ArticlePDF Available

Variations in the Encoding Conditions Can Affect Eyewitnesses' Vulnerability to Suggestive Influence

Authors:

Abstract

This is the latest study we have published using our 'Shift & Stick' Paradigm. The shift and stick paradigm involves two basic elements:1) Shift: After voicing their initial identification decision, but before documenting the final choice, participants are told, “Take another look, does anyone [else] look familiar?” . This prompt consistently leads a large majority of participant/witnesses to shift their identification decisions and selected a different picture. 2) Stick: After shifting, witnesses are given confirming feedback to reinforce the shifted selection, and after some delay they are administered the same lineup again. When given a second chance to identify the culprit, most of those witnesses who shifted to another picture and then had that altered decision reinforced with confirming feedback go on to select the same picture they had shifted to. In the current experiment, we examined how variations in encoding conditions could influence witnesses vulnerability to this sort of suggestive behavior by the lineup administrator
This arcle has been accepted for publicaon in Applied Cognive Psychology. Copyright noce: This arcle may
not exactly replicate the nal version published in ACP. It is not the copy of record.
Variations in the Encoding Conditions Can Affect Eyewitnesses’
Vulnerability to Suggestive Influence
Mitchell L. Eisen T’awna Williams
Jennifer Jones Rebecca Ying
California State University, Los Angeles
Acknowledgements:
Mitchell L. Eisen https://orcid.org/ 0000-0003-0395-6705
T’awna Q. Williams https://orcid.org/ 0000-0002-8497-2098
Jennifer M. Jones https://orcid.org/ 0000-0003-4112-9276
Rebecca C. Ying https://orcid.org/ 0000-0002-5666-7790
T’awna Williams is now at John Jay College of Criminal Justice
Jennifer Jones is now at is now at John Jay College of Criminal Justice
Rebecca Ying is now at Iowa State University.
We would like to acknowledge Marianne Lacsamana, Brittany Altuna, Alisheia Moreno,
Veronica Monarrez, Ethan Vaca, and Adriana Mendez for their work in collecting the data. The
work was supported by the RISE Program of the National Institutes of Health.
The data that support the findings of this study are available upon reasonable request.
We have no conflicts of interest to disclose.
Correspondence regarding this article should be directed to Professor Mitchell Eisen,
meisen@calstatela.edu.
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 2
Abstract
This experiment was designed to examine how viewing conditions could affect witnesses’
vulnerability to suggestive influence. It was predicted that when the encoding conditions were
stronger, accurate witnesses would be less likely to shift their decisions when prompted to
reexamine the lineup, and that confirming feedback would effectively solidify influenced
identifications. Participants (N=411) watched a simulated-crime from two different viewing
distances and were asked to identify the culprit from a photographic lineup. After voicing their
identification decisions, participants were prompted to reexamine the lineup. Half the
participants then received confirming-feedback for their decisions. Twelve-minutes later,
witnesses viewed the same lineup and were asked again to identify the culprit. As predicted,
accurate witnesses were less likely to shift their identification decisions when prompted to
reexamine the lineup in the close, but not the far viewing condition. Also, shifted identification
decisions that were reinforced with confirming feedback were asserted with higher confidence.
Keywords: Eyewitness, Lineups, Suggestion
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 3
On a Sunday morning in Los Angeles, a man stood in the middle of a busy street, pointed
a handgun at a city bus, and opened fire; striking the driver and causing the bus to crash. The
police could only find a single eyewitness who saw the crime and was willing to speak with
them, and this witness was nearly 100-yards away from the shooting. The eyewitness described
the shooter as a Hispanic male in his 20s with dark hair, grey shorts, and black tee shirt. They
combed the area for potential suspects who matched this general description, and found Carlos
Cruz hanging out in front of his home nearby, wearing black shorts and white tee shirt. Despite
the clothing mismatch, Cruz was detained in handcuffs and presented to the witness using a
highly suggestive identification procedure called a showup. At the showup, the witness
mistakenly identified Cruz as the shooter, and despite the absence of gun powder residue found
on his hands or clothing, Carlos was arrested and wrongfully prosecuted for this heinous crime
based on a single eyewitness identification, made using a highly suggestive procedure, by a
person who viewed the crime from nearly 100-yards away (People v. Carlos Cruz, Los Angeles,
County, BA435151).
The combination of using a suggestive identification procedure with a witness who
viewed the crime from such a far distance raises an interesting question; are witnesses more
vulnerable to suggestive influence when the encoding conditions are poor? If the witness had
gotten a better view of the shooter, would he have been less affected by the suggestive nature of
the identification procedure used, and less likely to have misidentified Carlos as the shooter?
Although relatively few studies in the eyewitness decision-making literature have directly
examined this question, there is some evidence supporting the notion that poor encoding
conditions can influence a witnesses’ vulnerability to suggestive influence. (Pezdek & Blandon-
Gitlin, 2005).
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 4
Discrepancy-Detection and Suggestion in Eyewitness Decision-Making.
Previous research examining the classic misinformation effect indicates that people are
more vulnerable to suggestive influence when their memory for the target event is weaker
(Loftus, 2005; Tousignant, et al., 1986). According to the detection discrepancy principle
(Tousignant et al., 1986), if a person’s memory for the event is weak, they will be less likely to
detect any discrepancy between their original memory and the suggested information; thus
increasing the likelihood of accepting the suggested information without question (Blank, 1998;
Loftus, 2005). Eisen et al. (2022) noted that misidentifying an innocent suspect from a
photographic lineup can be viewed as an example of misinformation acceptance. The implied
misleading suggestion from law enforcement is that the innocent suspect is the actual culprit, and
the act of choosing that suspect’s picture would constitute the acceptance of the suggested
misinformation. Since, poor viewing conditions limit witnesses’ ability to acquire a clear image
of the culprit, and witnesses are more vulnerable to suggestive influence when they are unable to
detect discrepancies between the suggested information and their memory for the event;
witnesses who get a poor view of the culprit should be more vulnerable to suggestive influence
when attempting to make an identification, because they will be less likely to detect
discrepancies between their relatively weak memory for the culprit and the innocent suspects
presented to them.
Suggestive influence can come in many forms and the person administering the
identification test can influence witness performance regardless of whether a showup is used,
like in the Carlos Cruz case, or when the suspect’s picture is presented in a photographic lineup
(Wells et al., 2020). The current study examined administrator influence when using
photographic lineups. Before describing this study in detail, we first review previous research
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 5
examining how encoding conditions can affect eyewitness decision-making, and then describe
studies looking at how various forms of suggestive influence from the lineup administrator can
affect a witness performance.
How Encoding Conditions Can Affect Eyewitness Decision-Making.
Smith et al. (2019) noted that when the witnessing conditions are poor, and match-to-
memory is weak or absent when viewing the lineup, witnesses will often adopt a more lenient
criterion for making identification decisions. Arguably, this more lenient decision criteria should
leave them more open to suggestive influence. The logic is straightforward: under culprit-absent
conditions, witnesses who are unable to acquire a clear image of the culprit when viewing the
crime should be less likely to experience strong internal cues detecting discrepancies between
their relatively weak memory for the culprit and the faces of the innocent suspects presented to
them. The absence of experiencing potent internal cues to guide their decision making should
leave witnesses more vulnerable to external influence from the lineup administrator.
