Handbook of Psychology of Investigative Interviewing: Current Developments and Future Directions

Home > Other > Handbook of Psychology of Investigative Interviewing: Current Developments and Future Directions > Page 50
Handbook of Psychology of Investigative Interviewing: Current Developments and Future Directions Page 50

by Ray Bull, Tim Valentine, Dr Tom Williamson


  Target Present

  Target Absent

  Hit

  Foil ID

  Miss

  CR

  Foil ID

  Overall

  Correct

  Demo Crime/Demo Line - up

  43%

  34%

  24%

  49%

  51%

  46%

  Real Crime/Demo Line - up

  52%

  24%

  24%

  48%

  52%

  50%

  Real Crime/Real Line - up

  47%

  24%

  28%

  53%

  47%

  50%

  260

  Handbook of Psychology of Investigative Interviewing

  Sequential line - ups were associated with a marked reduction in mistaken fi ller

  identifi cations.

  More informative data comes from a study by Steblay & Dysart (2008a) ,

  who report a meta - analysis of 46 papers with 70 comparisons of sequential

  and simultaneous procedures. These studies included judgements from over

  10,000 research participants. Across 45 studies the correct identifi cation rate

  for traditional simultaneous target - present photo arrays was 49%, the mistaken

  identifi cation rate of fi llers was 24% and the no identifi cation rate was 26%.

  (The results of the sequential line - ups are discussed below.) In short, one in

  three positive identifi cations using the traditional simultaneous method was

  clearly wrong – the witness ’ s ‘ guess ’ was wrong even though the target person

  was in the array. Experimental studies provide not only some insights into the

  rates of eyewitness error, they also provide some insights into methods that

  might reduce such errors. In the mid - 1990s a set of recommended identifi ca-

  tion practices were formulated by experimental researchers. These recommen-

  dations have generated new policy debates and new research on eyewitness

  error rates.

  Recommendations for c ollecting e yewitness e vidence

  Recommendations for how to conduct line - ups are not a new phenomenon,

  with the earliest published set of recommendations in the USA, produced by

  the District Attorney ’ s and Public Defender ’ s Offi ces of Clark County Nevada,

  appearing in 1967 (Procedure for line - up identifi cation, 1967 ). More recently,

  in light of the evidence provided by the analysis of DNA exonerations that

  mistaken eyewitness identifi cations play a signifi cant role in wrongful convic-

  tions and the burgeoning research literature assessing the effi cacy of different

  line - up procedures, a number of groups began reconsidering best practices for

  the collection of eyewitness evidence. The two most infl uential sets of recom-

  mendations came from a group of scholars sponsored by the American

  Psychology - Law Society (Wells et al ., 1998 ) and a group of scientists, legal

  practitioners and law enforcement personnel convened by then Attorney

  General Janet Reno under the auspices of the National Institute of Justice

  (Technical Working Group for Eyewitness Evidence, 1999 ). Although there

  is substantial overlap between the documents produced by these two groups,

  there are some notable differences in the recommendations that may be

  the result of the different constituents of committees that drafted the

  recommendations.

  In 1996, the Executive Committee of the American Psychology - Law Society

  (AP - LS; Division 41 of the American Psychological Association) appointed a

  subcommittee of scholars familiar with research on eyewitness identifi cation

  to produce recommended guidelines for the collection of eyewitness evidence

  using line - ups or photo arrays. The resulting paper describing those recom-

  mendations (Wells et al ., 1998 ) was accepted as an offi cial AP - LS Scientifi c

  Recent Developments in Identifi cation Science and Practice

  261

  Review paper by a unanimous vote of the Executive Committee. The recom-

  mendations were fourfold. First, all line

  - ups should be conducted using a

  double - blind procedure in which neither the witness nor the line - up adminis-

  trator knows the identity of the suspect. Second, the witness should receive

  instructions that the person who committed the crime may not be in the line -

  up or photo spread, and that the person administering the line - up does not

  know who the suspect is in the case. Third, the line - up should be constructed

  in such a way that the suspect is not salient in comparison to the other line - up

  members, based on the witness ’ s previous description of the perpetrator or

  any other feature that may call undue attention to the suspect. Fourth, the

  line - up administrators should take precise statements of confi dence from wit-

  nesses immediately after the identifi cation and before providing any feedback

  to the witness about their confi dence that the person that they identifi ed is

  the actual perpetrator.

