In the study, each student was paired with a member of the research team disguised as another participant. The pair sat side-by-side and played a computerized gambling game, which involved betting fake money based on the likelihood of answering a multiple choice question correctly. Each person was in charge of keeping track of their own wins; when a subject correctly answered a question, they got to take money from a shared “bank,” and when they incorrectly answered a question, they had to put money back. Participants were told that at the end of the game, the person who made the most money would win a prize.
After the gambling concluded, the researchers used Final Cut Pro to alter a video recording of the game and make it look like the partner had cheated. Five to seven hours after the first task, students were called back to the lab and told that their absent partner was suspected of cheating. One-third of the students were also told that the researchers had video evidence of the cheating, and another one-third got to watch the doctored video themselves.
Before asking participants to sign an eyewitness testimony, the researchers emphasized that no one should testify unless they were 100 percent sure they had seen their partner cheat, and they emphasized that the cheater would be punished. Students who watched the fake video were far more likely to give false testimony than students who heard about the video or were simply told that their partner was suspected of cheating.
When asked to describe what they had seen, some participants even invented memories. “One subject told us that the other person had acted suspiciously and taken money from the bank when there was clearly a cross on the screen,” Wade wrote. “So we are confident that a significant portion of people who saw the fake video genuinely believed—or even falsely remembered—that they had witnessed the cheating.”
http://www.wired.com/wiredscience/2009/09/falsetestimony/