Alternatively, if encoding conditions are more optimal and the witness is able to acquire a clear
image of the culprit to match with, they should be more able to detect discrepancies between
their relatively clear memory for the culprit and the faces of innocent suspects being presented to
them. When the decision to accurately reject the lineup is driven by strong internal cues detecting
clear discrepancies between their memory for the culprit and the innocent suspects being
considered, witnesses should be less open to external influence.
The relationship between viewing conditions and vulnerability to external influence
should arguably be the same when the culprit is present. If a witness had a very good view of the
perpetrator and was able to acquire a clear memory of the culprit’s face, they are more likely to
experience a clear match-to-memory when viewing the culprit’s photo, and this strong internal
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 6
sense of recognition should guide their decision making; leaving them less open to external
influence. Alternatively, if a witness was unable to acquire a clear image of the perpetrator when
viewing the crime, they will be less likely to experience a clear match-to-memory when viewing
culprit’s picture in a photo array; and the absence of strong internal cues to guide their decision
making will leave them more open to suggestive external influence.
Suggestive Influence from the Lineup Administrator
Broadly speaking, social influence from the lineup administrator can be broken down into
two general forms: pre-identification behaviors by the administrator that can influence decision-
making, and post-identification behaviors that can affect a witnesses’ confidence and
retrospective judgements related to the viewing experience and act of making the identification
(Eisen, et al., 2018). Pre-identification behaviors involve actions by the lineup administrator that
precede documentation of the final identification decision. These pre-identification behaviors
include social cues offered during an identification test that encourage choosing (Clark et al.,
2013), steering the witness to select a specific suspect’s picture when viewing a photo array
(Charman & Quiroz, 2016; Greathouse & Kovera, 2009; Kovera & Evelo, 2017), and/or
communications from the lineup administrator to shift away from their initial decision after they
have voiced an initial judgment, but before that identification has been documented and
memorialized (Eisen et al., 2018). Post-identification behaviors primarily involve feedback
provided to the witness about their identification after that decision has been finalized and
documented (Wells & Bradfield, 1998). This is most commonly investigated using the post-
identification feedback effect (see Steblay, Wells, & Douglass, 2014 for a review).
The Shift and Stick Paradigm
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 7
The current study examined how encoding conditions affected witnesses’ vulnerability to
social influence when considering both the influence cues that precede the finalized
identification decision, and feedback offered after the choice had been documented and
memorialized. This was accomplished by using the shift and stick paradigm (Eisen et al., 2018).
The shift and stick paradigm involves two basic elements: 1) A prompt for the witness to
reexamine the photos after initially voicing their decision, but before the choice has been
finalized; and 2) Confirming feedback delivered after a decision has been documented and
memorialized by the lineup administrator. In their original study using this paradigm, Eisen and
his colleagues had participant-witnesses watch a video of a simulated crime and asked them to
identify the culprit from a photographic lineup. After voicing their initial identification decision,
but before documenting the final choice, participants were told, “Take another look, does anyone
[else] look familiar?” After receiving this prompt to reexamine the lineup, a large majority of
participant/witnesses shifted their identification decisions and selected a different picture. Half
of the witnesses then received confirming feedback for their final decisions, regardless of
whether they shifted or not. After a 10-minute delay, a different experimenter administered the
same lineup to the witnesses again to see if they would stick with the picture they shifted to, or
revert back to their original decision. These investigators reported that when the witnesses were
given a second chance to identify the culprit, most of those who shifted to another picture and
then had that altered decision reinforced with confirming feedback went on to select the same
picture they had shifted to. These experiments demonstrated that seemingly innocuous
comments made by non-blind lineup administrators after witnesses had voiced an initial
decision, but before that selection had been documented and memorialized, can steer many
witnesses away from their initial identification decisions; and when this happens, the altered
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 8
choices can be easily solidified with suggestive confirming feedback. Essentially, this work
revealed the fluid nature of witness decision making prior to the final documentation and
memorialization of the final identification decision.
Are Accurate Witnesses Less Vulnerable to Suggestive Influence?
Theoretically, one would expect that accurate witnesses should be less likely to shift
when prompted to reexamine the lineup and reconsider their identification decisions, since the
decisions of accurate witnesses are more likely to be driven by internal cues; either recognizing
the culprit when present, or detecting discrepancies between their memory for the culprit and the
suspects presented. Surprisingly, this is not what Eisen and his colleagues found in their original
study using the shift and stick paradigm. Contrary to expectations, accurate witnesses were just
as likely to shift as their inaccurate counterparts when prompted to reexamine the lineup.
Remarkably, when the culprit was present, over three quarters of the witnesses who had correctly
selected his picture shifted their decisions when prompted to reexamine the photo array, and
went on to mistakenly identify a filler from the group (Exp. 1, 76%, Exp. 2, 78%). The very
high shift rate among accurate witnesses shows just how potent the suggestion to re-examine the
photo array can be in inducing witnesses to shift their identification decisions.
When Encoding Conditions Are Poor, Even Accurate Choices May Be More
Tentative and More Pliable. Eisen and his colleagues speculated that the absence of a
relationship between accuracy and shifting observed in their study may have been related to the
difficulty of the identification test used in that effort, which yielded a very modest 32.7%
accuracy rate across the two experiments. Notably, although shifting was not related to
accuracy, response latency did predict shifting in both experiments, as participants who made
quicker and presumably more definitive decisions were less likely to shift when prompted to
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 9
reexamine the lineup. Logically, response latency provides some insight into the strength of the
internal cues driving the witnesses’ decision making. Eisen et al. (2018) reasoned that under the
conditions examined, accuracy appeared to be less important than quick, decisive choosing when
it came to predicting how easily someone can be influenced to change their identification
decision. This pattern of results opened the door to the obvious question: What would happen if
the encoding conditions were improved and the task was not so difficult? Under better viewing
conditions would accurate witnesses be less vulnerable to the suggestive influence of the lineup
administrator, therefore making them less likely to shift when prompted to reexamine the lineup
and reconsider their identification decisions?
