line - ups or photo arrays produce substantial differences in identifi cation per-
formance ’ ( 1994 : 181).
232
Handbook of Psychology of Investigative Interviewing
Fairness of v ideo l ine - u ps
The research available suggests that video identifi cation is likely to be no more
or less sensitive than a live line - up. However, the number of video clips avail-
able in a database to construct video line - ups is much greater than the number
of volunteers available to stand on a live line - up. Therefore, there is good
reason to believe that in an operational context it should be possible to con-
struct video line - ups that are fairer than live line - ups. Indeed, research has
found that video line - ups from criminal cases were fairer to the suspects than
conventional ‘ live ’ line - ups (Valentine & Heaton, 1999 ). Furthermore, video
line - ups were equally fair to white European and African – Caribbean suspects
(Valentine, Harris, Colom Piera & Darling, 2003 ). In these studies, partici-
pants ( ‘ mock witnesses ’ ) were given the fi rst description of the offender pro-
vided by the original witness and were required to select the line - up member
whom they thought was most likely to be the suspect. As a mock witness has
not seen the perpetrator, the suspect should be chosen no more often than
predicted by chance if the line - up is perfectly fair: 11% (one in nine) of choices
should be of the suspect from line - ups containing eight foils. Valentine &
Heaton (1999) found that the mock witnesses identifi ed the suspect in live
line - ups more frequently (25%) than by chance, but were not able to select
the suspect from video line - ups (15%) signifi cantly more often than chance.
Valentine et al. (2003) found that video line - ups of African – Caribbeans and
of white Europeans were equally fair, using equal numbers of mock witness
from both ethnic backgrounds.
Perception of i dentifi cation o ffi cers
Hughes (2005) examined the opinions of 30 experienced police identifi cation
offi cers of video identifi cation procedures. Almost all identifi cation offi cers
regarded video identifi cation as a better method than a live line - up. The point
was made that it facilitated use of identifi cation evidence for volume crime
rather than just serious crime. Interestingly, 37% of identifi cation offi cers
thought identifi cation evidence was wholly reliable. This fi gure increases to 51%
if witness confi dence is high and the procedure is held shortly after the crime.
The majority of offi cers (66%) thought that video identifi cation is fair to both
suspect and witness; nobody thought it unfair to the suspect, but 33% thought
it unfair to the witness. The concern was that aspects of the procedure made
the task of identifi cation too diffi cult. There was a clear view (90%) that more
should be done to brief and support the witness prior to the procedure.
Evaluation of r evised p rocedures for v ideo l ine - u ps
Valentine, Darling
& Memon
(2007) tested whether adopting the strict
sequential presentation rules described above would enhance the reliability of
video identifi cation evidence. The strict sequential presentation procedure, by
Recent Developments in Eyewitness Identifi cation Procedures
233
which the witness was instructed to make a response to each face as it was
presented, was compared to the British procedure in which witnesses are told
to watch the whole video twice before making a decision. The line - up admin-
istrator was blind to the position of the suspect in a nine - person line - up under
both the ‘ strict sequential ’ and the ‘ existing ’ conditions. There was no reliable
effect of the presentation procedure on the number of mistaken identifi cations
from perpetrator - absent line - ups, but there were fewer correct identifi cations
of the perpetrator when he was present in the line - up under the ‘ strict sequen-
tial ’ condition.
Darling, Valentine & Memon (2008) tested the use of a culprit - description
strategy instead of suspect - resemblance strategy to select foils for a nine - person
video line - up. It was established, from pairwise ratings of similarity provided
by participants who did not take part in the main study, that the foils in the
culprit - description line - ups were more dissimilar to each other than were the
foils in the suspect resemblance line - ups. Nevertheless, there was no statistically
signifi cant difference in the rate of correct or mistaken identifi cation between
culprit - description and suspect - resemblance line - ups. The experiments reported
by Valentine et al. (2007) and Darling et al. (2008) followed British police
video identifi cation procedures and used line - ups constructed from a police
national database. These data explicitly compared procedures advocated in the
research literature with existing procedures in an operational context in the
UK. There was no evidence that either the sequential presentation instructions
or the culprit - description strategy would improve the existing video identifi ca-
tion procedures.
Suspects with d istinguishing m arks
How can a fair line
- up be constructed if a suspect has a distinguishing
mark, for example a tattoo, scar or distinctive mark on their face? A witness may
have described a mark or tattoo and the suspect may have been arrested
because they have a similar mark. Alternatively, the suspect may have a mark
that was not described by a witness. If the suspect is the only person in the line -
up with a scar or tattoo on his face, he will stand out in comparison to the foils,
rendering the line - up unfair. In the UK the suspect ’ s solicitor is likely to object
to a line - up in which his or her client is the only person with a scar or tattoo.
