The effect of certain variables on the reliability of eyewitness testimony
The single most important determinant of whether a case will be solved in a court of law is often the information supplied by the witness or victim of a crime (Yuille & Cutshall, 1986). Eyewitness testimony is viewed as powerful in courts, due to a belief in the ultimate accuracy of observation and human memory (Bartol & Bartol, 1994). Despite the credibility ascribed to eyewitness testimony by law, Shaw and Skolnick (1994) indicate that studies of human perception and memory have demonstrated that such testimony is susceptible to errors. A distinction is currently being made between two kinds of variables that impact on eyewitness testimony (Wells, 1978; Wells & Olson, 2003). Estimator variables, over which the criminal justice system exerts little or no control, are classified into witness factors (e.g. stress experienced by the witness during the event, as well as the occupation, age, gender, race, and individual differences of witnesses), target factors (e.g. the suspect’s physical appearance and weapon focus) and situational factors (e.g. variables associated with the incident, such as the severity of the incident and a witness’s opportunity to view the suspect). System variables, on the other hand, refer to factors that are directly under the control of the criminal justice system and can be altered to increase accuracy in court cases. These factors include the number of foils in a line-up, the selection of line-up members, post-event information, questioning techniques and court proceedings (Narby, Cutler & Penrod, 1996; Wingfield & Byrnes, 1981). Although the impact of the factors that can potentially influence eyewitness testimony appears to be generally recognised, it also seems to be true that in-depth knowledge concerning aspects such as the interaction between these factors is often, if not mostly, lacking in those professions (e.g. the legal profession and the police force) where such knowledge could play a significant role. The purpose of this study was to investigate the reliability of perception and memory and more specifically the variables that may affect the reliability of eyewitness testimony. The thesis consists of eight articles, namely three literature review articles and five empirical articles. The findings of these empirical articles are presented next. The first empirical study, which was aimed at investigating the influence of certain variables on the accuracy of eyewitness testimony after exposure to a real-life incident, yielded the following main findings: - There were no significant differences between the average accuracy scores of the short-term and long-term memory groups as a whole. - The short-term memory average scores of the students and the 18 to 25-year-olds were significantly better than those of the corresponding long-term memory group. No significant differences, however, had been found for the gender and racial groups. - The children (12 to 14-year-olds) obtained significantly higher average accuracy scores than the adults with regard to long-term memory. In respect of long-term memory, the scholars also performed significantly better than all other occupational groups, except the members of the police. The second study (which is linked to the first) aimed at investigating whether the analysis of items would provide more clarity on the relationship between certain biographical variables and the accuracy of eyewitness testimony, yielded the following main findings: - As a whole, the short-term and long-term memory groups showed no significant differences regarding details related to less pronounced physiognomic details. Significant differences were indicated for only certain items related to obvious physiognomic details, clothing details, circumstantial evidence and action-related details. - Regarding occupation, significant differences in the short-term were found for some of the items related to obvious physiognomy, less pronounced physiognomy, clothing, action, and circumstantial evidence. However, for the long-term groups no significant differences were found for less pronounced physiognomic details and circumstantial evidence. - The age group findings indicated significant differences in the short-term group concerning certain items related to obvious physiognomy, less pronounced physiognomy, clothing, action and circumstantial evidence. In the long-term age groups, no significant differences were found for items related to obvious and less pronounced physiognomy and circumstantial evidence. - With respect to gender, no significant differences regarding action-related and circumstantial evidence were found. However, significant differences were found regarding certain obvious physiognomic details, less pronounced physiognomic details and clothing details. - Regarding the different races in the short-term memory group, no significant differences were found regarding items related to action, less pronounced physiognomy and clothing. However, significant differences were indicated in certain items related to obvious physiognomic details and circumstantial evidence. Although no significant differences were found in any clothing details in the long-term memory group, certain items related to obvious physiognomy, less pronounced physiognomy, action and circumstantial evidence displayed significant differences. The main findings of the third study, which investigated memory performance after exposure to a violent and a non-violent incident, may be summarised as follows: - Regarding both the non-violent and violent video a significant decrease in the