Allport and Postman (1947) found some convincing results to support the view that eyewitness testimony is unreliable. They found that participants were more likely to recall that the photograph that they had seen previously had shown a black man holding a razor, when in fact it was the white man holding the razor in the photograph. The only justified explanation presumed for their findings is that this fitted in with the stereotype at that particular time. However, looking at the other point of view, if there is much publicity to fit this stereotype (such as the mass media which can be very influential), then it is bound to have an effect upon how some people perceive other people. Also, with some people it may have been a genuine mistake that they thought the black man was holding the razor in the photograph, not necessarily because they were acting in a prejudice manner. Both Bartlett's and Allport and Postman's studies are low in ecological validity (to everyday life).
Other theories of memory and forgetting can be used to support and explain the view that eyewitness testimony is unreliable. People who believe in valid research that supports the levels of processing theory would say that we have not processed the crime well enough if we do not recall. This could have a great impact to what extent that what people say is trustworthy and subsequently can be taken at face value. Similarly, the context-dependent forgetting theory states that our recall (for eyewitness testimony in this case) may possibly be unreliable if we do not recreate the event as an aide-memoir.
It has also been found that the manner in which leading questions are asked can have a huge impact on eyewitness's testimony. Loftus and Zanni (1975) demonstrated that even small differences in the way in which a question is asked could have an evident consequence. Their participants were shown a short film that showed a car accident.
