In a second experiment the researchers asked a group of 28 different undergrads to complete the Symptom Checklist-90 (SCL-90)—a 90-item self-report survey of general psychiatric stress, including difficulty concentrating, social anxiety, depression, sadness and panic attacks as well as somatic symptoms like headaches. Participants rated how frequently they experienced each item using a five-point scale, where 0 meant "not at all," 2 meant "occasionally" and 4 meant "all the time".
Once more, the experimenters had undergrads busy themselves with sudoku puzzles after completing the survey—but this time the researchers secretly changed some of the participants' answers while they were distracted. Specifically, the researchers manipulated two items: one about concentration difficulties and another about social anxiety, increasing low scores by two scale points or decreasing scores by two points if the participant had circled a 3 ("a lot") or 4.
Then the experimenters returned the answer sheets to the undergraduates and asked them to explain their scores on 10 items, including the two tweaked answers. Fifty-seven percent of the subjects failed to notice both manipulated answers, and more than two thirds of the subjects justified scores they had never actually reported. For example, if the researchers had switched a participant's original score on concentration difficulties from 0 to 2, the participant would explain the answer by citing an excess of coffee or anxiety about exams. The experimenters subsequently provided the participants with an abbreviated 30-item version of SCL-90 and found that on the second time around people who had justified their manipulated answers filled out the SCL-90 in the direction of the manipulation. Although this shift was statistically significant, Merckelbach and his colleagues would like to independently confirm it with more research, they noted in their study.
"If you play the role of having a disease, then at some point the symptoms may become very real to you," Merckelbach says, adding that the progression from purposefully faking symptoms to truly believing them could be exacerbated by doctor–patient relationships. "For example, when you talk about whiplash or chronic fatigue disorder, you can imagine a patient who starts out playing these symptoms, but when he is asked by a physician, 'Do you also have this or that?' and the questions are posed over and over again, the patient may lose sight of the fact the they are playing a role."
Loftus thinks the new study also shows how people can delude themselves. "The second experiment is more like the misinformation experiments I have done," Loftus said, referring to studies in which intentionally manipulative questions affected people's memory of footage from a car accident. "They don't know they are being deceived. But this study shows you can deceive yourself, too. In the first experiment, they [the participants] know they are deliberately faking. It's like a lie—a kind of a lie. But later on, they don't stop lying, even though they know they don't need to. They've deceived themselves."
Both Merckelbach and Loftus think the recent study is particularly relevant to malingering in legal procedures—when people feign illness or injury for a specific personal gain, like decreased criminal responsibility or increased financial compensation. Merckelbach drew inspiration for the study from the 20th-century Russian playwright Leonid Andreyev's The Dilemma, in which a character who malingers a disease eventually develops the symptoms that he faked. "Andreyev was the first to describe this phenomenon, and he was a court reporter," Merckelbach says. "He may have seen it with his own eyes."