  Two additional rules were considered by the subcommittee but they were

  not ultimately endorsed by the group. First, the group considered recommend-

  ing that all line - ups be conducted using a sequential presentation of the line - up

  members rather than the traditional simultaneous presentation. Because of

  concerns that sequential line - ups may be more prone to infl uence from a non -

  blind administrator and perceived reluctance on the part of law enforcement

  to adopt double - blind procedures, the group did not include sequential line -

  up presentation among its recommendations (research on administrator infl u-

  ence is discussed below). Second, the group considered recommending that

  all line - up or photo - spread administrations be videotaped and provided to the

  court. Again, the group stopped short of recommending videotaping line - up

  administrations as a rule because of concerns that videotaping increases costs

  to law enforcement agencies coupled with fears that videotaping might not

  capture improper infl uence because of the low - quality of amateur video pro-

  ductions and the need to have cameras capture multiple viewpoints – i.e., the

  line - up, the witness, the administrator.

  The NIJ guidelines were developed by the 34 members of the Technical

  Working Group on Eyewitness Evidence convened by NIJ to make recom-

  mendations about the collection of eyewitness evidence, including how to

  prepare mug books (photograph albums), develop composites, collect witness

  recollections of the event and conduct identifi cation procedures. Of interest

  here are the recommendations that the working group made regarding the

  administration of line - up and photo - spread identifi cation procedures.

  As mentioned previously, many of the recommendations made by this panel

  were similar to those made by the AP - LS subcommittee, although they often

  include greater detail. For example, the guidelines additionally recommend

  that only one suspect should appear in each line

  - up or photo

  - spread and

  that fi llers be chosen to match the witness ’ s description of the culprit. The

  suspect should also not stand out from the fi llers because of any unique fea-

  tures (e.g., tattoos, scars). In terms of instruct
ions, the NIJ guidelines not

  only recommended that administrators instruct witnesses that the perpetrator

  262

  Handbook of Psychology of Investigative Interviewing

  may not be in the line - up, but also that it is just as important to clear innocent

  suspects as to convict guilty ones, the perpetrator

  ’ s appearance may have

  changed since the time of the crime, and the police will continue to investigate

  the crime irrespective of whether the witness makes an identifi cation. Finally,

  the NIJ guidelines, like those from the AP - LS subcommittee, recommended

  obtaining statements from the witness regarding their level of certainty in

  any identifi cation made prior to providing them with feedback about their

  decision.

  The researchers of the working group identifi ed two major shortcomings

  of the guidelines (Wells, Malpass, Lindsay, Fisher, Turtle & Fulero, 2000 ).

  Unlike the AP - LS panel recommendations, the NIJ guidelines contained no

  recommendation for double - blind line - up administration. In the absence of

  such a recommendation, the guidelines did warn the administrator not to do

  or say anything to the witness that would infl uence his or her decision. Like

  the AP - LS recommendations, the NIJ guidelines also failed to recommend

  presenting line - up members sequentially rather than simultaneously, although

  both methods are described in the guide. Despite support from the social

  scientists for these recommendations, police offi cers and prosecutors voiced

  practical concerns, including the extra resources needed for double

  - blind

  administration and the potential for previous convictions obtained using simul-

  taneous line - ups to be appealed, which prevented the inclusion of these line - up

  features in the NIJ recommendations (Wells et al ., 2000 ).

  Does the available science, including the science produced since the publica-

  tion of the AP - LS white paper and the NIJ guidelines, support the recom-

  mendations for conducting line - ups and photo - spreads that were made by

  these two groups? In the following sections, we review the literature related

  to the recommendations on line - up composition, witness instruction, sequen-

  tial vs. simultaneous presentation of line - up members and double - blind line - up

  administration.

  Line - u p c onstruction or c omposition

  The composition of a line - up is important because the proper selection of

  line - up members can help control some of the damaging effects of witnesses

  guessing. If the suspect is the only line - up member, as when police conduct

  a show - up containing only the suspect, then witnesses who guess will neces-

  sarily pick the suspect. A line - up containing fi llers (i.e., line - up members who

  are not suspects) helps to minimize bad outcomes from guessing in target -

  absent line

  - ups as guesses should be distributed equally among line

  - up

  members, assuming that the line - up is unbiased, with only a fraction of the

  guesses resulting in an identifi cation of an innocent suspect. However, the

  choice of fi llers can bias witnesses towards choosing the suspect if the suspect

  is the only line

  - up member who has features mentioned in the witness

  ’ s

  description. Under these circumstances, although a line - up with one suspect

  and fi ve fi llers may have a nominal size of six, in effect, the suspect is the only

  Recent Developments in Identifi cation Science and Practice

  263

  viable line - up member for the witness to choose. Thus, the line - up becomes

  a test of the witness ’ s recall of his or her description rather than a test of the

  witness ’ s recognition of the perpetrator ’ s face (Luus & Wells, 1991 ).

  In recognition of these issues, both the AP - LS and NIJ guidelines recom-

  mended that a suspect should not stand out from the line - up distractors and

  that the decision about whether the suspect does stand out in problematic

  ways should be determined by comparing the line - up members to the descrip-

  tion provided by the witness. Care should be taken to ensure that the suspect

  is not salient, as might be the case if the suspect is the only line - up member

  to possess a feature mentioned in the witness ’ s description (Lindsay & Wells,

  1980 ).

