This is a critical review of Pamela Coxon and Tim Valentine’s (1997) study on “The effects of the age of eyewitnesses on the accuracy and suggestibility of their testimony”. I intend to critically evaluate the research carried out by Coxon and Valentine in terms of its contribution to the area of witness evidence, by examing its design, methodology and findings. The rationale behind the study was that everyday life requires us to retrieve information that may have been encoded some time before.
Some information seems easier to recollect than others. One might assume that it would be a relatively easy task to recollect a crime witnessed. At the time of encoding you may be aware that what you are witnessing is important, and so retrieving the information at a later time wouldn’t be a problem. But, this is not always the case, and many factors can influence the accuracy and reliability of eyewitness evidence. The way questions are phrased can have an enormous influence on witness responses as can the age of the witness.
Young children tend to remember less than older children or adults (Ceci & Bruck, 1993, cited in Westcott and Brace, 2002 p138). Research has shown that younger children seem to be more susceptible to suggestibility and therefore leading questions should be avoided during interviews (Ceci et al, 2001, cited in Westcott and Brace, 2002 p139). The experiment employed by Coxon and Valentine had a factorial design that was between participants. The participants were divided into three groups aged 7-9 (children), 16-19 (young adults) and 60-85 (elderly).
All participants watched a short video. They answered seventeen questions based on the video, four of which were misleading for those in the experimental condition (where misleading questions were introduced). The participants’ accuracy was assessed by the number of correct and incorrect answers given to the non-critical (non-misleading) questions asked. Suggestibility was assessed by the number of the four questions in which participants gave misleading information.
The design stated clearly that it was employing a factorial design and that it was a between participants experiment. The number of participants in each group was well balanced, but there appeared to be an inbalance in the number of females and males in each group. There were 95 females to 52 males overall. (Coxon and Valentine, 1997, p7). While the procedure used by Coxon and Valentine (1997) was well written and clearly described, there were some flaws and omissions in the design of the experiment which I found to be highly problematic.
Firstly, there was no detailed description given of the Independent Variables being used for this design. Nor did it say what the conditions were. It simply lists two “between-subject factors – age and experimental condition”. Likewise, the Dependent Variable was not stated explicitly, but it was described how it would be measured. There are two major omissions in the design of this experiment. Firstly, there was no description of any controls used in order to improve the experiment. Whether they were simply not listed, or whether there were none taken, is unclear.
Secondly, and more importantly, the study as outlined in the report would be totally unreplicable. There is no detailed information or script provided to inform the reader what exactly happened in the video. Furthermore, there is no list of the questions that were verbally given to the participants and no information on how these questions were constructed. Further questions must be raised about the appropriateness of the stimulus materials used. The video chosen consisted of a kidnapping of a newly born baby from a hospital ward.
Coxon and Valentine quote previous research from Goodmann et al (1990) (cited in Westcott and Brace, 2002 p140) as their reason for choosing this topic, saying that fear of separation from loved ones is a central concern for children. The length of the video the participants were asked to watch was was three minutes; a good choice especially for the younger participants as they may have had trouble concentrating for longer. We are not told how long the questioning after the video took, which may have had a negative impact on the findings of the group of young children. The next section of the article presented the findings of the experiment.
It was found that both the elderly and child participants gave fewer correct answers and more incorrect answers to non-misleading questions than did young adults. Children were found to be more susceptible to suggestibility than either the elderly or young adults. There was no relationship found between the accuracy of recall and suggestibility. We are told that when the data was collected, there was no significant difference between each age group to the non-critical questions asked. The data was therefore combined, using the experimental condition as the Independent Variable.
A one-way ANOVA was used. We are not, however, given any information on the original data collected to witness for ourselves the lack of significant difference between the age groups. Despite this, the remainder of the results were well presented and the differences in responses to the misleading questions may be clearly observed in the tables provided. Coxon and Valentine, in the interpretation of the results, say that their findings have demonstrated that young children and elderly people “perform less accurately than young adults on questions about an event viewed on video” (p. 14).
As we have noticed earlier, however, due to the lack of information on the video and system variables such as questions asked verbally, we have no way of knowing whether these results are accurate or have been influenced by the researcher. Likewise, there was quite a large inconsistency between the groups regarding the proportion of males to females. As yet, studies have been inconsistent as to whether the sex of a witness may have some bearing on the reliability of their ‘testimony’ (Wescott & Brace, 2002), but this may, also, have some bearing on the outcome of the results.