Effects of biased versus unbiased instructions on eyewitness identification accuracy

Eyewitness Identification Evidence

Police line ups, biased / unbiased instructions, confidence

Testimonies and identification of suspects by eyewitnesses is a critical component of criminal investigation and proceedings in court. The confidence of the eyewitness in a testimony is extremely important for investigators and jury. Eyewitness identification evidence, however, relies on memory, which is limited by internal and external influences. This study aimed at determining the eyewitness identification confidence and false identification rate of students at a university who were shown a robbery video and then asked to answer questions based on an easy or difficult memory test. The findings show that when provided with unbiased instructions, false identification rates are much lower. Eyewitness identification confidence was also higher with unbiased instructions. However, there is no effect of the difficulty of the memory test on false identification rates. In administering police lineups, it is important to consider using unbiased instructions in order to reduce the risk of convicting an innocent individual.

Eyewitness identification evidence

Introduction

Eyewitness identification evidence provided through police lineups is given considerable weight during court proceedings and criminal investigations. In cases where there is no other evidence, eyewitness identification, and testimonies act as the sole source of evidence in convicting criminals. However, often there are identification errors that occur as a result of this procedure and these are driven by several factors. These factors are divided into three major groups which are event-inherent factors, psychological issues affecting eyewitness identification such as recall and system variable which are can be controlled by the criminal justice system. This research focuses on two system variables of eyewitness identification -- identification confidence and false identification rates. These will be determined when the eyewitnesses are administered a difficult and easy memory test and when they are provided with biased or unbiased instructions in thief-present and thief-absent lineups.

Literature review

Human eyewitness identification relies majorly on memory. This makes it prone to mistakes because even when people express confidence in their memories, they often interpret their memory in different ways and often it is impossible to get a second account of the same event R.S. Malpass & Devine, 1981.

As stated by Wells, Malpass, Lindsay, Fisher, & Turtle, 2000(Wells and Loftus (2003)

, memory can be altered by external and internal influences and they fade as time goes. Witnesses also become uncertain or confused if an event is traumatic or unexpected leading to distortion of recall of events )

Forensic DNA testing has also shown the inherent problems in eyewitness identification because many of the exonerations that have been done after post-conviction DNA testing were as a result of the person being convicted after eyewitness testimony. This suggests that there is need to find ways to mitigate the imperfections of human memory that manifest in eyewitness identification. Wells et al. (1998)

suggest that three out of every four fall convictions are as a result of the imperfections of eyewitness identification which presents a strong case for Over the last 35 years, a lot of research has been conducted on flaws related to eyewitness identification and these have sadly undermined the accuracy, reliability, and validity of eyewitness identification. A huge wealth of knowledge on this subject has been amassed by scientists through research and they have been able to determine variables which affect eyewitness identifications and provided recommendations on how identification can be improved to reduce these errors.

Researchers have also greatly debated the applicability of research conducted in laboratory settings in heavily controlled environments to the real-life scenario founded on gross imperfections and unpredictability compared to the laboratory or research settings.

Police lineups

Most of the research on identification by eyewitnesses has focused on police lineups of live persons or photographs. In examining the procedures used in police lineups of photographs or individuals, researchers have developed recommendations for improving these lineups such as providing unbiased instructions using blind administration procedures and using a sequential and not simultaneous presentation method Leippe & Eisenstadt, 2013.

Blind administration is whereby the officer administering the lineup is unaware of which member of the lineup is a suspect. This leads to unbiased instructions when taking the test since they do not want to risk giving wrong instructions. Studies have shown that these blind procedures reduce influences of the administrator and lead to better identification of suspects Haw & Fisher, 2004(; Phillips, McAuliff, Kovera, & Cutler, 1999)

Research also suggests sequential and not simultaneous lineups may be better at identifying the suspects. This is because when the suspects are lined up simultaneously, there is a tendency to promote relative judgments where the witness compares all members of the lineup to see the one that closely resembles the culprit in their memory Wells, 1984.

This results in a higher rate of false identification of members of the lineup even when the culprit is not present

ADDIN EN.CITE

(Cutler & Penrod, 1988; Wells, 1984, 1993)

. Research has also shown that when simultaneous lineups are used with the suspect in the lineup, they produce more identification of the culprit compared to the sequential method Steblay, Dysart, Fulero, & Lindsay, 2001.

