Sunday, November 9, 2008

Legally Defensible Competency Modeling For Use In Pre-Employment Testing

Competency modeling is a technique that is increasingly embraced as an alternative to the more traditional techniques of job analysis that have long been the foundation for developing and validating employee selection procedures. This strategy is fraught with potential peril and may expose employers to an increased risk of litigation.

It is the purpose of this article to discuss potential problems facing competency modelers. This is not to say competency modeling cannot be used effectively. Rather, it is a call to caution and serious reflection. Strategies will be presented to help protect employers.

What is a “Competency Model?” A model is a “hypothetical recreation of a complex process” or “a representation of a larger entity but on a smaller scale.” Statistical modeling is the process of creating a simpler version of a complex process by including the major or most important information necessary to make predictions. A competency model, therefore, is an artificially constructed small-scale representation of the important competencies that tend to be possessed by those people who are most successful on the job.

On a purely common-sense level, competency modeling has great appeal. It speaks a language we all seem to understand. The focus on a match between a person’s characteristics and broad aspects of the job generally just sounds “right.”

Competency modeling. as a technique for understanding the person/job interface is not new. Researchers have stated varying opinions on it for many years. It is actually similar to a “person centered” job analysis.

Guion and Highhouse (2oo6) in Essentials of Personnel Assessment and Selection, write “The notion behind the competency modeling movement is that traditional, or ‘old school,’ job analysis cannot meet the demands of the changing workplace.” Such a claim is unfounded. According to Schippmann, 2000), the problem facing competency modeling proponents is that the actual meaning of the term “competency” remains unclear, the notion behind it is that one should identify the characteristics or attributes related to exceptional performance on the job.

Competency modeling’s foundation is the mistaken assumption that traditional job analysis focuses solely on those tasks done on the job and not on the attributes required for success on the job (Sacjett and Laczom 2003).

More seriously, competency modeling lacks “methodological rigor” (Lievans, Shachez, & Corte, 2004) in spite of attempts to make it look more quantitative and objective in nature.
Another serious potential problem with competency modeling as a replacement for traditional job analysis is the fact that this author is not aware of a single case where it has had to face serious legal scrutiny and so, in the event that an employer’s selection decisions are ever challenged, it is difficult to know how solid their footing will be.

Finally, the most significant problem with competency modeling is related to how it may be used to develop selection instruments. The Federal Uniform Guidelines on Employee Selection Procedures recognize three traditional approaches to establishing the validity of pre-employment tests. These are Content Validation, Criterion Validation, and Construct Validation and each has withstood the tempests of the courtroom.

By far, the most commonly used approach to establish pre-employment test validity is content validation. Approximately 80% of all employment tests are based on this strategy (Biddle, 2007). The reason for this is that content validation is a simple and easy to follow process that links test content to critical knowledge, skills, and abilities that are required to perform critical job tasks and duties which are performed on the job. The traditional approach to conducting a job analysis is designed to facilitate this test-to-job linkage.

However, the Guidelines clearly state that a content validation strategy is NOT appropriate for creating tests that purport to measure traits such as intelligence, aptitude, personality, common sense, judgment, and leadership (UGESP, 4[b]). In many cases, the competencies that have been established as the benchmarks against which people are assessed are made based on aggregated duties or abilities that are essentially traits—the very things that are not appropriate for use in content validation processes.

The only safe way to proceed at this point would be to conduct a criterion validation study which generates statistical evidence that the test predicts job performance. It should be noted that very few employers use this approach to test validation, however. The reason for this is that it is a much more complicated and costly approach and has methodological issues of its own.
If a competency modeler is intent on using content validation, it is critical that they do the following: 1) Create competencies that are as objective and observable as possible, 2) Establish links between critical job duties and each competency, 3) Establish links between each critical KSAPC that is required at entry to each critical job duty, 4) Link actual assessment questions to KSAPCs—not to competencies as you will get the greatest level of objectivity. The result is a job analysis with the addition of an umbrella grouping of “competency” under which duties and KSAPCs are grouped.


--Jim