Key Study: Leading questions and the misinformation effect – ” the car crash study” (Loftus and Palmer, 1974)

Travis DixonCognitive Psychology, Internal Assessment (IB), Key Studies, Studies and Theories

Loftus and Palmer's first experiment showed how leading questions can affect answers, but did it really show how it can affect memory?

 Memory is a reconstructive process, which means memories are actively and consciously rebuilt when we are trying to remember certain things. Elizabeth Loftus, her colleagues and others studying this cognitive phenomenon have shown that during the reconstruction phase our memories can be distorted if we are given false information about the event – this is called the misinformation effect.

Background Information

Some of Elizabeth Loftus’s first studies focused on how language can influence memories of particular events. Research prior to the following two 1974 experiments suggested that people are quite inaccurate when asked to report numerical details regarding events. Also, as memory has been shown to be reconstructive in nature, Loftus and Palmer predicted that the wording of a question could influence recall. They define a leading question as “one that, either by its form or content, suggests to the witness what answer is desired or leads him (sic) to the desired answer).”

Have you ever seen this in a film or on TV in a court-room drama? The lawyer asks the question and the opposing lawyer shouts, “Objection! Leading the witness, your Honour”. They are objecting to the use of a leading question – asking in a question that is guiding (or leading) the respondent towards a particular answer.

For example, I would be asking a leading question if I asked you, “how much do you like Psychology?” I’m already implying in my question that you do in fact like Psychology, I simply want to know how much. You’re lead to answer in a way that suggests you like this subject. What if you hate it, or find it immensely boring? It would be more difficult to respond this way to this particular question.

In the following two experiments, Loftus and Palmer first studied the effects verbs in questions on speed estimates and also if these verbs could impact memory in other ways.

The following information has been adapted from our textbook, IB Psychology: A Student’s Guide.

Key Study Experiment #1: 5 verbs in leading questions.

In this first experiment, 45 college participants were divided into five groups of nine and watched seven short videos (5 – 30 seconds) taken from driver’s education courses that involved a traffic accident of some kind. The participants were first asked an open-ended question: “Give an account of the accident you have just seen”, which was followed by a series of specific questions about the accident. There was one critical question that asked “About how fast were the cars going when they … each other”. The five groups were given five different verbs. I.e. one group was asked “hit”, one was asked, smashed, etc.

2 Car Crashed

The results were as follows (mph):

  • Contacted: 31.8
  • Hit: 34.0
  • Bumped: 38.1
  • Collided: 39.3
  • Smashed: 40.5

A note on the films and speed estimates: Four of the seven films were staged crashes made specifically for education purposes, and so the precise speed in mph (miles per hour) of the vehicles is known. The results below show the actual speed of the car in the video (first number) and the mean guesses from all participants (second number)

  • 20mph = 37.7mph
  • 30mph = 36.2mph
  • 40mph = 39.7/36.1 (there were two films of 40mph).

From the above results it shows that the different verbs can lead to different speed estimates. The researchers provided two possible explanations for these results. The first explanation is that the participants might not have been sure about the speed and the verb simply led them towards a particular answer. If they were not sure of the speed and thought it was around 30 to 40mph, the verb would have biased their answer in a particular direction. This doesn’t tell us much about the reconstructive nature of memory and is more a possible limitation in the research methodology, if anything.

However, they also hypothesized that perhaps the verb “smashed” caused the participants to remember the crash differently. During the process of imagining the crash in order to remember the details and answer the questions, the verb may have affected the memory itself. The participants might have actually been imagining a more severe crash and a faster speed than was really portrayed in the video because of the leading question; when remembering the incident and playing it over in their minds, the verb “smashed” might have led to an actual change in the memory of the video.

But this data doesn’t provide strong support for this hypothesis so they conducted a second experiment, which will be explained in the next section.

Experiment #2: The broken glass manipulation

In this study, 150 participants were put into three different groups but all watched the same film (in smaller groups). The film showed an accident involving many cars and the entire film lasted for less than one minute and the accident part of film lasted 4 seconds. After the participants watched the film, they were given a questionnaire. The first question was again open-ended and asked the participants to describe the accident in their own words. This was followed by a series of specific questions, with one critical question.

  • 50 participants were asked “About how fast were the cars going when they smashed into each other?”
  • 50 participants were asked “About how fast were the cars going when they hit each other?”
  • 50 participants weren’t asked any questions about speed.

    If the verb smashed significantly increased the memory of broken glass when there was none, this is stronger evidence to show that the verb was acting as false information which was actually changing the memories of participants in this condition.

One week later all participants returned and were asked a series of ten questions but they didn’t watch the film again. One of the ten questions appeared randomly in a different order for each participant and asked: “Did you see any broken glass?” And there was a check-box for Yes or No.

Once again the results showed that the speed estimates of those asked about the cars with the verb “smashed” were higher than those with the verb “hit” (10.46mph and 8.00mph respectively).

Here are the results regarding the memory of seeing broken glass:

Distribution of “Yes” and “NO” Responses for Different Conditions
Response Smashed Hit Control
Yes 16 (32%) 7 (14%) 6 (12%)
No 34 (68%) 43 (86%) 44 (88%)

These results provide some evidence for the explanation that the misinformation effect was occurring. Perhaps the verb “smashed” was influencing people’s recollections of the crash and they were remembering it as being more severe than it really was, which is why they could remember seeing broken glass even when there wasn’t any in the original video.

Loftus and Palmer argue that two types of information are influential in making up someone’s memory. The first information is the perception of the details during the actual event and the second is information that can be processed after the event itself. In this case, information from our environment might impact our memory processes, which could lead to distortions. They argue that the verb “smashed” provides additional external information because it shows that the cars did actually smash into each other. The verb that has connotations of a stronger and more severe impact than hit or collided could result in a memory of the incident that never happened, like remembering broken glass when there was none. Remember that the second question was asked an entire week after the original videos were viewed and the leading questions asked. The participants are reconstructing their memories after one week and the difference between the scores is quite significant.

Exam (and IA) Tips

  • These studies can be used to show the reconstructive nature of memory.
  • If asked about “one study” it would be fine to write about both of these versions of the same experiment – the focus should be on the second one, though. 
  • The second study is the important one to be able to explain in exams as it shows the reconstructive nature of memory.
  • For the IA, I would not use the broken glass version of the experiment as it gathers nominal data and this makes the inferential stats a little more difficult. The best option is to choose two verbs from the first study and replicate that.
  • This study could be used for schema theory, but I prefer other studies (e.g. Bransford and Johnson)

Critical Thinking Questions

  • What do these experiments show that memory is reconstructive?
  • Is this study limited in population validity? For example, look at the accuracy of their guesses in the first experiment – is this evidence that perhaps these results might not apply to other groups of people? (Think about experience).
  • What are the possible practical implications of these findings?
  • What are the ethical considerations involved in these experiments?
  • Can you find any other limitations with this study?


Loftus, Elizabeth F., and John C. Palmer. “Reconstruction of Automobile Destruction: An Example of the Interaction between Language and Memory.” Journal of Verbal Learning and Verbal Behavior 13.5 (1974): 585-89. Web. (Read full text here)