There are factors that affect the accuracy of eyewitness testimony such as emotions, fundamental attribution bias, face recognition in other races, leading questions and many more. An example of the affect factors such as leading questions can have on eyewitness testimonies is the Loftus and Palmed study (1974).
It's has been proposed that we store a series of incomplete memory fragments in our mind. When we need to recall a memory we unknowingly fill in the blanks to reconstruct a memory that can be fraught with inaccuracies. Our memory is shaped by our own beliefs of what has happened in the past which may not be an accurate depiction of events. When we have a incomplete memory or are faced with something unknown to us we use our memory of past experiences to create 'schemas'. For example in Carmichael et al (1932) study they showed two groups of participants the same set of drawings but each group was given a different set of descriptions. When the participants were asked to recall a drawing, the way in which the drawing was previously described affected the drawing that was subsequently produced by the participant. This shows that the language used affected our memory and conjures up a set of expectations about the object - schemas. Such schemas alter our recollection of something and produces an inaccurate memory. Another form of reconstructive memory is stereotypes. A stereotype is when we attribute inaccurate characteristics to a group of people e.g. all blondes are dim or all youths are yobs. This could affect our memory recall when it comes to remembering an incident. For example when remembering a fight between two people we may be inclined to believe the youth was being violent went in actual fact it was an adult who was being violent.
There has been debate over whether our initial perceptions are effected by schemas or whether it is our subsequent recall that is effected.