This is also supported by the result of a meta-analysis of research on simultaneous vs. sequential identification procedures, it was found that more correct identifications of the culprit were achieved using simultaneous lineups compared to sequential lineups Steblay et al., 2001.

R.S. Malpass, Tredoux, and McQuiston-Surrett (2009)

also showed that when there was an innocent suspect and the perpetrator in the lineup and the two resembled each other, simultaneous lineups were better at identifying the culprit compared to sequential ones.

The results of research on sequential lineup are also interesting, they show that sequential lineups reduce the rate of identifying the culprit falsely when the culprit is not present in the lineup while when the culprit is present, they have a lower accuracy rate Steblay, 1997.

This is because the eyewitness is presented with each individual member of the lineup alone and asked to make a decision whether the person was the culprit before they are allowed to move to the next person. However, research also shows that sequential lineups are ineffective when the witness is a child Steblay et al., 2001()

The findings described above are from laboratory settings where most factors are controlled. Few studies have attempted to check whether these findings are generalizable to the actual investigation setting. These field studies involved actual cases so they only use the words suspect and filler. The suspect is defined as the person that the police believe committed the crime. In these situations, the word culprit is avoided because the person is lawfully innocent until proven guilty. A study conducted in 2006 showed that sequential procedures when used for blind photo arrays were logistically possible and accurately identified suspects better Klobuchar, Steblay, & Caligiuri, 2006.

Another study in the same year showed that sequential, blind lineups had more identification of fillers than the other simultaneous, non-blind lineups. These results were disputed by many researchers since they were contrary to other studies. They were thought to be inherently flawed with study design failures. Another field study conducted by the New York Police Department found that simultaneous, non-blind lineups have similar identification rates to sequential, blind lineups.

A field study conducted in 2011 in four police departments, Charlotte-Mecklenburg, Tucson, San Diego and Austin Police Departments comparing blind sequential and simultaneous procedures found that eyewitness identification rates for both procedures was the same. These results can be attributed to the blinding that was used, which allowed for direct comparison of the two lineup procedures. However, there were statistically significant differences in how the eyewitnesses identified filler photographs incorrectly. The field study found that during simultaneous lineups, eyewitnesses were more likely to choose fillers wrongly than in a sequential lineup. This study, therefore, concluded that sequential lineups reduced the known errors in photo lineups significantly. This results should be interpreted with caution because even these sequential lineups led to a considerable percentage of false filler identification with more than 12% Wells, Steblay, & Dysart, 2011()

Wells et al. (2011)Grolund, Andersen, and Perry (2013)

compared the results of the field study to others done previously and stated that there is still a lot that needs to be researched in this field. The researchers are currently working on research on the potential implication of suspect placement in sequential lineups to determine whether the suspect is easier picked out correctly when they are viewed first or later in the sequence.

The use of open ended instructions for eyewitnesses has also been researched and recommended. Steblay, 1997(Steblay (1997)

also suggests for the instructions provided to the eyewitness to include a statement that the perpetrator may or may not be in the lineup. This finding is from a meta-analysis that shows that an unbiased instruction that the culprit may or may not be present, reduces mistaken identification in culprit-absent lineups by over 40% Wells, Rydell, and Seelau (1993)

state that it is important to add fillers who meet the description provided by the eyewitness. )

. Another meta-analysis conducted in 2013 supported these findings and showed that unbiased instructions reduced the identification errors in culprit-absent lineups by about 39% while in culprit-present situations, identification errors were reduced by over 50% after giving unbiased instructions.

Methodology

College students in a class were randomly assigned to an easy or difficult memory test that had either a target-present or target-absent lineup Leippe & Eisenstadt, 2013.

They were shown a video of a staged robbery then were taken through an unrelated lecture for about 30 minutes. They were then administered with a paper-and-pencil memory test of the crime video that asked them to circle the member of the lineup they thought was the culprit. The test included free-recall and directed-recall questions. The difficult questions required the participants to provide as much detail as they could remember about the culprit including their weight, height, hair and skin color, clothing and age. Questions surrounding the scene focused on the subtle details. Participants who were randomized to the easy questions were required to describe the thief's appearance in brief. Their other questions focused on obvious details of the scene. Participants were then provided with a second folder of materials, which provided them with sequential pages of 6 people in a photo spread where the thief was either present or absent and asked to rate their certainty of identification on a 100-point scale. In the culprit-absent photo spread, the culprit was replaced with a look-alike.

