loftus and palmer

Question 1

aim

Answer 1

To investigate the effect of questioning on witness memory of
a car accident.

Question 2

experiment 2

Answer 2

150 students underwent a similar procedure but
were asked about broken glass at the scene. When the word
‘smashed’ was used, participants still estimated a higher speed,
but, in addition, they wrongly remembered seeing broken glass at
the scene of the crash.

Question 3

experiment 1

Answer 3

45 students watched film of car crashes. They
were then asked to estimate the cars’ speeds, using different verbs
to describe the crash. Estimated speed varied according to the
verb used, with ‘smashed’ leading to the highest estimates.

Question 4

conclusion

Answer 4

Wording of questions can alter witness memories
of events.

Question 5

CONTEXT

Answer 5

How well do you remember events you have seen? Psychologists are concerned
with the accuracy of our memory of events. As far back as 1909, G.M. Whipple
reviewed evidence and concluded that eyewitnesses’ memory of events is
considerably less accurate that we would like to believe. This may be particularly
true when we are asked to recall numerical values, such as time, distance or
speed. By the 1970s, several studies had shown that people tend to over-estimate
the time and speed involved in complex events. In one study, Marshall (1969)
asked Air Force personnel to estimate the speed of a car that they had been
watching. Although the participants knew that they would be questioned, their
responses varied wildly and were inaccurate (their estimates were between
10–50mph, whereas the actual speed was 12mph).
Fillmore (1971) suggested that one such factor might be the language used to
describe the motion, and that using words such as ‘smashed’, rather than more
neutral words, such as ‘hit’, could lead people to judge speed to be greater.
Loftus and Palmer subsequently proposed that: ‘Given the inaccuracies in
estimates of speed, it seems likely that there are variables which are potentially
powerful in terms of influencing these estimates’ (1974: 585). In other words, if
we are poor at judging speed, then there must be factors other than the actual
speed that affect our judgement.
The inaccuracy of eyewitness memory, and the potential for memories
to be distorted by the use of language, have important practical applications. In
particular, the police and the courts often rely on eyewitness testimony in
order to make decisions about what actually took place and who was responsible
for what happened. By the time of Loftus and Palmer’s study, there was concern
in legal circles about the use of leading questions, and the likelihood that such
questions can cause inaccurate eyewitness testimony. Loftus and Palmer define
a leading question as ‘one that, either by its form or content, suggests to the
witness what answer is desired or leads him to the desired answer’ (1974: 585).
The present study is concerned with the effect on eyewitness memory of asking
leading questions about the speed of a car.

Question 6

GENERAL AIM

Answer 6

The overall aim of the study was to test whether the phrasing of questions about
a car accident could alter participants’ memory of an event.

Question 7

EXPERIMENT 1 : AIM

Answer 7

The aim of the first experiment was to see whether using different verbs to
describe a collision between two cars would affect estimates of the speed at
which they were travelling when the crash took place.

Question 8

METHOD - participants

Answer 8

Forty-five students took part in the first experiment. No details of age or gender
were recorded.

Question 9

METHOD - Design and procedure

Answer 9

The study was a laboratory experiment using an independent measures
design. Participants were shown seven films of car crashes, taken from training
films used by the Seattle Police Department and the Evergreen Safety Council.
In four of the films the speed of the car was known because the crashes were
staged for training purposes. The speeds in these films were 20mph, 30mph,
40mph and 40mph. After watching the films, all participants were asked to write
an account of the accident and then to answer a series of questions. All but one
of the questions were fillers, designed to make it harder to work out the aim of
the experiment. The other question was a critical question, meaning that it was
closely concerned with the aim of the study. This question was: ‘About how fast
were the cars going when they hit each other?’
The independent variable was the verb used in the critical question.
For one group this was ‘hit’. The other groups received the same question but
with the verb ‘contacted’, ‘bumped’, ‘collided’ or ‘smashed’ instead of ‘hit’. The
dependent variable was the mean estimated speed of the car.