accuracy of the average scores occurred after a five to six week retention period when the short-term and long-term memory groups were compared as a whole. - With respect to all the biographical groups that were exposed to the non-violent and violent video, significantly higher average scores were obtained by the short-term memory groups than by the long-term memory groups. - The 18 to 25-year-olds who viewed the non-violent video obtained significantly higher short-term and long-term average memory scores than the 26 years and older. No significant differences were found for the violent video. - No significant differences were indicated for either the short-term or long-term memory gender groups regarding both the non-violent and violent video. - Concerning the two racial groups, the only difference found was for the non-violent video where the whites obtained significantly higher short-term memory average scores than the blacks. No significant differences were indicated for the long-term groups or those groups who watched the violent video. The aim of the fourth study to investigate the relationship between the accuracy of responses and the method of questioning and the relationship between the method of questioning and the accuracy of responses of specific biographical groups, provided the following main findings: - A significant relationship exists between accuracy of memory and type of questioning. The average accuracy scores of the participants who responded to the closed-ended questions were significantly more accurate than those who answered open-ended questions. - Investigating the relationship between the average accuracy scores of biographical variables and type of questioning revealed that all groups except the public were significantly more accurate in responding to the closed-ended than open-ended questions. - The scholars obtained the lowest average accuracy scores compared with the other occupational groups in responding to both the open-ended and closed-ended questions. Their average scores were, however, not always significantly lower than those of the other groups. Regarding the open-ended questions the public obtained a significantly higher score than all the other occupational groups. - The 18 to 25-year-olds obtained a significantly higher average accuracy score than the other age groups with respect to the closed-ended questions. No significant difference was indicated for the open-ended questions in relation to age. - Regarding gender, no significant differences on either open-ended or closed-ended questions were indicated. - With respect to race, the whites performed significantly better than the blacks when responding to both open-ended and closed-ended questions. In the fifth study the relationship between confidence and accuracy, the method of questioning and accuracy, and the method of questioning and confidence were investigated. Firstly, regarding the relationship between confidence and accuracy of recall it was found that: - there was a significant correlation between the accuracy of memory and confidence regarding more than 70% of the items. - for 60% of the items there was a significantly larger proportion of participants who were very certain and responded accurately to their responses. - for more clothing and circumstantial details (75%) there was a significant relationship between accuracy and confidence than for central details (57%). Secondly, the findings for the relationship between type of questioning and accuracy were: - A significant relationship was found for 67% of the total number of items. - Ninety per cent of the items that displayed a significant relationship between type of questioning and accuracy, involved closed-ended questions. This means there was a significant higher degree of accuracy regarding the responses to closed-ended questions in comparison to open-ended questions. - With regard to the respective details, 70% of the total number of details that indicated a significant relationship between type of questioning and accuracy, concerned clothing and circumstantial details. Thirdly, the relationship between type of questioning and confidence indicated that: - there was a significant relationship in for 53% of the items. - all of the items which indicated a significant relationship concerned closed-ended questions. Therefore there was a significantly higher degree of confidence regarding the responses to closed-ended questions than to the open-ended questions. - with respect to the details that displayed a significant relationship between confidence and type of questioning 50% involved circumstantial and clothing details and 50% concerned central details. However, with regard to the items that displayed a significant relationship between confidence and type of questioning, the respondents were very certain about 67% of clothing and circumstantial details. These results indicate that estimator and system variables need to be further researched in the South African context, as most existing findings are based on American and European research. Doing research on eyewitness testimony remains difficult owing to ethical concerns and the question as to whether laboratory studies have the same impact as real-life studies on memory performance. To make an impact in the area of eyewitness testimony in coming up with results which are less affected by confounding variables, it would not only be important for researchers in this field to reconstruct events which have the qualities of real-life incidents, but also to ensure that the measuring instruments correspond with the way eyewitnesses are questioned in legal situations.