Hypotheses

The following hypotheses were being tested in this study. The first hypothesis is that identification confidence will be higher after an easy memory test compared to a more difficult memory test. The second hypothesis is that identification confidence would be higher in the lineup with biased instructions compared to unbiased instructions. The third hypothesis is that false identification from a culprit-present lineup would be more likely with biased instructions compared to unbiased instructions. The last hypothesis that was being tested is that false identification would more likely come from an easy memory test compared to the more difficult test.

Variables

In order to test these hypotheses successfully, the following variables were defined. The independent variables were target presence as either target present or target absent, lineup instructions as either biased or unbiased, and difficulty of the memory test as either easy or difficult. The dependent variables were accuracy of identification, identification rates, false identification and confidence in identification.

Results

Confidence

The mean identification confidence for those administered the easy memory test was 54.4 compared to 63.0 for those administered the harder memory test (Table 1). This suggested that participants administered with the difficult test had higher confidence levels compared to those who received the easy test. The test of effect of difficulty of the memory test on identification confidence was not significant (F = 1.551, p = 0.218). Table 2 below shows the tests of between-subjects effects for test difficulty. Hypothesis 1 was, therefore, not supported by these findings.

Table 1: Mean identification confidence by difficulty of memory test

Table 2: Test of between-subject effects for test difficulty

The mean identification confidence for participants given unbiased instructions was higher than for those given biased instructions (59.6 compared to 57.9, table 3). However, the test of effect of type of lineup instructions as either biased or unbiased on identification confidence was not significant (F = 0.070, p = 0.793). Table 4 below shows the tests of between-subjects effects for type of instruction provided.

Table 3: Mean identification confidence based on lineup instructions

Table 4: Test of between-subjects effect for type of instruction

False identification

The mean rate of false identification was similar for unbiased and biased instructions (0.367 for unbiased versus 0.464 for biased instructions, table 5). Based on type of lineup, the false identification rate for thief-absent lineups was 0.593 compared to 0.238 for thief-present lineups (table 6). There was a significant effect of type of lineup as either thief-absent or thief-present (F = 8.480, p = 0.005), and the type of instruction as biased or unbiased on the number of false identifications (F = 5.315, p = 0.025).

Table 5: Mean rate of false identification by type of instruction

Table 6: Mean false identification rate based on type of lineup

Table 7: Test of between-subjects effect for instruction and lineup on false identification rate

The mean false identification rate for easy memory tasks was 0.34 compared to 0.50 for those administered with the hard memory test (table 8). There was no significant effect of the difficult of memory test on false identification rate (F = 1.340, p = 0.252) as seen in table 9.

Table 8: Mean false identification rate by difficulty of memory tasks

Table 9: Test of between-subjects effect on false identification rates for difficulty of memory test

Discussion

Identification of culprits by eyewitnesses is one of the most important practices used in investigations and criminal justice. This method relies majorly on the memory of witnesses and victims to enable the identification of culprits. When administered with a difficult memory test, eyewitnesses are required to give more details about their recollection of the scene and the criminal. This enables them to have greater confidence because they are able to recollect the subtle details of the event better leading to higher identification confidence levels. This supports the cue-belief model that suggests that when persons have access to better memory cues, they are able to receive internal and situational information regarding the event that helps to put together pieces from their memory to make a better picture, which elicits greater confidence. However, there was no effect of the test difficulty on false identification rates which can be thought to be as a result of the imperfections of memory and their ability to recall events not being affected by the difficulty of the memory test.

Unbiased instructions help to improve false identification rates because once the participants are informed that the culprit may not be in the lineup, they take greater caution not to convict someone wrongfully. Unbiased instructions make the eyewitnesses cautious and they make deliberate attempts not to make mistakes in choosing their culprit. They, therefore, dig deeper into their memory leading to lower rates of false identification. This shows that how the police provide instructions to the eyewitnesses helps to improve identification confidence and reduce the false identification rates.

Conclusion

There is need to exercise care when administering lineups to eyewitnesses to identify suspects. The findings of this research show that police should provide unbiased instructions to the eyewitnesses in order to elicit care in their identification and reduce false identification rates. This will reduce the number of wrongful convictions and ensure only the real culprits are convicted. There is no effect of difficulty of memory test on false identification rates, therefore, there needs to be further research conducted on this before any recommendation regarding this can be made.