Participants were then asked to selected a verdict (guilty or not guilty) and rated their confidence in that verdict on a scale from 0 to 10. Participants also rated the witness's believability, trust, and honesty using three 7-point scales.
The results found that when both witnesses were correct with the evidence given it was the more confident witness who the participants considered to carry the most conviction. However, when it was found that the evidence given was in fact wrong the roles reversed and it was the unconfident witness who ended up being found to be more credible.
The second experiment saw the case changed from a criminal case to a civil one and the two witnesses were pitted against each other instead. One hundred and three undergraduate participants read testimonies from two conflicting witnesses and were asked to rate the credibility of each before it was found that some of the information was wrong. The confident witness was confident about both the central and peripheral details of the case whereas the unconfident witness remained confident on the central details however displayed unconfident behavior when it came to the peripheral details. When it was revealed that both had supported erroneous facts regarding the peripheral details it was the unconfident witness who remained the most credible in the eyes of the participants (who acted as jurors). However before it was revealed that some of the testimony was false on both accounts the participants believed the confident witness the most.
Elizabeth R. Tenney, Robert J. MacCoun, Barbara A. Spellman and Reid Hastie found that jurors felt that the unconfident witnesses knew there were limitations in their own knowledge and were therefore well calibrated so when they were found to be wrong about the evidence they were unsure on, it was easier to believe that they were correct on the evidence they were confident on. This is in contrast to the high confident witnesses who were wrong on the evidence they were confident on and are therefore poorly calibrated and are likely to be wrong on further evidence that they are also confident about. Jurors use the confidence as an inference into how well the witnesses are calibrated.
During this experiment Elizabeth R. Tenney, Robert J. MacCoun, Barbara A. Spellman and Reid Hastie only used the same 2 ‘witnesses’ throughout. This is a potential drawback as the two witnesses were not a decent representation of everyone who stands in front of a juror and may have displayed different traits in behaviour to others should they be in their position. For example the witness who was asked to be confident may have displayed this behaviour in such a way that they appeared cocky which could have led to the jurors being less likely to believe them due to their first impression.
A series of experiments by Princeton psychologists Janine Willis and Alexander Todorov have shown that all it takes is a tenth of a second to form an impression of a stranger from their face, and that longer exposures don’t significantly alter those impressions. Elizabeth R. Tenney, Robert J. MacCoun, Barbara A. Spellman and Reid Hastie could have used more witnesses during the experiment whilst repeating it in order to get a fairer representation of the population.
The fact that the experiment was not real and the participants knew that would also have led to certain demand characteristics and without the thought that there are real consequences the participants may not have taken it as seriously and simply tried to conform to what they believed the experimenter was trying to find.
References
Tenney, E. R., MacCoun, R. J., Spellman, A. B. & Hastie, R.,
Calibration Trumps Confidence as a Basis for Witness Credibility
Teabag, J. R. (1999). Effect of Winding-up Duration on Jaw Clenching
Willis, J., & Todorov. A,
First Impressions Making Up Your Mind After a 100-Ms Exposure to a Face