Don't trust the gutDoes intuition have any role to play in recruitment?
It’d be pretty useful if, when it came to finding new hires, you had a crystal ball to work out who will flop and who will not. But even if we can’t rely on sorcery, most people believe that they have a few insights into the future performance of candidates.
Most of the time however, this confidence is unfounded. Five years of data from Google, a company with deep pockets and talented employees, showed that their interviewers displayed only a 1% edge over sheer chance when it came to selecting quality hires.
It would be interesting to pick through exactly what’s behind that statistic, but Google never released the report. What it does show, though, is that even at the poster child company for pioneering practice and whip-smart thinkers, intuition on hiring is next to redundant.
Daniel Kahneman, the Nobel-prize winning psychologist has long been warning of the way we blind ourselves to our own abilities. In his own words, ‘intuition is almost useless when it comes to predicting future success’.
In his book Thinking, Fast and Slow, Kahneman analysed data from eight years of hedge fund management by twenty-five fund managers and discovered that their success was no better than random chance. (Despite this, the fund managers continued to believe themselves as skilled and fairly rewarded for their performance.)
This ‘illusion of skill’ is not just a dangerous phenomenon in wealth management. It is clearly alive and well in recruiting as well.
Interviewees are always putting on their best selves, hoping for interviewers to extrapolate competence from shiny shoes. And, thanks to the ‘halo effect’, they often do.
Data, data, data
As it happens, back in the mists of time (the late nineties), there was a large scale meta-analysis of all pre-employment testing across 85 years’ worth of recorded studies, and their validity analysed in relation to job performance.
There are some fascinating pieces of information to consider. For example, did you know that in France and Israel in the 1980s, graphology was the most used test to predict future job performance?
Yes, graphology – the method of reading personality through handwriting analysis. We’ve come a long way since then, and not just because nearly nobody uses a pen and paper anymore. (By the way, for the curious, the studies showed zero correlation.)
Another factor that had no bearing on future performance? Age. In case you ever had any lingering doubts, age showed no relevance to how well somebody could do their job, although presumably this didn’t include things like heavy manual labour or professional bingo.
What else basically had no relevance to performance? Standard CV fodder: years of education, interests, and years of job experience.
Let’s be honest: unless someone’s hobby is creating Excel spreadsheets, we probably weren’t taking it into account except for the purpose of awkward interview small talk.
In that pile you can also throw peer reviews as next to useless. Makes sense. Among your colleagues, not many people will be handing over the unvarnished truth. So what actually does work?
References are slight predictors, understandably a little more private and less biased than peer reviews. Interviews help too, but structured interviews are significantly more effective than unstructured (i.e. eliminating intuition).
The only others which demonstrated decent correlation with performance were: IQ tests, integrity tests, and work sample tests.
Comparisons are(n’t) odious
The best predictors of candidate quality all feed you data that is freer from bias, and allows candidates to be objectively compared to one another.
All the other metrics rely on preconceptions. We believe 20 years’ of experience is beneficial. We draw conclusions from what university somebody attended, how they came across in the interview, if they have interests that mesh with our own. But in reality, these things tell us very little about whether somebody can do a particular job.
An impeccable CV and interview manner are very often most of what is required to land a job, but it seems that both of these are practically wastes of time. Unsettling as it may seem, ranking candidates on objective measures in an interview with rigid targets, and playing examiner with tests, conclusively produces better outcomes.
Data jars with HR’s human orientation. We like face-to-face and empathy. Emotional intelligence is in, not statistical prowess, although we’d like the best of both.
Emotional intelligence is effective for many things that HR does, but in a short-term situation it can be very misleading. In that sense it is a double-edged sword.
To understand somebody and empathise with them — to feel we know them — actually requires a lot of data gathered over a semi-significant amount of time.
It isn’t data you can number crunch or plot into a pie chart, but it is data nonetheless. How we see a person react to a hundred different scenarios, their facial expressions, their words and reactions. It builds a picture of a person over a long time, allowing us to anticipate their reactions and thoughts.
Applying this skill to recruiting is however a disaster. We try to make similar conclusions from a tiny sample size, so we big up the importance of little things — smiles, handshakes, posture — anything.
It’s comparable to a statistician trying to work out the long term viability and profitability of a company from a single day’s worth of figures. The figures may look rosy, but for all you know, it could have been in freefall for months and has just announced liquidation — the joyfully named ‘dead cat bounce’. You lack the data, so your mind invents the rest in order to avoid cognitive dissonance.
‘This person seems nice and professional. Therefore they must be a nice and professional person all the time, and will continue to be for the foreseeable future.’ Put like that, it sounds quite silly. But it is something that plays out unconsciously in our heads from the moment we meet people.
IQ test or a work sample test would probably be an awful predictor of performance compared to a year’s careful observation of somebody working. But in a short-term environment, the hard data they provide is significantly more enlightening than standard interviews and CVs, because in a year’s time their results will likely have only changed slightly.
There is still a place for HR’s human touch and gut instincts, but they play out over a longer time frame.
When it comes to pre-employment testing, they are a handicap HR needs to overcome.