An old friend in the assessment business recently shared with me that a team of marketing consultants had advised him that any assessment, regardless of technical quality, can be a sales success in the employment selection market, if it is marketed well. This is a sad statement regarding perceptions of the value our industry provides to businesses–given the value we all should be providing.
My response to the consultants was that their statement may be true for consumer applications, e.g., dating sites, but in the employment industry, our tests must meet two criteria, or they will not sell. If they do sell, they will not sell for very long. These are the criteria that I provided:
First, an assessment (test) used in a process to make employment decisions must be fair to all people. Period. Not only is this criteria mandated by law, but there is no good reason for anyone to sell or use tests that discriminate against any group of people. It is the 21st century; everyone knows better than to use tests that are biased. Furthermore, our industry now has over 100 years of test development experience, assessment data, and research results showing us how to build, implement, and evaluate fair tests. Today, the technology to run the necessary analyses to identify test bias is available to us all. However, having served on several State and National level technical review panels over the past 15 years reviewing the technical aspects of hundreds of published assessments, I have learned that it is rare to find an unbiased test. Apparently most test publishers either do not know how to develop fair test content, or they just don’t care enough about the issue to devote sufficient resources to develop fair tests. And, there appear to be very few test publishers who know how to do a defensible evaluation of the bias in their own tests. The fairness hurdle for a test must be met, or the buyers of the tests will most likely get sued for discriminatory employment practices and lose in court. On the other hand, without the benefit of fair and valid assessment data, employment decisions are biased by design. For example, interviews are based on human judgments. Human judgments regarding the qualities of other people are obviously biased by pre-existing social schema used by our brains to automatically process basic information about whoever is in our presence. Our schemata are in place for our own protection and survival; they are nearly impossible to extinguish. The greatest value of the fairness criteria for tests, in my opinion, is that a good assessment in an employment decision process will greatly reduce the inherent bias of human judgments.
Second, assessments (tests) must be valid; the results must predict criteria of importance to the business using the assessment for making employment decisions; including risk factors such as turnover, absenteeism, unsafe acts, theft, error rates, destruction of property, illegal drug use, and harassment of other employees; and value added factors such as job performance, customer satisfaction, production, process improvements, training/teaching others, organizational citizenship behaviors, and leadership. This is not a hurdle by law, unless the test is unfair, but it is a business requirement of buyers. The value of an employment test to a business is based completely on the degree to which the assessment predicts criteria of importance to the business, (i.e., predictive validity). Test data that do not predict business outcomes cannot improve business outcomes, and will have no return for the buyer on the investment made in the test use. In fact, given that tests with high predictive validity are an available alternative, the use of a test that does not predict business outcomes comes at the cost of the test use, plus the much greater relative cost of the lost opportunities for the business to improve its outcomes (e.g., profitability, competitive advantage) through risk reduction and value added factors. Again, based on reviews of their technical data and validation studies, very few assessments have ever been developed with the specific purpose of improving target business outcomes for the users of the tests. In fact, the history of the testing industry is filled with examples of tests sold for making employment decisions, even though the tests in question were designed for other purposes (e.g., clinical diagnosis, theory testing, parlor games, amusement of children). And again, there is no reason for any business to continue to use tests for employment decisions that do not have a positive impact on their target business outcomes, when tests are now available for use that were specifically designed to improve business outcomes and have been validated to do so.
Thanks for reading. I hope this is useful information for your work and/or business success. There will always be those who do not see the value of data for decision making–but that does not make the use of data wrong–it just increases the need to use fair and valid data in our decision making. Please remember that fair and valid assessments for use in employment decisions do exist and will help a business succeed–as an example, we spent the last five years of our own lives developing and validating the PointLeader assessments for this purpose and have seen the impact of their use on our clients’ business outcomes. I look forward to hearing from you…