  One way to examine the bias in arrays is to examine the rates at which

  ‘ designated innocent

  ’ individuals who resemble perpetrators/targets are

  chosen relative to the (other) fi llers. In situations where one wants to assess

  line - up fairness in the absence of choosing data, the most common procedure

  begins by obtaining the witness ’ s original description of the perpetrator and

  then presenting that description to 30 or more ‘ mock witnesses ’ (Doob &

  Kirshenbaum, 1973 ). These participants have not seen the crime or the per-

  petrator but are asked to make an identifi cation from the line - up based on the

  description provided by the actual witness to the crime. If mock witnesses,

  who did not see the perpetrator, pick the suspect at levels greater than would

  be expected by chance (e.g., greater than 20% of the time for fi ve - person line -

  ups and 17% of the time for six - person line - ups), then there is evidence that

  the line - up is biased against the suspect (Doob & Kirshenbaum, 1973 ). The

  functional size of the line - up, or the number of plausible line - up members, is

  obtained by dividing the number of mock witnesses by the number of suspect

  identifi cations (Wells, Leippe & Ostrom, 1979 ). For example, if 11 of 30

  mock witnesses (37%) identify the suspect, then the line - up has a functional

  size of 2.73 (30/11). Other methods of assessing line

  - up fairness include

  effective size (Malpass, Tredoux & McQuiston - Surrett, 2007 ) and E ’ (Tredoux,

  1998 ).

  Studies of the fairness of line - ups used in actual cases suggest that many

  line - ups are biased against defendants. Brigham, Meissner

  & Wasserman

  (1999) tested the proportion of mock witnesses who chose the suspect in 18

  US cases. Suspects were selected an average of 2.6 times as often as one would

  expect by chance alone. Wells & Bradfi eld (1999) , in a study of ten photo

  arrays and line - ups from US cases, found suspects were selected, on average,

  more than 2.1 times as often as expected by chance. Of course, arrays in these

  studies were not randomly selected – they were from contested identifi cations

  – so may overestimate the level of bias in the majority of actual cases.

  In a mock witness study of 25 actual fi ve - person line - ups used in France,

  the observed proportion of mock witnesses who chose the suspect was 48%

  in contrast to the expected rate of 20% – nearly 2.5 times the chance rate (Py

  et al ., 2003 ). Similarly, Valentine & Heaton (1999) studied suspect bias using

  photographs of a representative set of 25 line - ups and 16 video parades from

  264

  Handbook of Psychology of Investigative Interviewing

  the UK. The video parades consist of 15 - second clips of individuals and are

  selected from an archive of more than 15,000 clips. In live line - ups
individuals

  are shown simultaneously; video clips are shown sequentially. All arrays con-

  tained nine members, with an expected guess rate of 11% per face. Mock

  witnesses selected line - up suspects 25% of the time, a rate 2.2 times chance,

  and video parade suspects 15% of the time, 1.4 times chance. The difference

  between line - ups and video parades might be attributable to the methods of

  fi ller selection (perhaps more fi llers resembling witness descriptions are avail-

  able in videos – a point noted by Valentine & Heaton, 1999 ), the method of

  presentation (simultaneous vs. sequential) or other factors. That the line - up

  multiplier is lower than in the USA may be attributable to the larger arrays,

  to fairer arrays or both.

  Given that the line - ups used in actual cases appear to be relatively biased,

  is there an optimal strategy for choosing line - up fi llers that will decrease the

  likelihood that an innocent suspect will be identifi ed? There are two primary

  strategies for selecting fi llers for line - ups or photo - spreads: match - to - suspect

  and match

  - to - description. The match

  - to - suspect strategy involves selecting

  fi llers with features that are similar to the suspect ’ s features. Although match-

  ing fi llers to the suspect may appear to be a reasonable method, taken to the

  extreme the strategy would result in a line - up with a suspect and fi llers that

  are indistinguishable from one another.

  An alternative method for choosing fi llers is the match - to - description (or

  match - to - culprit) strategy. With this strategy, line

  - up constructors choose

  fi llers who share the features of the perpetrator that the witness mentioned

  in his or her description of the culprit but who vary on features that were

  not mentioned by the witness (Luus & Wells, 1991 ). So if the witness described

  the culprit as in his late twenties, 6 feet tall, medium build, with blond

  hair and light eyes, then all the fi llers should share these features but could

  vary on whether their hair is curly or straight and whether their eyes are green

  or blue.

  There is some research support for an advantage of the match - to - description

  strategy. For example, one study tested the rate of suspect identifi cations when

  fi llers were selected from a pool of potential fi llers using one of three proce-

  dures (Wells, Rydell & Seelau, 1993 ). In the mismatch - description condition,

 

‹ Prev