The Current Study
The current study was designed to examine how manipulating the encoding conditions
would affect a witness’ vulnerability to suggestive influence by the lineup administrator when
using the shift and stick paradigm. Theoretically, when the viewing conditions are poor, and the
witness is unable to acquire a clear image of the culprit to match with, they should be less likely
to experience potent internal cues when viewing the lineup to guide their decision making;
leaving them more open to suggestive external influence from the lineup administrator to
reconsider their identification decisions. In contrast, under better viewing conditions, witnesses
are more likely to acquire a clear image of the culprit to match to, and as a result should be more
likely to experience potent internal cues when viewing the lineup to guide their decision making;
either recognizing the culprit when he is present, or detecting discrepancies between their
memory for the culprit and the innocent suspects presented, making them less vulnerable to
suggestive influence. Specifically, it was predicted that when encoding conditions were better
(i.e., viewing the crime from a closer distance), accurate witnesses would be less vulnerable than
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 10
inaccurate witnesses to the suggestive effect of being prompted to reexamine the lineup, and
would therefore be less likely to shift their decisions. However, when encoding conditions were
poorer (i.e., viewing the crime from a further distance), we expected to replicate the findings of
Eisen et al. (2018), and predicted that accuracy would not predict shifting. We also expected to
replicate the findings of Eisen et al. (2018), showing that regardless of the viewing conditions,
response-latency should predict shifting. This is based on the notion that response-latency is
highly correlated to confidence, and provides the best indication of whether the witness is
experiencing a strong match-to-memory when the culprit’s picture is present in the photo array
(Brewer & Weber, 2008; Wixted & Wells, 2017); or readily detects a discrepancy between their
memory or the perpetrator and the innocent suspects pictures when the culprit is absent.
Specifically, we predicted that witnesses who make quicker decisions will be less likely to shift
when prompted to reexamine the lineup regardless of viewing conditions or accuracy.
We also examined whether viewing conditions influenced witnesses’ vulnerability to
post-identification feedback effects. In the current study, post-identification feedback effects
were measured by sticking with shifted choices that were reinforced with confirming feedback,
and confidence in reinforced identification decisions. If viewing conditions influenced witness
vulnerability to feedback effects, then witnesses in the poor viewing condition would be more
likely to stick with their reinforced identification decisions and would be more confident in their
final decisions.
Method
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 11
Participants
Participants were 421 undergraduate students recruited from introductory psychology
courses at California State University, Los Angeles (F = 69.6%, M = 30.4%). Overall, their ages
spanned from 17 – 47 years (M = 19.25 years, SD = 2.32 years). The racial background of the
sample varied, with 73.9% identifying themselves as Hispanic, 18.1% as Asian, 3.3% as White,
2.9% African American, and the remaining 1.9% identifying themselves as something other than
these groups.
Procedure
Overview. Participants were randomly assigned to watch a video of a staged crime shot
from either 13- or 30-feet away. After a 10-minute delay, participants were asked to make an
identification from either a culprit-present or -absent photographic lineup. After voicing their
initial identification decision, participants were prompted to re-examine the lineup. Specifically,
they were told, “Take another look, does anyone [else] look familiar?”. After hearing this
prompt, participants either shifted from their initial decision (“shifters”) or stuck with their initial
choice (“non-shifters”). Shifters and non-shifters were then randomly assigned to either receive
confirming feedback or no feedback following this second identification. After a 10-minute
delay, participants were taken to a different room and met with a new experimenter, were shown
the same photo array again, and were asked to identify the culprit. This experimenter also
secured ratings of confidence in the decision.
Design
The experiment was a 2 (30 ft. view vs. 13 ft. view) × 2 (culprit present vs. absent) × 2
(feedback vs. no feedback) between-subjects design, in which each participant received the
prompt to re-examine the lineup.
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 12
Viewing The Crime. Groups of up to 10 participants gathered together to read and sign
the informed consent. Each participant was escorted to a separate room one at a time to view the
crime video individually on a 21.5 inch monitor under the supervision of an experimenter.
Before each viewing, the researcher read the following instruction, “I am going to show you a
video of a staged crime. Please pay close attention. Later on, you will be asked to identify the
culprit in the video from a lineup”. In actual cases, there is rarely forewarning that a crime will
occur, and, as a result, many people do not process the details of the crime as well as they might
if they somehow knew beforehand that they were about to witness a crime. We forewarned
participants in this study in an effort to maximize the encoding experience across conditions so
there would be less chance that some participants simply did not attend to the details of the event
as well because they did not know they would be tested on it later. The individual viewing
sessions coupled with the forewarning were designed to reduce random variability in attention
among participants across conditions in an effort to enhance the chances of getting reliable
differences in processing the event across the two different viewing conditions. After viewing
the video, participants returned to the central waiting room where they were monitored to make
sure they did not discuss the procedures with one another.
The initial identification. After a 10-minute delay, each participant was directed into a
new room where they viewed a photo array and attempted to make an identification of the
culprit. Participants were read the Los Angeles Police Department pre-lineup admonition prior
to viewing the photo array:
“I’m going to show you a set of photographs to see if you can identify the
culprit from the carjack video you watched. This group may or may not
contain a picture of the person who committed the crime. Pay no attention to
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 13
markings or numbers that may appear on the photos or any other differences
in the type of style of photographs”.
The lineup administrator recorded the time it took each participant to select a photo.
After voicing their initial identification decision, participants were read a prompt to reexamine
the lineup. The phrasing of the prompt differed depending on if the witness selected a photo or
rejected the lineup. If the participant selected a photo, the lineup administrator pointed to that
photo and said, “You selected number __, is that right?” After confirming the selection, the
participant was told, “Take another look, does anyone else look familiar?” If the participant
rejected the lineup, they were told, “You did not see him in there, is that right?” The participant
was then told, “Take another look, does anyone look familiar?”.
Confirming Feedback. After the prompt to re-examine the photo array, some
participants shifted away from their initial decisions (shift) while others opted to stick with their
original decisions (stick). Both shifters and non-shifters were randomly assigned to receive
confirming feedback (“Good, you identified suspect”) or not (“Okay”). Participants then
returned to the central waiting room where they were monitored to make sure they did not
discuss the procedures with others.
Final identification. After a 12-minute delay, each participant met with a new lineup
administrator who presented them with the same photo array again and asked them to identify
the culprit. Participants were shown the same photo array again and were asked “Can you tell
me which one of these is the man that you saw in the video with the gun who stole the car?”
Participants were then asked to rate their confidence in that final selection on a 0-100 scale, with
zero being not confident at all and 100 being absolutely confident. Following this identification,
participants returned to the central waiting room where they completed a final questionnaire.
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 14
Final questionnaire. For their final task, each participant responded to 10 questions that
asked for their retrospective judgements of their witnessing experience and memory quality
(Wells & Bradfield, 1998).
Attention check. At the end of the final questionnaire, participants were asked to recall
the hair color of the carjacker as an attention check. They were also asked whether they had
prior knowledge of the experiment. Those who could not answer the hair color question
correctly or disclosed having prior knowledge of the experiment were dropped from the analyses.
A total of ten participants were dropped based on this screening.
Materials
Crime Video. Participants viewed a 30-second video of a staged crime filmed from
either 30-feet or 13-feet away. The video depicted a young Latino male carjacking a young
White male at gunpoint in an empty parking lot. The perpetrator’s face was clearly visible for
19-seconds.
Lineup Construction. A large pool of photos of young Latino males who generally
matched the appearance of the culprit was compiled by our research team. A group of 10
researchers rated the similarity of the photos to the culprit, and these judgments were used to
narrow the pool to eight potential fillers. A pilot group of 42 undergraduates watched the crime
video (13-foot version), were shown the eight remaining photos, and were asked to identify the
culprit. Photos that were selected less than 7% of the time were excluded. This process was
used to exclude two of the potential fillers and narrowed the group to six. A new set of 21
students were given a mock lineup task with the remaining six filler photographs. These students
were presented with a general description of the culprit (i.e., Male, Latino, 18-25, dark hair, light
skin, medium height and weight) and were asked to select the photograph that best fit this
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 15
description. This task yielded a Tredoux’s E of 4.37, indicating that the lineup had a reasonable
degree of fairness. The most selected filler was selected to replace the culprit in the culprit-
absent condition. Six culprit-present lineups were created, with the culprit rotated to a different
position in each. For the culprit-absent lineups, six versions of this lineup were also created by
rotating the pictures to different positions.
Worst-Case Scenario Analysis. Results revealed that the most selected photo in the
culprit-absent condition was different in the 13-foot and 30-foot condition. As such, a worst-
case scenario analysis was used (Clark, Marshall, & Rosenthal, 2009) and the nominated suspect
was determined by the filler that was chosen most often under culprit-absent conditions in each
viewing-distance condition.
Results
The results are presented in three general sections. The first section examines accuracy,
the second section examines shifting, and the third section describes the effects of post-
identification feedback on solidifying influenced decisions.
Accuracy.
Table 1 displays the proportion of witnesses who identified the culprit, a filler, or made no
identification as functions of culprit presence and viewing conditions (Close: 13 ft. versus Far:
30 ft.). It was expected that accuracy would be higher in the close versus distant viewing
conditions, and that this would be consistent in both culprit-present and culprit-absent conditions,
as evidenced by a greater likelihood of identifying the culprit when he was present in the lineup,
and greater correct rejection rates when the culprit was absent. A binary hierarchical logistic
regression analysis was conducted with accuracy as the dependent variable, and with both
viewing distance (13 ft. vs. 30 ft.) and culprit-presence as the predictors. When the culprit was
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 16
present, selection of the target was coded as accurate, and both filler choices and rejections were
coded as inaccurate. When the culprit was absent, rejections were coded as accurate, and falsely
identifying any lineup member was coded as inaccurate. Before conducting the regression
analyses, we inspected the Variance Inflation Factor (VIF) for each predictor. Results of these
test showed no multicollinearity concerns (VIFs < 1.12).
In the first model, the main effects were entered, and, in the second model, the two-way
interaction was entered. For the first model, the predictor model was a significant improvement
over the constant-only model, χ2 (2, N = 411) = 76.85, p < .001. As expected, a main effect for
distance was revealed, as participants in the 13 ft. viewing condition were significantly more
likely to make an accurate identification decision than those in the 30 ft. group, B = .89, SE =
.24, Wald’s χ2(1) = 13.56, p < .001, OR = 2.44, 95% CI [1.52, 3.91]). A main effect was also
revealed for culprit presence, as participants were more likely to be accurate when the culprit
was present compared to when he was absent, B = .1.94, SE = .26, Wald’s χ2(1) = 55.74, p
< .001, OR = 6.94, 95% CI [4.17, 11.55]. The second model was not a significant improvement
over the constant-only model, χ2(1, N = 411) = .55, p = .46, and no interaction was found
between distance and culprit-presence, B = .39, SE = .52, Wald’s χ2(1) = .56, p < .001, OR =
1.48.
Shifting
The next set of analyses examined factors that influenced shifting. It was predicted that
accuracy would predict shifting in the close (13 ft.) but not the far (30 ft.) viewing condition.
Also, based on the results reported by Eisen et al. (2018), we predicted that response latency
would predict shifting regardless of viewing distance. A binary hierarchical logistic regression
analysis was conducted with shifting as the dependent variable and the following predictors:
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 17
accuracy, viewing distance (13 ft. vs. 30 ft.), culprit presence, and response latency. In the first
model, the main effects were entered, and two- and three-way interactions were entered in
subsequent models. For the first model, the predictor model was a significant improvement over
the constant-only model, χ2 (4, N = 411) = 28.55, p < .001. As expected, a main effect for
response latency was revealed, as participants who took more than 12 seconds to make an
identification were significantly more likely to shift than those who took less than 12 seconds, B
= 1.04, SE = .25, Wald’s χ2(1) = 16.60, p < .001, OR = 2.82, 95% CI [1.71, 4.65]. A main effect
for accuracy was also revealed, as inaccurate participants were significantly more likely to shift
than their accurate counterparts, B = -.70, SE = .26, Wald’s χ2(1) = 7.11, p = .008, OR = .50,
95% CI [.30, .83]. Table 2 shows that the accuracy affect was primary driven by differences in
shifting between accurate and inaccurate witnesses in the close (13ft.) condition. Neither culprit
presence nor view predicted shifting, Bs < .18, SE < .25, Wald’s χ2(1) < .66, p < .67, OR < .11.
Although the second model examining two-way interactions was not a significant
improvement over the constant only model, χ2 (6, N = 411) = 7.76, p = .26, our a priori
prediction was that accuracy would predict shifting in the close (13 ft.) but not the distant (30 ft.)
viewing conditions; thus, we examined the simple slopes of accuracy as a function of viewing
distance. These analyses revealed the predicted interaction between view and accuracy on
shifting. Table 2 shows that although three quarters of inaccurate witnesses shifted regardless of
viewing distance, in the closer 13 ft. viewing condition, accurate witnesses shifted substantially
less often (55.1%) than their inaccurate counterparts (76.9%), B = .92, SE = .34, Wald’s χ2(1) =
7.45, p = .006, OR = 2.52, 95% CI [1.30, 4.88]. In contrast, as expected, in the more distant 30
ft. viewing condition, there was no significant difference in shifting when comparing accurate
witnesses (68.1%) and their inaccurate counterparts (75%), B = .41, SE = .39, Wald’s χ2(1) =
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 18
1.14, p = .29, OR = 1.51, 95% CI [.71, 3.21]. Notably, Table 2 shows that for inaccurate
witnesses, shifting was generally comparable across the two viewing distances.
Interaction Between View, Response Latency and Accuracy. The next model
examining a three-way interaction between view, response latency and accuracy was a
significant improvement over the constant only model, χ2 (7, N = 411) = 4.29, p = .04, showing
that the two-way interaction between view and accuracy was qualified by this three-way
interaction, B = .08, SE = .04, Wald’s χ2(1) = 3.82, p = .05, OR = 1.08, 95% CI [1.00, 1.18].
Table 2 shows that accurate witnesses in the close viewing condition were less likely to shift than
accurate witnesses in the distant condition (55.1% vs. 68.1%), and witnesses who made quicker
decisions were significantly less likely to shift overall regardless of accuracy or viewing
conditions. Not surprisingly, accurate witnesses in the close viewing condition who made
decisions in under 12-seconds were the least vulnerable to suggestions by the lineup
administrator to shift their decisions. Table 2 shows that 26% of participants in the close viewing
group made accurate decisions under 12-seconds (54/208), and of that group of quick accurate
deciders, only 46.5% of those participants shifted when prompted to reexamine the lineup
(25/54). However, in the distant viewing group, quick accurate deciders shifted significantly far
more often. Table 2 shows that, in the distant viewing condition, only 14.7% of participants
made accurate decisions in under 12-seconds, (30/203), and of those quick accurate deciders,
60% shifted when prompted to reexamine the lineup (19/30). Indeed, even when only
considering witnesses who made accurate decisions under 12-seconds, quick accurate decisions
in the close-viewing group were less vulnerable to suggestive influence from the lineup
administrator than those in the far-viewing group.
Sticking: Solidifying Shifted Decisions with Confirming Feedback
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 19
This next section examines the effect of confirming feedback on solidifying shifted
identification decisions across conditions. After being directed to reexamine the lineup, both
shifters and non-shifters in each viewing condition were randomly assigned to either receive
feedback confirming the picture they settled on, or to get no feedback at all. Then, 12-minutes
later, participants met with a new experimenter who showed them a copy of the same photo array
they had viewed earlier and asked them to identify the culprit. It was expected that shifters in
both conditions who received confirming feedback would be more likely to stay with the photos
they shifted to when asked to identify the culprit in the final session.
Shifters. Shifters fell into one of three categories: (1) Shift & Stick – Participants who
shifted when prompted to take another look at the photos, and then selected the same pictures
they shifted to when asked to identify the culprit in the final session; (2) Shift revert
Participants who shifted when prompted to take another look at the photos, but then reverted
back to their original selection (i.e., the decision made before shifting) when asked to identify
the culprit in the final lineup session; and (3) Shift change – Participants who shifted when
prompted to take another look at the photos, and then made a third choice when asked to identify
the culprit in the final lineup session (not the original choice nor the choice they shifted to). For
the purposes of these analyses, the shift revert and shift change groups were combined. This
resulted in a two-level variable that we called Sticking. Shifters who stuck with their shifted
decisions in the final-lineup session were coded as 1 (Stickers), while shifters who either
reverted back to their initial response or changed their decision on the final identification test
were coded as 0 (Changers).
A hierarchical logistic regression analysis was conducted with sticking as the dependent
variable, and the following predictors: Feedback, Viewing distance (13 ft. vs. 30 ft.), Culprit-
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 20
presence, Response latency, and Accuracy on the final lineup (i.e., if the final choice made on
the final identification test was accurate, regardless of whether they shifted to that response or
not). For the first model, the predictor model was a significant improvement over the constant-
only model, χ2 (5, N = 191) = 27.50, p < .001. As expected, a main effect for feedback was
revealed, as participants who had their shifted choices reinforced with confirming feedback were
significantly more likely to stick with the choice they switched to compared to those in the no-
feedback condition, B = 1.21, SE = .25, Wald’s χ2(1) = 24.10, p < .001, OR = 3.34, 95% CI
[2.07, 5.41]). Neither culprit presence, accuracy, response latency, or viewing condition
predicted sticking, Bs < .29, SE < .34, Wald’s χ2(1) < .87, p > .39, OR < 1.01. The second model
was not a significant improvement over the constant-only model, χ2(9, N = 191) = 5.15, p = .12,
and no interactions were revealed.
Non-shifters. We also examined the consistency of identification decisions made by
participants who did not shift when prompted to reexamine the lineup (non-shifters). Table 3
shows that in both viewing conditions, non-shifters were very consistent in their selections when
shown the lineup again in the final session; so much so, that the lack of variance in their
performance precluded formal statistical analyses. In the 30 ft. condition, 95.8% of the non-
shifters who received confirming feedback chose the same pictures when asked to make a
selection in the final session. In the 13 ft. condition, consistency was a bit lower, but still very
high, as 90.3% of non-shifters who received feedback stayed consistent in the final lineup.
Confidence in the Final Identification Decisions.
It was predicted that shifters who received confirming feedback would be more confident
in their final identification decisions than shifters who did not have their altered responses
reinforced with feedback. To test this prediction, a 2 (Confirming feedback) × 2 (Shift vs. No
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 21
shift) × 2 (Viewing condition) × 2 (Final choice accuracy) ANOVA was conducted to examine
the relative effects of confirming feedback, accuracy, viewing conditions, and shifting on
participants’ confidence in their selections made in the final session. These analyses revealed a
main effect for feedback, as participants who received confirming feedback were more confident
in their final identification decisions than those who did not receive feedback F(1, 410) = 16.51,
p < .001, ηp2 = 0.04. A main effect for shift was also found, as shifters were less confident
overall compared to non-shifters, F(1, 410) = 13.03, p < .001, ηp2 = .32. A main effect for final
choice accuracy also emerged, as witnesses who had settled on an accurate identification
decision were more confident in their final identification decisions (M = 76.18, SD = 19.86), than
those who were inaccurate (M = 71.40, SD = 20.87), F(1, 410) = 5.80, p = .02, ηp2 = .14.
However, viewing distance was unrelated to confidence in the final decision, F(1, 410) = .007, p
= .93.
The main effects for shift and accuracy were qualified by a significant interaction
between shifting and accuracy, F(1, 410) = 3.85, p = .05, ηp2 = .01. Table 5 shows that accurate
non-shifters (i.e., those who chose correctly and then steadfastly did not shift when prompted to
reexamine the lineup) were more confident (M = 84.36, SD = 15.64), than inaccurate non-
shifters, (M = 74.71, SD = 22.20), F(1, 118) = 6.87, p = 0.01, ηp2 = .06, d = 0.50. However,
when considering shifters, there was no difference in confidence between accurate (M = 70.54,
SD = 20.59) and inaccurate witnesses (M = 70.35, SD = 20.37), F(1, 291) = .001, p = .95, ηp2
= .0001, d = .001. Notably, witnesses who shifted to accurate decisions reported very similar
confidence ratings to those who shifted to make a false identification, regardless of whether they
received confirming feedback.
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 22
Discussion
This experiment was designed to examine whether variations in the encoding conditions
would affect witnesses’ vulnerability to suggestive influence. As predicted, accurate witnesses in
the close viewing condition were significantly less likely to shift their decisions when prompted
to reexamine the lineup compared to their inaccurate counterparts. Theoretically, the better
encoding conditions provided witnesses with the opportunity to acquire a clearer memory for the
perpetrator’s face, which in turn increased the likelihood that witnesses would experience a clear
match-to-memory when viewing his image in culprit-present conditions, or detecting
discrepancies between their relatively strong memory of the perpetrator and the photos of the
innocent suspects presented to them in culprit-absent conditions. Based on this, it was expected
that accurate witnesses in the close viewing condition would be more likely to experience
stronger internal cues guiding their initial correct identification decisions, therefore reducing
their vulnerability to external influence from the lineup administrator when prompted to
reexamine the lineup and reconsider their initial choices.
As expected, results from the distant viewing condition replicated the findings reported
by Eisen et al. (2018), who also found that accuracy did not predict shifting. Notably, the video
stimulus used in that previous study provided witnesses with a relatively poor view of the culprit,
resulting in relatively low accuracy rates which were comparable to the performance of
witnesses in the distant viewing condition of the current study. Logically, if witnesses have a
limited ability to acquire a clear image of the perpetrator when viewing the crime, they should be
less likely to experience a match-to-memory when viewing photos of potential suspects.
Theoretically, when witnesses view a photographic lineup and do not experience a strong match-
to-memory, they will often adopt a more lenient decision criterion for making identifications,
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 23
which leaves them more open to influence from the lineup administrator suggesting that their
initial choice may have been mistaken, and that they should consider shifting their decision and
making a new selection. Results from the response latency analyses further indicate that
witnesses who experienced a stronger match-to-memory when viewing the suspect were less
vulnerable to suggestive influence.
Response Latency, Confidence and Vulnerability to Suggestive Influence
In the current study, response latency is arguably the best indicator of a witness either
recognizing the culprit when present or clearly detecting discrepancies between their memory for
the actual perpetrator and the innocent suspects presented in culprit absent conditions. As
predicted, witnesses who made quicker decisions were less likely to shift when prompted to
reexamine photos and reconsider their identification decisions. Since response latency is highly
correlated with confidence (Brewer & Weber, 2008), it is likely that witnesses who made quick
and definitive decisions were more confident in their decisions, which in turn made them more
resistant to suggestions from the lineup administrator that could influence their decision-making.
This replicated the findings of Eisen et al. (2018) and is consistent with the work of Clark et al.
(2013), who also found that witnesses who made quicker decisions were more resistant to pre-
identification suggestive influence from the lineup administrator.
The interaction between accuracy, response latency and viewing conditions further
informs our understanding of how poor viewing conditions can increase vulnerability to
suggestive influence. Notably, when considering only witnesses who made quick accurate
decisions, those witnesses in the distant-viewing group were still more likely to shift when
prompted to reconsider their decisions than quick accurate deciders in the close-viewing group.
This pattern of results suggests that even when restricting our analyses to the very best
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 24
performing witnesses (i.e., those who made accurate identification decisions in under 12-
seconds), the accurate decisions of witnesses in the poor viewing conditions were generally more
tentative, pliable, and vulnerable to suggestion from the lineup administrator.
Rather than using response latency as an indicator of confidence, it certainly would have
been better to directly assess confidence in the initial decisions before prompting participant-
witnesses to reexamine the photos and reconsider their choices. Unfortunately, it was not
practical to assess confidence in the initial identification decision when using this paradigm, as
previous research examining the accessibility hypothesis suggests that asking witnesses to make
a confidence judgements before prompting them to reexamine the lineup would likely have
attenuated the effect of the influence prompt (Wells & Bradfield, 1999). According to the
accessibility hypothesis, people are not fully aware of their confidence judgments until they are
formally asserted. Thus, asking a witness to voice their confidence brings that judgment to
awareness (i.e., makes it accessible), and to some degree solidifies that judgment, making it less
malleable. Neuschatz et al. (2007) referred to this as the feedback-prophylactic effect. We
believe that the prophylactic effect also applies to the shift and stick paradigm. As such, when a
choice is initially voiced, but not yet finalized, decision making is still in a fluid state, as they
have not yet committed to the choice. However, once the witness formalizes the choice by
stating how confident they are in the decision, judgments related to the identification decision
should be less malleable from that point forward. From an applied perspective, in actual cases, it
is highly unlikely that an officer who is trying to steer a witness away from a tentative decision
would first ask the witness to state their confidence in their initial choice, before suggesting that
they consider shifting their decision.
Post-identification feedback effects: Sticking and Confidence
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 25
In the current study, post-identification feedback effects were measured by sticking with
shifted choices that were reinforced with confirming feedback, and confidence in reinforced
identification decisions. Although poor viewing conditions in this study increased witnesses’
vulnerability to pre-identification suggestive influence as measured by shifting when prompted
to reexamine the photos and reconsider their decision, viewing conditions did not appear to
influence the post-identification feedback effects examined. Specifically, witnesses in the poor-
viewing group who received confirming feedback for their shifted decisions were not more likely
to stick with those choices and were not more confident in their final identification decisions
when compared to their counterparts no feedback condition.
Policy Implications
From a practical perspective, these data show that prompting witnesses to take another
look at the photos after they have voiced an initial decision but before this choice has been
documented and finalized can prompt individuals who had initially accurately rejected a lineup
that contained a picture of an innocent suspect to make an erroneous selection. Notably,
witnesses who do not get a good view of the culprit appear to be particularly vulnerable to
shifting when prompted to reexamine the lineup. Previous research indicates that this type of
prompting is not uncommon in law enforcement practice. Indeed, in some jurisdictions, when a
witness rejects the lineup, it is actually prescribed practice for the lineup administrator to ask the
witness if anyone in the group looks like the culprit after an initial decision has been voiced, but
before it has been memorialized (Fitzgerald, Rubínová, & Juncu, 2021). Results of this study
and Eisen et al, (2018) show that this sort of additional probing and prompting after an initial
tentative decision has been voiced but before the final choice has been memorialized can have
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 26
dangerous consequences, resulting in some witnesses who had initially rejected the lineup to
make an identification.
Although double-blind controls are a crucial first step in working against the potential
damaging effects of influence from the lineup administrator, double-blind controls alone may not
be sufficient to prevent these sorts of errors from occurring. Indeed, the risk of this sort of
prompt resulting in an innocent-suspect identification still exists, even when the procedures are
administered in a double-blind manner. For example, if a witness correctly rejected a lineup that
included four viable options1 and was prompted to re-examine the photos and reconsider their
initial decision, shifting to make a selection would result in a suspect identification 25% of the
time by chance alone. If the lineup had three only viable options, then the chance of shifting to
the suspect is 33%; and with two viable options, shifting from a rejection to a selection would
result in a suspect identification 50% of the time, purely by chance alone. Of course, if there is
only one true viable option in the group, shifting over to select a picture after initially rejecting
the lineup will presumably result in a suspect identification 100% of the time. Data from the
current study shows that once a witness has been influenced by the lineup administrator to shift
their decision and to select the suspect’s photo, confirming feedback can be very effective in
solidifying that altered decision.
Summary and Conclusions
Results of this study suggest that variations in viewing conditions can affect witnesses’
vulnerability to suggestive influence when making identification decisions. Although the current
study focused solely on how likely witnesses were to shift when prompted to reexamine the
lineup, these findings can also inform our understanding of witness vulnerability to other well
1 A viable option is a member of the line-up (filler or suspect) who matches the general characteristics of the culprit
that the witness is looking for Steblay & Wells (2020)
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 27
documented forms of pre-identification influence, such as steering witness to select a specific
suspect’s picture when viewing a photo array (Charman & Quiroz, 2016; Greathouse & Kovera,
2009; Kovera & Evelo, 2017), or offering social cues during an identification test that encourage
choosing (Clark et al., 2013).
This study also replicated the findings of Eisen et al., (2018) showing that prompting
people to reexamine the photos after they voice an initial decision but before that choice was
memorialized led a clear majority of witnesses to shift their identification decisions. Moreover,
data from these studies indicate that when these altered choices are reinforced with confirming
feedback, it is virtually impossible to distinguish identifications that were influenced by the
lineup administrator from uninfluenced decisions. In actual cases, witnesses will always
eventually learn that the government is so confident that the witness identified the actual culprit
that they are proceeding with the prosecution and need the witness to testify at trial and restate
their identification decision in court, in front of a jury. For many witnesses, learning that the
government is prosecuting the case can function as a potent source of confirming feedback.
The results of this study also expand our understanding of how estimator variables can
interact with system variables; in this case, how variations in the viewing conditions can make
witnesses more or less vulnerable to suggestive influence from the lineup administrator. Future
studies should examine other estimator variables that could influence the shift and stick effect
and other forms of suggestive influence.
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 28
References
Bradfield, A. L., Wells, G. L., & Olson, E. A. (2002). The damaging effect of confirming
feedback on the relation between eyewitness certainty and identification accuracy.
Journal of Applied Psychology, 87(1), 112–120. https://doi.org/10.1037/0021-
9010.87.1.112
Blank, H. (1998). Memory states and memory tasks: An integrative framework for eyewitness
memory and suggestibility, Memory, 6 (5), 481-529. https://doi.org/10.1080/741943086
Brewer, N. & Weber, N. (2008). Eyewitness confidence and latency: Indices of memory
processes not just markers of accuracy. Applied Cognitive Psychology, 22, 827-840.
https://doi.org/10.1002/acp.1486
Charman, S. D., & Quiroz, V. (2016). Blind sequential lineup administration reduces both false
identifications and confidence in those false identifications. Law and Human Behavior,
40(5), 477–487. https://doi.org/10.1037/lhb0000197
Charman, S. D., & Wells, G. L. (2012). The moderating effect of ecphoric experience on post-
identification feedback: A critical test of the cues-based inference. Applied Cognitive
Psychology, 26(2), 243–250. https://doi.org/10.1002/acp.1815
Clark, S. E. (2003). A memory and decision model for eyewitness identification. Applied
Cognitive Psychology, 17(6), 629–654. https://doi.org/10.1002/acp.891
Clark, S. E., Brower, G. L., Rosenthal, R., Hicks, J. M., & Moreland, M. B. (2013). Lineup
administrator influences on eyewitness identification and eyewitness confidence. Journal
of Applied Research in Memory and Cognition, 2(3), 158–165.
https://doi.org/10.1016/j.jarmac.2013.06.003
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 29
Clark, S. E., Marshall, T. E., & Rosenthal, R. (2009). Lineup administrator influences on
eyewitness identification decisions. Journal of Applied Research in Memory and
Cognition, 15(1), 63–75. https://doi.org/10.1037/a0015185
Deffenbacher, K. A., Bornstein, B. H., & Penrod, S. D. (2006). Mugshot exposure effects:
Retroactive interference, mugshot commitment, source confusion, and unconscious
transference. Law and Human Behavior, 30(3), 287–307. https://doi.org/10.1007/s10979-
006-9008-1
Dysart J. E., Lindsay R. C. L., Hammond, R., Dupuis, P. (2001). Mug shot exposure prior to
lineup identification: Interference, transference, and commitment effects. Journal of
Applied Psychology, 86(6), 1280–1284.
Eisen, M. L., Cedré, G. C., Williams, T. Q., & Jones, J. M. (2018). Does anyone else look
familiar? Influencing identification decisions by asking witnesses to re-examine the
lineup. Law and Human Behavior, 42(4), 306–320. https://doi.org/10.1037/lhb0000291
Eisen, M. L., Williams, J., Ying, R., Owen, J., & Jones, J. (2017). Does anyone else look
familiar? How subtle disconfirming feedback can alter identification decisions. Paper to
be presented at the American Psychology and Law Society, Seattle, Washington.
Fitzgerald, R. J., Rubínová, E., & Juncu, S. (accepted). Eyewitness identification around the
world. In A. M. Smith, M. P. Toglia, & J. M. Lampinen (Eds.), Methods, measures, and
theories in eyewitness identification tasks. Taylor and Francis.
Glanzer, M., Adams, J.K. (1985) The mirror effect in recognition memory. Memory &
Cognition 13, 8–20. https://doi.org/10.3758/BF03198438
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 30
Glanzer, M., Adams, J. K., Iverson, G. J., & Kim, K. (1993). The regularities of recognition
memory. Psychological Review, 100(3), 546–567. https://doi.org/10.1037/0033-
295X.100.3.546
Greathouse, S. M., & Kovera, M. B. (2009). Instruction bias and lineup presentation moderate
the effects of administrator knowledge on eyewitness identification. Law and Human
Behavior, 33(1), 70–82. https://doi.org/10.1007/s10979-008-9136-x
Kovera, M. B., & Evelo, A. J. (2017). The case for double-blind lineup administration.
Psychology, Public Policy, and Law, 23(4), 421–437.
https://doi.org/10.1037/law0000139
Lampinen, J. M., Erickson, W. B., Moore, K. N., & Hittson, A. (2014). Effects of distance on
face recognition: Implications for eyewitness identification. Psychonomic Bulletin &
Review, 21(6), 1489–1494. https://doi.org/10.3758/s13423-014-0641-2
Loftus, E. F. (2005). Planting misinformation in the human mind: A 30-year investigation of the
malleability of memory. Learning & Memory, 12(4), 361–366.
https://doi.org/10.1101/lm.94705
Memon, A., Bull, R., & Hope, L. (2003). Exposure duration: effects on eyewitness accuracy and
confidence. British Journal of Clinical Psychology, 94(Pt 3), 339-354.
https://doi.org/10.1348/000712603767876262
Neuschatz, J. S., Lawson, D. S., Swanner, J. K., Meissner, C. A., & Neuschatz, J. S. (2008). The
effects of accomplice witnesses and jailhouse informants on jury decision making. Law
and Human Behavior, 32(2), 137–149. https://doi.org/10.1007/s10979-007-9100-1
People vs. Carlos Cruz (2017). Los Angeles Superior Court, BA435151.
