Once a week I get called by someone, usually an administrative assistant, asking if I have a personality test. When I try to get more information, she usually tells me her boss wants the test and she just wants the price.
& l- f2 g% K* z9 M( ?, d
I tell her, "Sorry. I sell solutions, not tests."
% I' ~3 S, y: m& y f
She hangs up absolutely convinced I am a jerk.
* s5 |0 M; w% n( x' FPutting aside the fact she's probably right, let's examine her request.
Y$ m h+ \; F b, M8 M% k2 a
When bosses want to buy a hiring test, it's usually because they want to reduce hiring mistakes. With typical don't-know-what-they-don't-know aplomb, they send their hired help on a fishing expedition to find a test to "fix" what they don't know. Then things get problematic.
O) }8 l8 K# \; j
Why is it so important to make a distinction between selling tests and selling solutions? Because selling tests benefits the seller; selling solutions benefits the buyer.
; y @2 I) R$ m; i* B K1 b/ ~Hiding in Plain SightTesting falls along a continuum:
: }; i5 U- N ?$ h# H: h
- Type 1 (chance): Hire anyone who applies and fire people who fail the job-test (a popular variant is to aggressively recruit based on personal opinion and cull the employee base 10% annually).
- Type 2 (also, chance): Ask questions, give tests, compare applicants to each other, and disqualify applicants based on choosing the wrong animal, tree, or legume they would most like to be.
- Type 3 (about 25% accurate): Study the job thoroughly, ask hard-to-fake-questions, and systematically compare results to job requirements.
- Type 4 (about 50% accurate): Study the job thoroughly, give a solid, legitimate, and validated test that directly applies to job performance and systematically score results.
- Type 5 (about 75% to 80% accurate): Combine several validated methods that cover the entire job domain.
Buy-a-test people are like patients who walk into a doctor's office expecting to be treated but not saying where, why, or how it hurts. It's hard to recommend a solution when there's no clear understanding of what needs fixing.
0 U, c& I$ c) R$ z
Hiring and Performance Theory Best practices start with having a sound theory of performance behind the test. This is psycho-babble for: it's always a good idea to test for something that makes a difference between high and low performance. Basically, this means if you use a test like the DISC or MBTI to make better hiring decisions, you better have hard data to support your decision (especially since reputable vendors of both tests recommend against using the DISC or MBTI for hiring).
, }; P' D( m {% t
Putting people in little style boxes may be fun stuff, but it takes homework to prove XYZ style actually predicts job performance (it usually doesn't).
5 t& y- p$ m" \) m" E' TAre theories behind the DISC, MBTI, and MMPI based on job performance? No. Read their manuals. These tests describe personal differences, not job skills. Don't make hiring decisions based on any test score unless you could produce data proving it predicts performance. And if the test authors advise against using their test as a hiring tool, then it's probably a good idea to follow their recommendations.
5 l& e1 _2 J. X C; y+ u5 LWhat if the authors suggest only using the test to
help make an informed hiring decision?
Huh? What part of "testing" is unclear? If a score is used to "help" make a hiring decision, then they are being used as a test.
e0 O$ @3 ?' |( M4 q0 b" y
Next StepOk, suppose the test is backed with a sound job performance theory…what's next? Face validity, reliability, and internal construction.
: s% F' z! [8 Z' c! L& {: O+ oFace validity means the test questions and format should resemble a legitimate test. That is, the test should never include invasive questions about the subject's personal lifestyle choices (e.g., do you belong to a spouse-swap club?) nor questions that have little to do with job performance (such as asking the person to stand on their head and sing The Star-Spangled Banner in a public place). Not only is a lack of face validity bad test practice, it makes the organization look silly and unprofessional.
+ [3 u/ {/ B, l# y6 D
Reliability is more complex. A good test should produce similar results time after time, like a bathroom scale. If you weigh yourself on five consecutive days, you would expect your scale to produce similar results every day, not vary between 100 and 200 from one day to the next. Likewise, a hiring test should produce roughly the same results from one time to the next.
2 R% ~( R8 D% g4 W! hReliability also applies to interviews. If three interviewers asked the same questions of an applicant and they heard dissimilar answers, then the interview can be considered unreliable. There are many popular tests that produce unreliable results. How do we know this? University researchers are paid to run investigations.
# ]) p, M+ }3 `" ROne sure sign of unreliability is short tests that produce long narrative reports, or long tests that try to measure too much, or tests that rely on opinions, or complicated scales.
