Loftus and Palmer conducted (1974) a laboratory experiment which was strictly controlled, giving it high experimental validity as they could make sure that IV affected the DV. Both experiments proved that memory can be unreliable giving results concurrent validity. It also matched what Bartlett said, again, giving it concurrent validity. Furthermore, when they repeated the experiments and the results showed a consistent trend, meaning it had concordant validity. They produced quantitative data which is objectively analysable (e.g the mph predicted) The could carry out statistical analysis of the results. The experiment also had standardised instructions for every participant. For example all 45 participants had the same question - “About how fast were the cars going when they X into each other” It suggests that eyewitness reports are susceptible to people rationalising and confabulating their memories. However, a shortcoming of the experiment was the fact it was held in a lab. It lacked ecological validity as behaviour may not have been spontaneous nor would the participants have been in the same emotional states. Further, psychology student were used where they could have guessed the aim. They could have shown demand characteristics to please the experimenter. As witnessing a real crime is probably more stressful than taking part in an experiment, memory accuracy may well be even more affected in real life. Clifford and Scott (1978) improved on Loftus and Palmer which didn’t look into stress as a factor, since the participants watching a safety video wouldn’t have felt the same levels of stress. They found that people who saw the stressful film remembered less. It highlights how verdicts shouldn’t be based on eyewitness testimonies because the witnesses are very likely to forget important details. However, studies outside of the laboratory have found that people who had been in high-anxiety, real life situations produce more accurate and detailed EWT. Yuille and Cutshall (1986) looked into witnesses of a real life incident. Recall was found to be accurate and two misleading questions had no effect on recall accuracy, even after a long time. Yuille and Cutshall’s experiment had ecological validity as it’s a field study (in a real environment and a real situation), meaning the behaviour was more likely to be spontaneous and natural. The researchers also replicated the police interviews to a great extent and did achieve similar results which suggests that it had reliability. Yuille and Cutshall were critiqued, however, because the participants who experienced the highest levels of stress were actually closer to the event. The IV becomes how close they were to the event (which was what enabled them to remember) rather than levels of stress; construct validity becomes an issue. There is also a limited reliability due to the small sample size of just 13. The study does rely on on witness self reports which means that it may have been subject to lying or social desirability. It suggest that eyewitness testimony can be reliable if the event isn’t held in a laboratory experiment. The study supports the use of cognitive interviews as they include open questioning which allows more details to be drawn from witnesses.