Question 10

key term : Laboratory experiments

Answer 10

take place under
controlled conditions. They test cause and
effect by comparing two or more conditions.

Question 11

RESULTS

Answer 11

Results were in the form of quantitative data. Participants’ estimates of the speed
at which the cars were travelling were not affected by the actual speed. The
mean estimates for each of the crashes in which the speed was known are shown
in Table 2.1. This shows that we are generally poor at estimating speed. However,
estimates of the cars’ speeds did vary according to the verb used in the critical
question. These results are shown in Table 2.2.

TABLE 2.1 MEAN ESTIMATES AND ACTUAL SPEED IN FOUR CRASHES

1 - actual speed 20mph - estimated 37.7
2 - actual speed 30mph - estimated 36.2
3 - actual speed 40mph - estimated 39.7
4 - actual speed 40mph - estimated 36.1

TABLE 2.2 MEAN ESTIMATES OF
SPEED IN ANSWER TO THE CRITICAL
QUESTION

verb - smashed - 40.5mph
verb - collided - 39.2
verb - bumped - 38.1
verb - hit - 43.0
verb - contacted 31.8

Question 12

CONCLUSIONS

Answer 12

Participants’ estimates of the speed at which the cars were travelling when the
accident took place varied according to the verb used to describe the crash. There
are two possible reasons for this:

1 Response bias. When a participant is unclear what speed to estimate, the
verb gives them a clue as to whether they should estimate a high or low
figure.
2 Memory distortion. The verb used in the question actually alters a
participant’s memory of the crash.

Question 13

EXPERIMENT 2 : AIM

Answer 13

The aim here was to investigate whether the different speed estimates found in
Experiment 1 were, in fact, the result of a distortion in memory. This was done
by seeing whether participants who heard the words associated with high-speed
estimates would be more likely to incorrectly remember broken glass at the
crash site.

Question 14

METHOD- Participants

Answer 14

One hundred and fifty students took part in the second experiment. No details of
age or gender were recorded.

Question 15

METHOD - Design and procedure

Answer 15

As in the first experiment, the method was a laboratory experiment with an
independent measures design. All participants watched a film of a car crash.
The entire film lasted less than one minute, and the accident itself lasted four
seconds. All participants were given a questionnaire that first asked them
to describe the accident in their own words, and then to answer a series of
questions. As in the first experiment, there was a critical question. The first 50
participants received the question: ‘About how fast were the cars going when
they smashed into each other?’ Another 50 participants received the question:
‘About how fast were the cars going when they hit each other?’ Finally, a control
group of 50 participants received questions that did not ask about the speed
of the cars. A week later, the participants returned and answered a further 10
questions. The critical question among these was: ‘Did you see any broken glass?’

Question 16

RESULTS

Answer 16

As in Experiment 1, participants who heard the word ‘smashed’ in the critical
question estimated a higher speed (10.46mph as opposed to 8.0mph in the ‘hit’
condition). The numbers of participants reporting that they had seen broken
glass in each condition is shown in Table 2.3. More than twice as many people
incorrectly remembered seeing broken glass having heard the word ‘smashed’
in the question compared to those who heard ‘hit’ or who heard no question
about speed.

response - yes smashed : 16 hit : 7 control : 6

response no smashed : 34 hit : 43 control : 44

Question 17

CONCLUSIONS

Answer 17

The general conclusion from the two experiments is that the way in which
questions about events are worded can affect the way in which those events are
remembered. The results of Experiment 2 are important because they strongly
suggest that this is not simply due to response bias. Instead it seems that post-
event questions actually become part of the memory for that event. Therefore,
the wording of questions can actually distort event memory.

