Sunday, August 03, 2014

Personality test software is getting sophisticated, but may lead to overly conforming workforces

Steven Pearlstein has an interesting report in the Business section of the Washington Post on Aug. 3, “People Analytics: ‘Moneyball’ for human resources”, link here. The article title of course invokes the film about baseball with Brad Pitt.

Companies are going back to using tests even more than before, and interviewing fewer people face to face.  Oddly, for contracting jobs in IT, I still found that the final interview with a client was usually by (land-line) phone.    
Personality tests used to be popular even in the 1950s. Starting about 2001, with the 9-11 recession, some companies started using true-false personality tests.  ATT would disqualify someone from re-applying for six months (for a call center job) if the person “failed” the 370-question test.  The TSA used the same personality test.
Pearlstein believes that overuse of “people analytics” can result in an overly homogenous or conforming workforce, one that “preaches to the choir” and becomes insular, setting a company up for eventual competitive decline.

Still, there were interesting observations.  People Analytics 2.0  (link for conference site   the basic domain name doesn’t work!) is “dynamic” and learns from its own experience in an AI mode.  Evolv has debunked some common misperceptions (that felons are risky, or that IQ tests and job hopping matter;  on the other hand long commutes do matter, as well as curiosity at home in computer use.  Evolv found that people who installed Chrome or Firefox at home did better at work than those who relied on factory installed IE or Safari.  “Creativity” mattered in sales jobs (and odd paradox, since you are selling somebody else’s agenda or product) but personal rapport mattered more in call centers.  Pegged Software found that leadership in community or volunteer organizations matters for acute care nurses but not for nursing home workers.  

Testing companies also reported that in IT, evaluation of open-source code written by a candidate, or of performance in video games can be predictive of job success.  I never got into those areas. 

No comments: