A recent study published in Human Communication Research by researchers at Korea University, Michigan State University, and Texas State University -- San Marcos found that using active questioning of individuals yielded near-perfect results, 97.8%, in detecting deception.

An expert using the Reid Technique interrogated participants in the first study, this expert was 100% accurate (33 of 33) in determining who had cheated and who had not. That kind of accuracy has 100 million to one odds. The second group of participants were then interviewed by five US federal agents with substantial polygraph and interrogation expertise. Using a more flexible and free approach (interviews lasted from three minutes to 17 minutes), these experts were able to accurately detect whether or not a participant cheated in 87 of 89 interviews (97.8%). In the third study, non-experts were shown taped interrogations of the experts from the previous two experiments. These non-experts were able to determine deception at a greater-than-chance rate -- 79.1% (experiment 1), and 93.6% (experiment 2).

Previous studies with "experts" usually used passive deception detection where they watched videotapes. In the few studies where experts were allowed to question potential liars, either they had to follow questions scripted by researchers (this study had no scripts) or confession seeking was precluded. Previous studies found that accuracy was near chance -- just above 50%.

"This research suggests that effective questioning is critical to deception detection," Levine said. "Asking bad questions can actually make people worse than chance at lie detection, and you can make honest people appear guilty. But, fairly minor changes in the questions can really improve accuracy, even in brief interviews. This has huge implications for intelligence and law enforcement.