There have been two solutions to this problem. Most frequently the area
of the distinguishing feature is masked on the face of the suspect and the foils.
For video identifi cation the area is pixelated, using a mosaic of squares in
a grid that have an average colour and luminance to occlude the area.
This process can be automated in moving video so that the mosaic occludes
the same area of the face as the view changes. The alternative strategy is to
replicate the distinguishing mark on the faces of the foils. This process is time -
consuming and expensive, so the mark is usually replicated only on full - face
and profi le views. Under these circumstances still images are shown to the
witness instead of moving images.
234
Handbook of Psychology of Investigative Interviewing
Is it better to mask or replicate? The disadvantage of masking a distinguish-
ing feature is that it changes the appearance of the suspect ’ s face. A distinctive
feature will be a salient cue to recognizing the face of a perpetrator (Winograd,
1981 ; Valentine, 1991 ). Therefore, a better strategy is to replicate the feature
on the foils. If replication is used, the identical feature is normally replicated.
However, this strategy means that the witness cannot use their knowledge of
the distinguishing mark to
recognize the perpetrator. A better approach would
be to use the culprit - description strategy described above. The feature repli-
cated on each face should be consistent with the witness ’ s description, but the
precise properties (location, size, colour or style) may vary across foils within
the constraints set by the description. This variation would allow the witness
to use recognition of a distinguishing mark to support recognition of the
perpetrator. However, variation in the characteristics of the distinguishing
mark would not introduce any bias against an innocent suspect whom the
witness has not seen before.
As replication can only be implemented on still images, the question arises
of whether use of still images rather than moving video would impair identi-
fi cation accuracy. The effectiveness of moving video and a single, still, full - face
image was compared in an otherwise identical video identifi cation procedure
by Valentine et al. (2007) and Darling et al. (2008) . No effect on identifi ca-
tion of perpetrators from culprit - present line - ups was observed. A foil was less
likely to be mistakenly identifi ed from culprit - absent line - up when moving
images were used in one experiment (Valentine et al. , 2007 ), but the effect
was not replicated in a second experiment (Darling et al. , 2008 ). The data
suggest that use of moving images has little if any reliable infl uence on the
outcome of a line - up.
Eyewitness c onfi dence and a ccuracy
It has been appreciated for a long time that a confi dent witness may be mis-
taken (Devlin,
1976 ). Indeed, a caution to this effect is included in the
Turnbull judgement on eyewitness evidence. In recent years it has been dem-
onstrated that the relationship between the confi dence of an eyewitness and
the accuracy of their identifi cation is moderately strong for witnesses who
identify somebody at a line - up. However, the relationship is weaker amongst
witnesses who reject the line
- up (Sporer, Penrod, Read
& Cutler,
1995 ).
Confi dence is most closely associated with accuracy when measured immedi-
ately after an identifi cation has been made (Cutler & Penrod, 1989 ), and
critically before the witness acquires any further information about their iden-
tifi cation. However, as the association between confi dence and accuracy is far
from perfect, confi dent but mistaken eyewitnesses will be encountered fairly
frequently.
Recent Developments in Eyewitness Identifi cation Procedures
235
Malleability of w itness c onfi dence
A very important research fi nding is that witness confi dence is changeable
and is infl uenced by information that the witness acquires after attending an
identifi cation procedure. Receiving feedback that the person identifi ed is the
suspect, or that somebody else made the same identifi cation, will increase the
witness ’ s confi dence in their identifi cation. Not only does confi rming feedback
tend to make the witness subsequently more confi dent in their identifi cation,
but it also tends to infl ate estimates of a range of subsequent testimony, includ-
ing how long the culprit was seen for, how close they were and how much
attention the witness paid (Wells & Bradfi eld, 1998 ). Furthermore, confi rming
post - identifi cation feedback tends to make eyewitnesses overconfi dent; that is,
they now express more confi dence in their identifi cation than is warranted
(Semmler, Brewer & Wells, 2004 ). By the time a witness gives evidence in
court they are likely to have received confi rming feedback. Therefore, the
confi dence a witness displays in court may well be determined by the feedback
they have received in addition to their initial confi dence at the identity
procedure.