Question 18

EVALUATION : research method

Answer 18

The study was a laboratory experiment. Because the
procedure took place in a highly controlled environment,
with precise timing of films, presentation of the question
order, the inclusion of fillers, etc., it was possible to eliminate
many extraneous variables. The researchers could therefore
be reasonably confident that it was the independent variable
of the verb that was affecting the dependent variables of
speed and recall of broken glass. The potential weaknesses
of laboratory studies such as that of Loftus and Palmer lie in
the realism of the environment and the participants’ tasks.
It is hard to set up laboratory procedures in such a way as to
ensure that people behave as they would in real life. In this
case, watching a film is not the same experience as witnessing
a real event. Actual car crash witnesses would be likely to
experience much more intense feelings, such as fear or shock,
and emotions are known to affect memory. Witnesses to a
real crash would also have much more significant motives
for accurate recall – their testimonies would have genuine
consequences for convictions.

Question 19

EVALUATION : Qualitative and quantitative data

Answer 19

The data gathered in this study were quantitative. This is both
a strength and a weakness. On the plus side, the statistics
allow easy comparison of the conditions, clearly showing that
memory is affected by the wording of questions. On the other
hand, there was no opportunity for participants to comment
either on what they remembered, or on their experience of being questioned in this way. Such qualitative data might have
added to the completeness of the findings.

Question 20

EVALUATION : Ethical considerations

Answer 20

This is a straightforward laboratory experiment, with minimal
ethical issues. One possible ethical issue might concern any
participants who happened to have experienced real car
crashes. As the participants were students, they might have
felt obliged to participate even if their experiences meant that
they didn’t want to. If so, this would raise a question about
whether their informed consent was genuine.

Question 21

EVALUATION : Validity

Answer 21

The high levels of control imposed by conducting the study in
laboratory conditions ensured that few extraneous variables
could influence the outcomes, for example the filler questions
reduced the likelihood that the participants worked out the
aim. Some of the films were of real accidents, so in this respect
the context was realistic. These factors raise validity. However,
the overall validity would be reduced both by the lack of
realism, in terms of the artificiality of the remaining films and
the context, and by the possibility that some participants
might have worked out the aim of the experiment, especially
in Experiment 2. These factors threaten the ecological validity
of the study as both the environment and the task were quite
artificial. Participants had a better view of the crash than is
typical in real-life situations, but they were probably more
relaxed and less motivated to remember details, given that they were in a familiar and safe situation and knew that they
were taking part in a study. Remember that it can be quite
upsetting to witness an accident in real life, and this emotional
response can make the event more (or less) memorable.
It is therefore possible that participants’ memory was
unrepresentative for these reasons.

Question 22

EVALUATION : Reliability

Answer 22

Laboratory procedures are highly standardised, for example
the length of the films was specified, which makes them
reliable. The standardisation also ensures that the procedure
is replicable. In the case of Loftus and Palmer’s study, it
is interesting to note that very similar results were found
for speed estimates with the verbs ‘smashed’ and ‘hit’ in
Experiments 1 and 2. This suggests that the findings are
reliable.

Question 23

EVALUATION : Sampling bias

Answer 23

The participants were all students, not chosen by any
representative sampling method. This means that they were
unlikely to be truly representative of the population. They
are likely to have been predominantly white, middle-class
and within a narrow age-range, and they all had the same
occupation. This is important for the following reasons:

• Because the participants were students they could have
  been particularly vulnerable to demand characteristics.
  In other words, they might have been strongly influenced
  by cues suggesting what the researchers expected to
  find. However, the independent measures design would
  possibly have eliminated the worst of this problem.
  *Because the participants were all students, they were
  very used to taking in information and being tested on it,
  so they might have been better able to recall accurately
  than most people.
  *The participants were less likely to be drivers than
  the population as a whole, and their speed estimates
  might have been less accurate as a result of their lack of
  experience with cars.
  These sample characteristics could mean that the results were
  partly a product of the sample.

Question 24

EVALUATION : practical applications

Answer 24

Studies such as this are important in helping authorities to
understand how to question witnesses to important events
such as accidents and crimes. Following the work of Loftus
and her colleagues, the use of leading questions – both by the
police immediately after an event, and later in the courtroom
– is now tightly controlled. This is likely to have improved the
rate of successful criminal convictions.