Subjects who were asked "how fast were the cars going when they smashed into each other?" estimated that the cars were going 40 mph, while subjects who were asked "how fast were the cars going when they contacted each other?" estimated that the cars were going 30 mph. Simply using the word "smashed" rather than "contacted" seems to have increased people's estimates by 10 mph.
Furthermore, when later asked "do you see any broken glass?" subjects in the "smashed" condition were more than twice as likely (about 33%) to say yes than those who had been in the "contacted" condition (about 14%). However, there was NO broken glass in the film subjects saw. Loftus's work indicates how influential post-event information can be on memories.
In another study, Loftus, Burns, and Miller (1978) had subjects view a series of slides depicting an automobile accident. Later a test was given about what depicted, and in the test was either a question that referred to a yield sign or did not refer to a yield sign. In fact, there was a slide with a stop sign, but none with a yield sign. Later in identifying slides that they had seen before in a forced choice situation (a slide with a yield sign v. a slide with a stop sign), subjects' responses were significantly different based on whether they had been asked about a yield sign or not.
Altering people's responses based on questions that are misleading is known as the misinformation effect.
There are two leading explanations for the misinformation effect. One is that the information in post-event questions is accepted in place of what was actually seen; this is known as misinformation acceptance. The second explanation is that people sometimes have difficulty identifying the actual source of the memory being recalled (is it from what was seen or what was asked?); this is known as source confusion.
Moreover, simply informing people that their memories may be influenced by post-event information will not eliminate those influences. Indeed, cognitive psychologists, who one would expect to be least likely to be susceptible to this memory distortion have been shown to be just as vulnerable as anyone else. Vicente & Brewer (1993) surveyed some cognitive psychologists for their knowledge of a well known study by deGroot (1965) that showed grandmaster chess players had superior memory for chess pieces on a chess board with just a brief glance at the board than novice players, which the cognitive psychologists were able to recall as deGroot's main finding.
However, these cognitive psychologists recalled that deGroot also had a condition in which chess pieces were put on the board in random order, which resulted in no difference in recall between the grandmasters and the novices, which would rule out general memory superiority as an explanation of the results. But deGroot did not have that condition. In their partial defense, a well known replication of deGroot's work by Chase and Simon (1973) did include that control condition.