promoted from the ranks.
Not only did researchers follow participant advancement during the ensuing years, but a second assessment also was conducted eight years after the first (Howard & Bray, 1988). The criterion used was advancement to the fourth level of management in a seven-level hierarchy. The eight-year prediction is more valid—an expected finding since most individuals would have begun to consolidate their management skills after eight years in management. Yet the original assessment ratings were still valid—even after 20 years.
Thornton and Byham (1982) reviewed 29 studies of the validity of assessment center methodology. The authors found more support for the assessment center method than for other selection methodologies, while lamenting the fact that most of the studies were done by a few large organizations (AT&T, GE, IBM, SOHIO, and Sears).
In 1985 Thornton and his associates at Colorado State University processed 220 validity coefficients from 50 studies using a statistical approach called meta-analysis. They estimated the method’s validity at .37 (Gaugler, Rosenthal, Thornton, & Bentson, 1985). Working independently of Thornton, Wayne Cascio of the University of Colorado arrived at the same figure (.37) in studying the validity of first-level assessment centers in an operating company of the Bell System. Cascio’s main interest, however, was in measuring the "bottom-line impact" of promotion decisions based on assessment center information versus decisions based on criteria extracted from other methods (Cascio & Ramos, 1984).
To determine the dollar impact of assessment centers, Cascio needed more than validity information; he needed cost data (fully loaded costs of the assessment process), plus job performance data expressed in dollars. Over a four-year period he developed a simple methodology for expressing in dollar terms the job performance levels of managers. Using information provided by more than 700 line managers, Cascio combined data on the validity and cost of the assessment center with the dollar-valued job performance of first-level managers. With this data, he produced an estimate of the organization’s net gain in dollars resulting from the use of assessment center information in the promotion process. Over a four-year period, the gain to the company in terms of the improved job performance of new managers was estimated at $13.4 million, or approximately $2,700 each year for each of the 1,100 people promoted in first-level management jobs.
Compared to other selection methodologies, the assessment center method generally is seen as more fair and objective in terms of gender, race, and age than other methodologies. Some differential performance has been found, but this usually is the result of differential applicant populations. The method is uniquely fair because of its emphasis on actual behavior rather than psychological constructs.
There is consistent research showing that assessment centers are unbiased in their predictions of future performance. These studies considered a candidate’s age, race, and gender and found that predictions by assessment center methodology are equally valid for all candidates. (See Thornton & Byham, 1982, for a complete discussion of these issues.)
Federal courts have viewed assessment centers as valid and fair. Indeed, they often have mandated assessment centers to overcome selection problems stemming from the use of paper-and-pencil and other selection instruments.