This study demonstrates that our memories are not entirely reliable. Another experiment by Loftus and Palmer (1974) shows how memories can be corrupted when presented with contradictory external stimuli. The aim was to investigate the effects of leading questions on the accuracy of speed estimates and perceived consequences of a car crash. In one laboratory experiment, subjects were shown seven films of traffic accidents (5-30 seconds long), and then asked to answer specific questions about the footage. The independent variable was the critical question, “About how fast were the cars going when they _______ (into) each other?”, where the words ‘hit’, ‘smashed’, ‘collided’, ‘bumped’ or ‘contacted’ were randomly inserted into the gap. Results showed that the mean speed estimate when the word ‘smashed’ was used, was higher (40.8) than ‘collided’, ‘bumped’, ‘hit’ and ‘contacted’ (31.8) respectively.
In a second experiment each subject was shown a short film that showed a multiple car accident. Loftus and Palmer asked 50 subjects, “How fast were the cars going when they hit each other?”, and 50 subjects were asked the same question with the word ‘smashed’ substituted instead of ‘hit’. A control group of 50 subjects were not interrogated about the speed of the vehicles. The subjects were then asked to return a week later and without re-viewing the film, answered a series of questions about the accident. In this case the critical question was, “Did you see any broken glass?”.
Results showed that a greater number of subjects claimed to have seen broken glass when previously asked to estimate the speed of the cars when they ‘smashed’ into each other, compared to when they ‘hit’ each other. In actual fact, there was no broken glass in the film.
Loftus and Palmer suggested that two kinds of information go into a person’s memory of a complex event; information obtained from perceiving the event, and information supplied after the event.
This study aims to see if the above effects can also occur when subjects simply read about an event, rather than actually see it, i.e. can their response to the event still be biased? Subjects will be presented with a vignette in which a character (John) drove his car after drinking and either ‘smashed into’ or ‘bumped into’ a garden wall. Subjects will be required to estimate John’s blood alcohol content relative to the British legal limit.
Based on the evidence given above:
The Experimental hypothesis for this investigation is, “The estimate for Driver’s blood alcohol levels will be higher when the word ‘smashed’ is used than when the word ‘bumped’ is used.”
The Null hypothesis for this investigation is “The estimate for the driver’s blood alcohol levels will not be higher when the word ‘smashed’ is used than when the word ‘bumped’ is used.”
Method
Subjects: 20 undergraduate drivers (10 male and 10 female) all with clean driving licences.
Materials: A vignette in which a character named John left his friends’ house at 2:20am. He was unable to get a taxi so he decided to drive home, despite having been drinking. He was travelling at 38 miles per hour, lost control of the car and hit a garden wall. The house owner alerted the police, who then breathalysed John.
Design: Subjects read a vignette in which the driver either ‘smashed into’ or ‘bumped into’ a wall.
Procedure: Subjects were required to read the vignette and then estimate John’s blood alcohol content relative to the British legal limit. For example, 200% would mean twice the limit, 110% would mean 10% over the limit. Subjects were allowed to give any number.
Results
The group of subjects who received the vignette that stated that John had ‘smashed into’ the wall gave a higher estimate (Average= 143%), compared to those who had received the vignette stating John had ‘bumped into’ the wall (Average= 108%). Using an independent samples t-test, the results revealed that the difference between the means t(18)= 2.172, p<0.05, is statically significant.
The independent groups t-test was implemented in this practical because ratio level data was used alongside the independent groups design. The independent groups t-test, assumes homogeneity of variance, even though the means may be different. A problem with this test is that it is difficult to decide if the assumption is satisfied. However as long as the estimated variance is not four times greater than the other sample, the test can be used successfully.
As stated above p was found to be significant at <0.05, therefore we reject the null hypothesis and accept the experimental hypothesis.
That is,
The estimate for the driver’s blood alcohol levels will be higher when the word ‘smashed’ is used than when the word ‘bumped’ is used.”
Discussion
Statistical analysis of the results data shows that p<0.05. This means that the results are statistically significant and have not occurred by chance. This further means that we can accept the experimental hypothesis, and reject the null hypothesis. These findings also support previous work by Loftus and Palmer (1974), on leading questions (which will be discussed in more detail).
The independent groups design was used in this study, which has both advantages and disadvantages. The advantages of this design are that order effects, such as boredom and fatigue, (which can alter concentration) and demand characteristics are eliminated, because different participants can only do one condition of the experiment. Disadvantages with this design, include individual differences between the subjects. These can cause different results being obtained and thus decrease the validity of the experiment. It is however difficult to avoid these individual differences when randomly selecting a sample group. In order to obtain a sample group of similar individuals, subjects would have to be chosen, and this would cause a whole array of experimental problems, for example experimenter bias.
Limitations with the study included the sample range being too small. This means that it is difficult to make generalisations because 20 undergraduate students cannot possibly be representative of the sample population. In order to improve the experiment, a wider sample range could be used, including people from different age groups and ethnic backgrounds.
Another limitation was gender bias. The driver in the vignette was male which could have meant that male participants may have been more sympathetic to him when giving an estimation. If a female character had been used, the results may have been different. This could be improved by using two sample groups, both with female characters, but again with either the ‘smashed into’ condition or ‘bumped into’ condition.
The results of this and previous studies have a number of implications in today’s society. Loftus and Palmer’s work (1974), showed how changing a verb in a sentence, can alter eyewitness recall of a car accident. These findings have major implications for the Legal Justice Systems. In a study by Rattner (1988), a review of 205 cases of wrongful arrest showed that 52% of the cases were associated with mistaken eyewitness testimonies. These results highlighted the unreliability of eyewitness testimony.
Other studies have demonstrated that witnesses sometimes cannot attribute memory to its appropriate sources, or that they make source attribution errors. When witnesses get information from other witnesses and from the police, then their own recollection is likely to be contaminated (Fisher, 1995). This is yet further evidence suggesting that there should be less dependence on eyewitness testimony, and with the arrival of advanced gene technology, it is now possible to use alternative sources of evidence.
References
Carlson, N. R., Buskist, W., Martin, G. N. (2000). Psychology The Science of Behaviour, 266-267
Loftus, E. F. and Palmer, J. E. (1974). Reconstruction of automobile destruction: an example of the interaction between language and memory. Journal of Verbal Learning and Verbal Behaviour, 13, 585-589.
Fisher, R. P., (1995). Interviewing victims and witnesses of crime. Psychology, Public Policy, and Law, 1(4), 732-764.