Loftus and Palmer

To investigate the effect of leading questions on the accuracy of speed estimates in, and observed consequences of, a car crash.

Experiment 1) 45 students from the University of Washington.

• They were split into 5 groups of 9.

Experiment 2) 150 students from the University of Washington.

• They were split into 3 groups of 50.

• Participants are from the University of Washington.

• Those who attended University at the time of this experiment were usually white males.

• This means that the results are difficult to generalise as the results are not representative of a range of ethnic groups.

• The size of the sample, in experiment one especially, is very small.

• This means that the results are difficult to generalise as the results are not representative of the whole population.

• The memory capacity of students differs to the memory capacity of the general population.

• Students are practiced at memorising information.

• Or they have too much important school information on their mind to fully commit themselves to, what they consider to be, a less important experiment.

• They may be eager to please senior faculty members, this is a confounding variable.

• Results are difficult to generalise as they're not representative of the whole population.

Participants were shown 7 videos of car crashes:

• Ranging from 4 to 30 seconds long.

• They were shown in a random order.

• They were taken from films made for driver education courses so that researchers were aware of the speeds involved.

Participants were asked to complete a questionnaire:

• They had to give an account of the film they had seen.

• They had to complete a number of "filler" questions, which would not be analysed.

• The question that would be analysed was:

  "How fast were the cars going when they ***** into each other?"

Participants had been divided randomly into 5 groups of 9.

• Each group was allocated a different verb that fitted into the question.

• Each verb indicated a different level of drama.

Smashed, Collided, Bumped, Hit, Contacted.

The results show that the more dramatic the verb, the higher the speed of the cars was estimated to be.

• Smashed - 40.8 (mean speed estimate)
• Collided - 39.3
• Bumped - 38.1
• Hit - 34.0
• Contacted - 31.8

• In the first experiment there was a level of uncertainty of whether prompts given by the researcher (verbs) had led the participants to:
• Alter their memories (distortion),
• Or whether it had led them to give a certain answer (response-bias). 
• Due to this a second experiment was carried out.

The verbal label in the question given to the participants could have cognitively amended their memory.

The participants may not have been sure about the speed of the car to start with, and may have adjusted their answer to fit in with the expectations of the researcher (a form of demand characteristics).

To show that information provided after an event is capable of distorting memories.

All 150 participants were shown a short film of a multiple car crash.

• The video lasted 1 minute.
• The actual action only lasted 4 seconds.

Participants were split into 3 groups of 50.

• Group 1 - were asked "how fast were the cars going when they smashed into each other?"

• Group 2 - were asked "how fast were the cars going when they hit each other?"

• Group 3 - were not asked a question on the speed of the cars; they formed a control group.

Participants were called back 1 week later.

• They were asked another series of questions about the film they had watched.

• 9 out of 10 of these questions were "filler" questions.

• The question which the answer of would be analysed was "did you see any broken glass?" (There was no glass in the actual film).

• If participants with the harsher verb, smashed, answered yes to seeing broken glass,

• and participants with the less dramatic verb, hit, answered no to seeing broken glass,

• then this would demonstrate that our memories can be distorted by information provided after an event, and the results of experiment one were due to distortion and not response-bias.

Researchers had a high degree of control over confounding variables

• The study was lab based.
• Researchers could ensure that they could control a wide range of variables better so than if they were based outside a laboratory.
• This makes results more reliable.

The experiment lacks ecological validity.

• Watching a video in a laboratory is different to experiencing a car crash in real life.
• Less emotional involvement in watching a video.
• Participants were cued to watch the video.
• A real life crash would be unexpected.
• This may affect the performance of participants, changing the results.

The setting up of a car crash on a test track:

• The experiment would be in a more 'real life' surrounding and the way a participant views a car crash would be different than watching a video.
• If a group of people watch a video they'll all witness the same thing, but if they all witness a car crash in real life they'll all see it from different perspectives. 
• This could change the way they answer the questionnaire and the results of the experiment may change.
• It is almost impossible to make the experiment 100% ecologically valid as this would break ethical guidelines.
• By transferring the car crash to a test track would also be expensive and difficult to maintain control of confounding variables.

By increasing the sample size the results will be more representative of the population and easier to generalise.

By adding more people to the sample of different ethnic origins the sample would be less ethnocentric and the results would therefore be easier to generalise and more representative of the whole population.

These results supported those of the previous study, with participants overestimating the speed of the cars when asked how fast they "smashed" into each other.

Smashed:

Yes - 16

No - 34

• This was the largest amount of people from a group to say yes, they had seen broken glass, which shows that a selection of participant's memories had been distorted by the leading question.
• However, 34 participants said no, which was the correct answer, showing that distortion only affected a selection of participants.

Hit:

Yes - 7

No - 43

There is no distortion here as this was the less dramatic verb and the correct answer was no.

Control Group:

Yes - 6

No - 44

Only 6 out of 50 participants said yes, maybe they had forgotten and guessed, as it had been a week since they saw the video.

• Both used a laboratory setting to test participants.

• Both have useful practical applications. 

- L&P has helped develop the approach to police eyewitness testimony.

- S-R has helped us see how language is developed in small children.

• Both collected quantitative data (numbers).

• Both were intended to be experiments.

• Loftus and Palmer experimented with humans.

• Savage-Rumbaugh experimented with chimps.

• Loftus and Palmer was based in a laboratory.

Independent Variable - The verbs (smashed, collided, bumped, hit, contacted)

• Baron-Cohen was a Quasi experiment and took advantage of naturally occurring independent variables.

Independent Variable - 

Experiment 1)

Independent Variable - The 5 verbs (smashed, collided, bumped, hit, contacted).

Dependent Variable - The speed estimated

Experiment 2)

Independent Variable - The verbs hit and smashed or none at all.

Dependent Variable - The answer to "Did you see any broken glass?"