When you consider investing in an employment testing system to help make hiring and advancement decisions, it's important to recognize that not all tests are created equal.

Here's what you need to know to make the best choice.

The Standards for Educational and Psychological Testing describe many types of information that testing companies should be able to provide to test users. Although The Standards do describe employers' best practices, it should be stressed that they are not laws. In specific situations, legal and practical considerations will take precedence.

A testing company should provide test users or potential test users sufficient information so that, before they adopt a test, they can evaluate the suitability of the test. In particular, the test user should be able to learn the following:

Kurt Ballard
Sr. VP Global Sales & Chief Marketing Officer
Qwiz

• the purpose of the test

• what the test is intended to measure

• how the test is administered

• whether the test has a stringent time limit

• the average time required for a test session

• how many questions are on the test

• what type of questions are on the test

• the intended test takers

• the score(s) generated by the test

• how the score(s) should be interpreted

• common misuses of the test, if any

The testing company should explain the proper interpretation of test scores. This is especially important if, instead of being reported in raw form (for example, number of correct answers), the scores are converted to a special scale (examples: percentiles, stanines, a scale of 0 to 10, or a scale of minus 3 to plus 3).

If the test includes overall or composite scores that combine the results from several subscales, the testing company should explain how these composite scores are calculated, and how they should be interpreted. For example, do all subscales receive equal weight? If particular misinterpretations of the scores are likely, test users should be forewarned. If the testing company scores the tests and the scoring system automatically assigns labels, categories, or descriptions to test takers, such as "unqualified" or "advanced programmer," the testing company should be able to provide a rationale for these assignments.

Support for Claims
If the testing company claims that use of the test will result in a specific outcome, such as reduced turnover or increased productivity, they should also report the basis for that claim, including supporting evidence. The evidence should be detailed enough that the test user can judge whether the same results were likely to be obtained in the test user's own setting with the test user's own candidates and/or employees.

If the testing company claims that test scores can predict job performance (that is, that there is a correlation between test scores and a job performance measure), they should also report the basis for that claim, including supporting evidence. The evidence should be detailed enough that the test user can judge whether the same results were likely to be obtained in the test user's own setting with the test user's own candidates and/or employees, and using the test user's own preferred performance measures.

Accuracy of Measurement
There are two common indicators of a test's accuracy of measurement: its reliability and its Standard Error of Measure (SEM). The test company should be able to provide this information on every test. If a test generates scores on more than one scale, the reliability and SEM should be available for every scale. The reliability is reported as a decimal number between zero and one, with higher numbers being better. The reliability allows a comparison of one test with another, even though the two tests may be of different types or different lengths.

The SEM gives the likely margin of error in an individual's test score, in terms of the scale in which the score is reported. The SEM makes it possible to define a "confidence zone" around an individual's score on a test. However, the SEMs of different tests usually cannot be compared because SEM is influenced by factors such as test length.
 

What Test Users Should Do To Ensure the Validity of Their Tests
The Standards include a number of guidelines that, if followed, will maximize the validity of the test user's testing process. This validity concerns the entire process, including the test itself, the test administration procedures, the interpretation of the test results, and the decisions that are based on the test results.

Test users should note that, by increasing the validity of their testing program, they are simultaneously increasing its effectiveness and its legal defensibility. Further, the test users are legally responsible for the validity of their own testing process.

Before adopting a test, a test user should be comfortable that the test is appropriate for its intended use. In particular, the test should:

• provide the information necessary to support the test user's employment decisions

• be suitable for the job for which it would be used

• be suitable for the test takers who will take it

• generate scores that are relevant and comprehensible

• have appropriate reliability

• be convenient to administer and score

If the population of test takers contains a large number of individuals who are members of protected classes (for example, minorities, women, and people with disabilities), the test user should investigate whether the test would be suitable for those individuals.

Beware of test score misinterpretations. There are many ways that a test score can be misinterpreted, but they can all be avoided if the test user knows enough about how the test is constructed and what the scores mean. One common source of misinterpretation is scale names that are ambiguous or misleading. For example, if a candidate scores high on a scale titled "Assertiveness," does that mean she takes initiative, or does it mean she is argumentative? Does it mean she takes the lead in social situations, or does it mean she defends her ideas energetically? If the test user has investigated the intent and the content of the Assertiveness scale, he or she will be able to interpret this score more accurately.

Depending on the subject matter, tests may become out of date as jobs and technology change. The test may no longer be applicable to how the job is performed, or it may contain references that reveal its age (for example, math problems involving 10-cent cups of coffee). In the first case, the test is no longer job-relevant; in the second case the test may remain relevant but the obsolete references may be distracting to test takers. The test user is responsible for using only tests that are job relevant and up to date.

About the Author
Kurt Ballard brings more than seventeen years of innovative sales, marketing, and leadership capabilities to the Qwiz team as Sr. Vice President & Chief Marketing Officer. Most recently he led the development of the sales and marketing methodology for Exult's Gunn Research Group, working with Fortune 500 companies to implement best practice-based improvements to human resource and finance processes. Mr. Ballard has held leadership positions in the staffing industry with Norrell and Staffmark in strategic sales, process and facility outsourcing, and client relations.

About the Company
Qwiz provides competency assessment solutions serving: corporate human resources, the staffing industry, Call Center, Legal, Finance, Healthcare and IT markets. With over fifteen years of proven success, Qwiz offers competency assessment solutions for office professionals in five different languages: English, Spanish, French, German, and Dutch, as well as a complete line of training products in English.