There's a problem with assuming the most intelligent candidates make the best employees

Workplace research through the 20th Century suggested that selecting for intelligence is the best way to identify good performers. General mental ability (GMA), a popular recruitment measure that maps closely to the colloquial meaning of "intelligence", is strongly correlated with on-the job performance, well ahead of any other single measure.

This consistent finding came from studies that mostly defined job performance as carrying out the duties expected in that role. Although intuitive, this neglects two types of "extra-role" behaviours identified and studied in more recent years: citizenship behaviours, such as volunteering time or treating colleagues with courtesy; and counter-productive work behaviours, such as spreading rumours, shirking, or theft. Now a new meta-analysis suggests that GMA isn't the best predictor of these crucial aspects of performance. In fact, intelligence may be of little use in predicting who will behave badly at work – although it may predict who can get away with it.

The meta-analysis winnowed the available literature down to 35 relevant studies that looked at citizenship and counterproductive behaviours in real organisations. Intelligence (GMA) was correlated with engaging in more citizenship behaviours, but the association was far weaker than between intelligence and traditional task-based measures of performance. The researchers led by Erik Gonzalez-Mulé then cross-compared their results with previous meta-analyses focused on personality, and concluded that personality and GMA each account for about half the variance in citizenship behaviours. Put another way, you're just as likely to do good because you're inclined that way, as you are because you're smart.

Turning to counterproductive workplace behaviours, the authors predicted a relationship here with intelligence/GMA based on evidence from criminology that's shown helping people see the consequences of their actions has an inhibitory effect on aberrant behaviour. In fact, the new analysis found no association between intelligence and aberrant behaviour. It's possible that this discrepancy with the criminology findings is because of differences in samples: there may be low-intelligence individuals who are more disposed to malfeasance, but they are underrepresented in workplaces because of adolescent anti-social issues, such as truancy or criminal behaviour. Meanwhile, personality, particularly the trait of agreeableness (but also conscientiousness and openness to experience) was strongly associated with performing fewer unhelpful behaviours at work.

An interesting footnote – when self-ratings of counterproductive behaviour were removed from the analysis (leaving only third-party ratings), the results showed a significant relationship between intelligence and (fewer) unhelpful workplace behaviours. This means that smarter people report engaging in just as much bad behaviour as the rest of us, but others, such as work supervisors, notice less of it.

In summary, while GMA is the undisputed king of predicting better task performance, it holds equal footing with personality in predicting helpful, altruistic work behaviour, and cedes the ground almost entirely to personality for bad behaviour. Looking at performance as a composite of these three areas, Mulé's team conclude that when it comes to workplace selection, GMA still has a prominent role, but a much diminished one.

Further reading

A Meta-Analysis of the Relationship Between General Mental Ability and Nontask Performance.