High stakes lies: police and non-police accuracy in detecting deception

To date, the majority of investigations into accuracy in detecting deception has used low-stakes lies as stimulus materials, and findings from these studies suggest that people are generally poor at detecting deception. The research presented here utilised real-life, high-stakes lies as stimulus materials, to investigate the accuracy of police and non-police observers in detecting deception. It was hypothesised that both police and non-police observers would achieve above chance levels of accuracy in detecting deception, that police officers would be more accurate at detecting deception than non-police observers, that confidence in veracity judgements would be positively related to accuracy and that consensus judgements would predict veracity. One hundred and seven observers (70 police officers and 37 non-police participants) watched 36 videos of people lying or telling the truth in an extremely high-stakes, real-life situation. Police observers achieved mean accuracy in detecting deception of 72%, non-police observers achieved 68% mean accuracy, and confidence in veracity judgements was positively related to accuracy. Consensus judgements correctly predicted veracity in 92% of cases.