Smart Testing Doubles Performance

Paul Starr, a Director at People Measures, takes on the hot topic of the value of psychometric testing. He argues that companies need to use reputable psychologists who employ sophisticated tools and who can remain impartial when reporting the test results of those at the pointy end of a business. Here, he responds to a story that recently appeared in The Australian Financial Review.

It was great to see Geoff Campbell having the courage (and insight) to say what organisational psychologists have known for years (AFR 28.11.19), namely that intelligence tests (and other psychometrics, such as personality tests) are invaluable tools for the selection of CEOs.

Several myths and errors must, however, be challenged.

The average organisational psychology student can sprout at least a dozen studies that show properly chosen and conducted psychometrics are at least twice as predictive as interviews, references and other more commonly used processes. The fact that references helped Elizabeth Wettlaufer get another job after she’d secretly murdered seven patients in an Ontario nursing home should give anyone who places weight on them food for thought. Properly managed psychometrics become even more powerful when coupled with business simulations.

It’s been more than twenty years since Schmidt and Hunter’s seminal research in 1998 which showed that a senior executive whose intelligence is in the top 15 per cent will be 96 per cent more productive than someone who is below average. They showed that intelligence has double the impact on CEO and C-suite performance as it does at middle management (where ironically testing is more commonplace). Unfortunately, there were a number of myths and errors in the article that shouldn’t stand uncorrected. Firstly, the suggestion that humility and conscientiousness, (and curiosity and agreeableness) are the same thing is not only contrary to psychological research but any fifteen-year-old can tell you they have friends at school who are incredibly conscientious but not especially humble.

Similarly, the view that “the results can vary depending on an individual’s mood” is one frequently used to denigrate psychometrics by those who know little about them. Of course, psychometrics can be affected by ‘mood’, but so is every form of selection, such as interviews (which in fact are affected significantly more). Unlike interviews or reference checks however, psychologists have actually studied the impact of ‘mood’ or time of day or other confounding variables on psychometric results, which show that they are significantly more consistent over time than other forms of selection. To suggest that psychometrics are affected by mood without acknowledging that other forms of selection are is both misinformed and misleading.

A final important piece of information missing from the article is that several of the views expressed were those of executive search firms or recruiters, who collect their fees when a successful candidate is recruited. If psychometrics show that their recommended candidate is not suitable they stand to lose, in the case of CEO’s, potentially hundreds of thousands of dollars. The lay reader may not be aware when hearing the views of these experts of this inevitable conflict of interest. Psychometrics aren’t infallible (an argument we hear often is “I did a test once and it showed I was X when really I’m Y so they can’t be any good”). But of course, nothing that attempts to measure human behaviour is perfect. What makes people interesting also makes them confounding; we’re not completely predictable (just ask any parent). But at least science can show us what is more or less accurate when trying to predict how people will perform.

At People Measures we are very happy to discuss this science, rather than just share our opinions.