The c onfi dence of r eal w itnesses
Wright & Skagerberg (2007) took advantage of the practice of Sussex Police
to routinely inform witnesses at a video line - up whether they had identifi ed
the suspect. After the witness had made an identifi cation but before they
received feedback each witness provided a rating on a 10 - point scale for three
questions, one each about their opportunity to view the culprit, the identifi ca-
tion they had just made and how good they believed their general memory to
be. After receiving feedback, each witness provided a rating on three questions,
and again one question addressed each of the same three aspects. Witnesses
evaluated the identifi cation task as more diffi cult after feedback that their
identifi cation was mistaken, but witnesses who were told that they had identi-
fi ed the suspect evaluated the task as easier. Wright & Skegerberg ( ibid .) make
an important contribution by demonstrating that malleability of confi dence
previously observed in laboratory studies is also observed in real witnesses and
victims of crime. There is no provision in the current PACE code of practice
to record a statement of confi dence, although anything the witness does say
must be written down. In contrast, the US Department of Justice guide on
eyewitness identifi cation does recommend taking a clear statement of confi -
dence immediately after the witness makes an identifi cation and before any
feedback is given (Technical Working Group for Eyewitness Identifi cation,
1999 ).
The code of practice for England and Wales requires that witnesses are
instructed: ‘ If you cannot make a positive identifi cation, you should say so. ’
Hughes ’ (2005) survey showed that the meaning of the word ‘ positive ’ was
236
Handbook of Psychology of Investigative Interviewing
ambiguous even amongst identifi cation offi cers. Interpretation ranged from
‘ any identifi cation ’ to ‘ a defi nite 100% ’ confi dence. Only 7% of identifi cation
offi cers said that they ask the eyewitness about their confi dence in their iden-
tifi cation before they leave the ID suite. Ninety - three per cent of offi cers know
the position of the suspect in the video line - up (i.e., the identifi cation proce-
dure is not run blind). The majority (70%) did not think that blind testing
would enhance justice. Nineteen per cent of police forces have a policy of
telling witnesses whether they identifi ed the suspect. One reason given for this
policy is to enhance witness care as most witnesses want to know if they have
identifi ed the suspect. Unfortunately, only 36% of respondents who inform
the witness of the outcome of an identifi cation procedure take a statement
before providing the feedback. Making a written record of confi dence before
providing feedback is essential to preserve an accurate record of confi dence at
the time of identifi cation. Prior to trial a witness is likely to receive or deduce
feedback that will bolster their confi dence.
Conclusions and r ecommendations
The evidence is very clear that mistaken identifi cation is the leading cause of
wrongful conviction. Furthermore, wrongful conviction is a major problem
– there have now been more than 230 DNA exonerations in the USA. Whilst
this evidence relates specifi cally to that country, there is good cause to
believe
that a similar problem exists in the UK. Approximately 20% of eyewitnesses in
the UK make a known mistaken identifi cation. However, a mistaken identifi ca-
tion of a foil in a line - up will not lead to a wrongful conviction. It is the
unknown mistaken identifi cation of the police suspect that leads to a miscar-
riage of justice. The legal system in the UK is very different from that in the
USA. Identifi cation procedures in England and Wales are regulated by the
code of practice (code D) required by the Police and Criminal Evidence Act.
In addition, a special warning about the frailty of eyewitness identifi cation must
be given to the jury by the trial judge. Nevertheless, uncorroborated eyewit-
ness identifi cation evidence remains suffi cient to secure a conviction in England
and Wales.
The PACE code of practice does include a number of elements of best
practice which are not often discussed in the research literature on eyewitness
identifi cation. The suspect ’ s legal representative has the right to be present
when a witness views a video or live line - up. If no representative is present,
the procedure must be videotaped. This provision has the potential to be an
effective protection of the suspect ’ s rights and to discourage overtly leading
behaviour. A further provision that allows the suspect or their legal representa-
tive the opportunity to object to line - up members provides a practical means
to improve the selection of foils, which is likely to enhance the fairness of
line - ups.
Recent Developments in Eyewitness Identifi cation Procedures
237
Although the PACE code of practice is extensive, there are a number of
areas in which practice could be improved. As virtually all identifi cation pro-
cedures are run on video double - blind testing, in which the line - up administra-
tor does not know who the suspect is, can be easily implemented (see Valentine,
2006 , for a practical suggestion of implementing this provision). Blind testing
would enhance the integrity of identifi cation procedures and safeguard the
police against accusations of malpractice. The code of practice should be
amended to include provision to take a clear statement of confi dence imme-
Handbook of Psychology of Investigative Interviewing: Current Developments and Future Directions Page 45