Assessment Center Method
- Introduced by
- George C. Thornton III & William C. Byham
- Year
- 1982
- Type
- Behavioral selection and development assessment procedure
- Family
- Organizational behavior / personnel assessment
What does it do?
The assessment center method evaluates people, most often candidates for managerial roles, by observing their behavior across multiple job-relevant simulations and pooling the judgments of several trained assessors.
It is a method, not a place: a standardized procedure in which candidates work through exercises such as in-baskets, role-plays, leaderless group discussions, and case analyses while assessors rate defined behavioral dimensions. George Thornton and William Byham's 1982 book consolidated the approach and its evidence, establishing assessment centers as a rigorous, behavior-based alternative to interviews and paper tests for selection and development. The method's logic is to sample behavior directly under realistic conditions and to triangulate across exercises and raters to reach defensible judgments. Arthur, Day, McNelly, and Edens's 2003 meta-analysis quantified the criterion-related validity of the underlying dimensions, sharpening understanding of what assessment centers actually measure. Professional guidelines from the International Task Force on Assessment Center Guidelines govern sound practice.
The intuition behind the method
If you want to know whether someone can lead a meeting, organize a flood of competing priorities, or handle a difficult employee, the most direct evidence is to watch them try. Interviews ask people to describe what they would do; tests measure traits and knowledge; the assessment center instead puts candidates into realistic situations and observes what they actually do. The clever part is the layering of evidence. No single exercise is decisive, so candidates face several different simulations, each tapping the skills the job demands, and no single observer's impression is trusted, so several trained assessors watch and rate. By recording specific behaviors against clearly defined dimensions, and by having different assessors observe different exercises, the method guards against the snap judgments and idiosyncratic biases that plague ordinary evaluation. The result is a structured, repeatable way to convert messy observed behavior into defensible ratings, much as a panel of judges scoring a performance from multiple angles produces a fairer verdict than any one judge alone.
History
The assessment center method traces to the multi-method, multi-assessor selection programs of the German and U.S. military and the British War Office Selection Boards in the 1940s, and to the U.S. Office of Strategic Services. AT&T's Management Progress Study, begun in 1956 under Douglas Bray, brought the approach into industry and demonstrated its predictive power for managerial careers, spawning wide corporate adoption. George Thornton and William Byham's 1982 book Assessment Centers and Managerial Performance consolidated theory, practice, and accumulated validity evidence, becoming a standard reference. The International Task Force on Assessment Center Guidelines issued and periodically revised professional standards for ethical and methodologically sound practice. Later research, culminating in meta-analyses such as Arthur, Day, McNelly, and Edens's 2003 study, scrutinized what the dimensions actually measure and confirmed their criterion-related validity, even as the construct-validity debate over dimension-versus-exercise variance continued.
When to use it
Use the assessment center method when you need a high-fidelity, behavior-based assessment of candidates for roles, especially managerial and leadership roles, where the relevant competencies are best observed in action and where the stakes justify a resource-intensive procedure. It is well suited to selection, promotion, and developmental diagnosis when job analysis yields clear behavioral dimensions and trained assessors and realistic exercises can be provided. It is less appropriate for high-volume entry-level screening where cost per candidate is prohibitive, for jobs whose key requirements are knowledge or cognitive ability better measured by tests, or where assessors cannot be properly trained, since untrained raters and poorly designed exercises strip the method of the rigor that justifies it.
Limitations and considerations
- Is expensive and time-consuming to design and run, limiting its use to high-stakes or low-volume decisions.
- Suffers a persistent construct-validity problem: variance is often dominated by exercises rather than the intended dimensions.
- Validity depends heavily on assessor training and exercise quality, which vary widely across implementations.
- Overall ratings can be influenced by cognitive ability and other constructs, complicating claims about what is uniquely assessed.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Samples job-relevant behavior directly under realistic conditions rather than relying on self-report or trait inference.
- Triangulates across multiple exercises and multiple trained assessors, reducing single-source and single-occasion bias.
- Demonstrates criterion-related and incremental validity for predicting managerial performance, as confirmed meta-analytically.
- Provides rich, behaviorally specific feedback that supports development as well as selection decisions.
Applications
- Selecting and promoting candidates for managerial and leadership positions.
- Diagnosing developmental needs and providing behaviorally grounded feedback in development centers.
- Identifying high-potential employees for succession planning and talent pipelines.
- Assessing competencies for roles where in-action behavior is the most valid evidence, such as supervision and negotiation.
- Validating and benchmarking selection systems by combining center ratings with on-the-job performance criteria.
How it works
How it works
The method starts from a job analysis that identifies the behavioral dimensions, the competencies or performance requirements, that matter for the target role, and defines each one in concrete, observable terms. Thornton and Byham stress that dimensions must be grounded in the actual demands of the job rather than chosen for convenience, and that they be defined behaviorally enough for assessors to recognize and score them. This foundation determines what the center measures; vague or job-irrelevant dimensions undermine everything downstream, so deriving a small set of clear, job-anchored dimensions is the essential first step.
