Discussing the results of assessment tests with candidates--or even giving them the full report--is increasingly popular. "The trend has really been to lay it all on the table between the second and third interviews," says James Hazen. This gives candidates the chance to explain themselves, gives the interviewer a chance to address weak spots, and, if someone is hired, points out ways he or she might best be managed.
There are, by some estimates, 2,500 employment tests on the market. One of the biggest mistakes companies make is using the wrong test. A classic example is the Myers-Briggs Type Indicator, that ubiquitous test that sorts people into 16 personality categories. Myers-Briggs, a test created by a Pennsylvania woman who was fascinated by how her merry personality differed from that of her straightforward husband, has a weak record of predicting job success. Indeed, its publisher warns that "It is unethical and in many cases illegal to require job applicants to take the Indicator if the results will be used to screen out applicants."
With so many tests available, it's not a surprise that employers use tests meant for other purposes, like Myers-Briggs (which is fine, by the way, for employee development), or even design their own tests. But choosing the wrong one can mean dismissing qualified candidates and even getting sued for discrimination. Employers need to know whether a test is appropriate for hiring, what it measures, and how it's designed, along with making sure it's legal. Psychologists evaluate a psychological test by two measures, called reliability and validity. Reliability examines whether items that supposedly measure the same thing (agreeableness, say, or conscientiousness) correlate highly with one another. Validity asks, in this case, for proof that scores on tests are related to success in specific jobs. "If you go out on the Net and look at the hundreds of tests out there, a very small percentage have validity data," says Seymour Adler, a senior vice president at Aon Consulting and a teacher of organizational psychology at New York University.
Recent psychological research supports going beyond validity and reliability data. First, both for legal purposes and to ensure usefulness, make certain the test is designed for selecting--as distinct from developing or training--employees. It should be created or adapted for the workplace, not for clinical or medical diagnosis. Pre-employment tests are more predictive when they compare an individual's score against a group (they use "normative" scales, in the lexicon) instead of just presenting it on its own ("ipsative" scales). For the best results, too, employers should continue to evaluate and revalidate the tests within their companies to make sure they are still predicting top performers.
A note about testing for hourly employees. There, employers might care most about who's punctual and honest. Rock Bottom Restaurants, a 29-store chain based in Louisville, Colorado, switched three years ago from a pencil-and-paper application for its hourly employees to a test from Unicru. (Kenexa and PreVisor are two other assessment companies focusing on entry-level and hourly applicants.) For waiters, it tests for sociability and team orientation; for the back of the house, it asks applicants whether they've worked in on-their-feet jobs before; for all job candidates, it looks at integrity. Applicants in each pool--cooks, bartenders, and so on--are ranked according to their assessment scores, which gives the Rock Bottom management a good starting point. "It's not 100 percent predictive, and that's why we interview people, but it's at least an indicator," says Ted Williams, senior vice president of the brewery division at Rock Bottom. Rock Bottom's turnover for its 6,000 hourly employees has dropped by 20 percent, which Williams thinks is largely because of the system.
Step 3
In which the process starts to imitate finding World War II spies
In 1943, a pretty countryside residence in Fairfax, Virginia, was renamed Station S and repurposed as a testing site for Office of Strategic Services recruits. In an atmosphere of intense secrecy--candidates were stripped of their clothes and given military fatigues, then driven in a windowless van to Fairfax, where they would invent a cover story and fake name--the OSS studied their performance during job simulations. One test had "couriers" giving candidates a map, which they'd need to memorize in eight minutes. Other exercises included interrogating ersatz prisoners of war, devising propaganda plans, and recovering papers from an agent's room (and, aggravatingly, getting interrupted by a rifle-wielding "German" midway). The tests went on for three and a half days.
Inspired by that work-based approach, corporations such as AT&T starting using assessment centers to select executives. By the late 1950s, the candidate in the gray flannel suit was performing in-basket assessments in which he'd be graded on how he handled a set of letters, papers, tasks, and telephone calls that mimicked what he'd get on the job.
Today's work samples are essentially updates of those AT&T tests. Work samples are a proven predictor of success and can be simple to arrange. A company can design its own by laying out the criteria for a job and asking a candidate to perform a task based on those criteria. For example: "Explain how you would sell this product to Target, step by step," or "Tell me how you'd improve these lines of C++ code."
4 Number of weeks capital H Group dedicates to hiring a single consultant
At Sterling Communications, a technology PR firm in Los Gatos, California, CEO Marianne O'Connor knows her account reps have to be good at understanding technical information, at figuring out how to pitch to a media outlet, and at writing. Logical enough. So she's started giving job candidates a two-hour test before she even meets with them. It describes a client's technology, identifies a target publication and its readership, and asks a candidate to distill the salient technical points and write a pitch to the magazine. Three staffers review the pitch, and that decides whether the candidate will get an interview. "If they can't write in my business, it's not going to work," O'Connor says.