Formula
D = \{d_1,\dots,d_p\} \text{ from job analysis};\quad \text{each } d \text{ behaviorally defined}Exercises are then designed to elicit the behaviors relevant to those dimensions, and the linkage is made explicit in a dimension-by-exercise matrix that specifies which dimensions are observed in which simulations. A balanced center uses several different exercise types, in-baskets, leaderless group discussions, role-plays, presentations, so that each important dimension is sampled in more than one situation. Thornton and Byham present this matrix as central to design discipline: it ensures coverage, prevents any dimension from resting on a single observation, and gives assessors a clear map of what to look for where.
Formula
E = \{e_1,\dots,e_q\},\quad M_{D\times E}:\ d_i \text{ observed in } e_j \text{ (dimension-by-exercise matrix)}During the center, trained assessors observe candidates and record specific behaviors, then translate those observations into ratings of each dimension within each exercise. Assessors are typically rotated so that no candidate is judged by only one observer and no assessor sees a candidate across every exercise, which limits halo and personal bias. Thornton and Byham emphasize assessor training, including frame-of-reference training and the disciplined separation of observation from evaluation, because the method's accuracy depends on raters reliably detecting and classifying behavior rather than forming global impressions.
Formula
r_{aij} = \text{rating by assessor } a \text{ of dimension } i \text{ in exercise } j,\quad a=1,\dots,AThe many ratings, across dimensions, exercises, and assessors, are integrated into overall assessment ratings and feedback. Traditionally this is done in a consensus discussion in which assessors share observations and reconcile their judgments, though research and Thornton and Byham's treatment also consider mechanical, statistical combination of ratings. The integration step is where the triangulated evidence is consolidated into the decisions the center exists to support, selection, promotion, or development planning, and how it is conducted materially affects the reliability and validity of the final judgments.
Formula
\text{OAR} = g\bigl(\{r_{aij}\}\bigr) \text{ via consensus or mechanical combination}Finally, the method's claims are tested against outcomes, and meta-analytic evidence refines what assessment centers measure. Arthur, Day, McNelly, and Edens collapsed the many idiosyncratic dimension labels into a small set of overriding dimensions, problem solving, influencing others, organizing and planning, communication, drive, and consideration of others, and estimated their criterion-related validity, finding meaningful prediction of performance and incremental validity over cognitive ability and personality. This validation step both justifies the method's use and addresses the long-standing construct-validity puzzle that exercise effects often dominate dimension effects in the ratings.
Formula
\bar{\rho}_{d,\text{criterion}} \approx 0.25\text{--}0.39 \text{ (dimension-level meta-analytic validity)}Common pitfalls
- Choosing dimensions for convenience rather than deriving them from a proper job analysis.
- Sampling a dimension in only one exercise, so dimension ratings cannot be separated from exercise-specific performance.
- Skimping on assessor training, allowing global impressions and halo to replace disciplined behavioral observation.
- Interpreting dimension scores as pure trait measures despite strong evidence of exercise-driven variance.
Frequently asked questions
Why use multiple exercises and multiple assessors instead of one good test?
Because the method's validity comes from triangulation. Any single exercise samples behavior in just one situation, and any single assessor brings personal biases and limited attention. Using several different exercises ensures each important dimension is observed in more than one realistic context, and rotating multiple trained assessors so that different observers watch different exercises dilutes halo and idiosyncratic error. Thornton and Byham present this layered design as the core of the method: it converts fallible individual observations into more reliable, defensible judgments, in much the same way that averaging independent measurements reduces error.
What is the construct-validity problem in assessment centers?
Studies repeatedly find that ratings cluster by exercise rather than by dimension: a candidate's scores on different dimensions within the same exercise correlate more highly than scores on the same dimension across different exercises. This 'exercise effect' suggests that centers may be measuring performance in particular situations more than stable, cross-situational dimensions, which challenges the assumption that they assess enduring competencies. The finding does not negate the method's value, since centers still predict job performance well, as Arthur, Day, McNelly, and Edens showed, but it has reshaped how practitioners interpret and report dimension scores and spurred interest in task- or role-based, rather than purely dimension-based, designs.
Are assessment centers valid predictors of job performance?
Yes, the evidence supports criterion-related validity. Arthur, Day, McNelly, and Edens's meta-analysis estimated meaningful validities for the major dimensions and found that assessment center dimensions predicted performance with incremental validity beyond cognitive ability and personality. The AT&T Management Progress Study earlier showed long-run predictive value for managerial advancement. Validity is not automatic, however: it depends on job-anchored dimensions, well-designed exercises, properly trained assessors, and sound integration of ratings. A center that neglects these design and training requirements can fall well short of the validity the method is capable of achieving.
Related methods
Sources
Thornton, G. C., III, & Byham, W. C. (1982). Assessment Centers and Managerial Performance. Academic Press.
ISBN: 9780126906202Copied.Arthur, W., Jr., Day, E. A., McNelly, T. L., & Edens, P. S. (2003). A meta-analysis of the criterion-related validity of assessment center dimensions. Personnel Psychology, 56(1), 125-153.
Open DOI referenceCopied.