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 31
Pezdek, K., & Blandon-Gitlin, I. (2005). When is an intervening line-up most likely to affect
eyewitness identification accuracy? Legal and Criminological Psychology, 10(2), 247–
263. https://doi.org/10.1348/135532505X49846
Rodriguez, D. N., and Berry, M. A. (2013). "Eyewitness science and the call for double-blind
lineup administration". Journal of Criminology. https://doi.org/10.1155/2013/530523
Smith, A. M., Wilford, M. M., Quigley-McBride, A., & Wells, G. L. (2019). Mistaken
eyewitness identification rates increase when either witnessing or testing conditions get
worse. Law and Human Behavior, 43(4), 358–368. https://doi.org/10.1037/lhb0000334
Steblay, N. K., Wells, G. L., & Douglass, A. B. (2014). The eyewitness post identification
feedback effect 15 years later: Theoretical and policy implications. Psychology, Public
Policy, and Law, 20(1), 1–18. https://doi.org/10.1037/law0000001
Steblay, N. K., & Dysart, J. E. (2016). Repeated eyewitness identification procedures with the
same suspect. Journal of Applied Research in Memory and Cognition, 5(3), 284–
289. https://doi.org/10.1016/j.jarmac.2016.06.010
Tousignant, J. P., Hall, D., & Loftus, E. F. (1986). Discrepancy detection and vulnerability to
misleading postevent information. Memory & Cognition, 14(4), 329–338. https://doi.org/
10.3758/BF03202511
Wells, G. L., & Bradfield, A. L. (1998). “Good, you identified the suspect”: Feedback to
eyewitnesses distorts their reports of the witnessing experience. Journal of Applied
Psychology, 83(3), 360–376. https://doi.org/10.1037/0021-9010.83.3.360
Wells, G. L., & Luus, C. E. (1990). Police lineups as experiments: Social methodology as a
framework for properly conducted lineups. Personality and Social Psychology Bulletin,
16(1), 106–117. https://doi.org/10.1177/0146167290161008
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 32
Wells, G. L., Bull Kovera, M., Bradfield Douglas, A, Brewer, N., Meissner, C. A., Wixted, J. T.
(2020). Policy and Procedure Recommendations for the Collection and Preservation of
Eyewitness Evidence. Law and Human Behavior, 44(1), 3-36.
http://dx.doi.org/10.1037/lhb0000359
Wixted, J. T. & Gaitan, S. (2002). Cognitive theories as reinforcement history surrogates: The
case of likelihood ratio models of human recognition memory. Animal Learning &
Behavior, 30, 289-305.
Wixted, J. T., & Wells, G. L. (2017). The relationship between eyewitness confidence and
identification accuracy: A new synthesis. Psychological Science in the Public Interest,
18(1), 10–65. https://doi.org/10.1177/1529100616686966
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE 33
Table 1. Identification Decisions for Close versus Distant Viewing Conditions
13 Foot 30 Foot
ID Viewing condition Viewing condition
n = 203 n = 208 .
Culprit-present
Culprit ID 60.4% 35.8%
Filler ID 30.7% 51.9%
No ID 8.9% 12.3%
Culprit-absent
Filler ID 84.1% 90.7%
No ID 15.9% 9.3%
34
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE
Table 2
Rates of Shifting by Viewing Condition for Accurate and Inaccurate Participants who Made
Initial Identification Decisions Over and Under 12-seconds
Viewing Conditions
13 Feet 30 Feet
Shift Rate Shift Rate
n = 208 n = 208
All Participants 68.8% (143/208) 73.4% (149/203)
1st Choice Accurate (all) 55.1% (43/78) 68.1% (32/47)
Under 12-seconds 46.3% (25/54) 60.0% (18/30)
Over 12-seconds 75.0% (18/24) 82.4% (14/17)
1st Choice Error (all) 76.9% (100/130) 75.0% (117/156)
Under 12-seconds 75.0% (54/72) 65.3% (66/101)
Over 12-seconds 77.6% (45/58) 92.7% (51/55)
35
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE
Table 3
Pattern of Shifting for Culprit/Suspect IDs, Filler Picks, and Rejections, From the Initial Decision Through the Final Identification
All Shifters-Second Choice Final Identification Decision (Room #2)
Initial Participants Culprit/ Culprit/
ID Decision who Shifted Suspect ID Filler ID No ID Suspect ID Filler ID No ID
% (n) % (n) % (n) % (n) % (n) % (n) % (n) % (n)
30 Feet (n = 203)
CP-Culprit ID 35.8% (38) 65.8% (25) ----- ----- 100% (25) 0% (0) 44.0% (11) 56.0% (14) 0% (0)
Filler ID 51.9%. (55) 76.4% (42) 47.6% (20) 52.4% (22) 0% (0) 47.6% (20) 50.0% (21) 2.4% (1)
No ID 12.3% (13) 92.3% (12) 66.7% (8) 33.3% (4) ----- ----- 50.0% (6) 41.7% (5) 8.3% (1)
CA-Suspect ID* 47.4% (46) 65.2% (30) ----- ----- 100% (30) 0% (0) 46.7% (14) 53.3% (16) 0% (0)
Filler ID 43.3% (42) 71.6% (33) 48.5% (16) 98.4% (16) 1.6% (1) 33.3% (11) 66.7% (22) 0.0% (0)
No ID 9.3% (9) 77.8% (7) 71.4% (5) 28.6% (2) ----- ----- 71.4% (5) 14.3% (1) 14.3% (1)
13 Feet (n = 208)
CP-Culprit ID 60.4% (61) 52.5% (32) ----- ----- 96.9% (31) 3.1% (1) 59.4% (19) 40.6% (13) 0.0% (0)
Filler ID 29.7% (30) 80.0% (24) 45.8% (11) 54.2% (13) 0.0% (0) 33.3% (8) 66.7% (16) 0.0% (0)
No ID 8.9% (9) 66.7% (6) 50.0% (3) 50.0% (3) ----- ----- 50.0% (3) 33.3% (2) 16.7% (1)
CA- Suspect ID 28.0% (30) 70.0% (21) ----- ----- 100% (21) 0.0% (0) 47.6% (10) 52.4% (11) 0.0% (0)
Filler ID 56.5% (61) 80.3% (49) 40.8% (20) 58.3% (28) 0.0% (1) 28.6% (14) 71.4% (35) 0.0% (0)
No ID 15.9% (17) 64.7% (11) 18.2% (2) 81.8% (9) ----- ----- 18.2% (2) 54.5% (6) 27.3% (3)
Suspect ID* = Worst-Case Scenario. Suspect was the most selected filler at each distance.
36
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE
Table 4
Final Identification Decisions
Final Decisions for All Shifters Final Decisions for All Non-Shifters
Committed Reverted Changed Consistently Changed
to the photo to the to select chose the to select
they shifted to initial selection a new photo original selection a new photo
% (n) % (n) % (n) % (n) % (n)
30 Feet: All Participants 49.7% (74) 34.9% (52) 15.4% (23) 94.4% (51) 5.6% (3)
Confirming Feedback 63.2% (48) 32.9% (25) 3.9% (3) 95.8% (23) 4.2% (1)
No Feedback 35.6% (26) 37.0% (27) 27.4% (20) 93.3% (28) 6.7% (2)
13 Feet: All Participants 48.3% (69) 41.3% (59) 10.5% (15) 84.6% (55) 15.4% (10)
Confirming Feedback 63.9% (46) 36.1% (26) 0.0% (0) 90.3% (28) 9.7% (3)
No Feedback 32.4% (23) 46.5% (33) 21.1% (15) 79.4% (27) 20.6% (7)
37
Running Head: VULNERABILITY TO SUGGESTIVE INFLUENCE
Table 5
Confidence in Accurate and Inaccurate Final Decisions for Shifters and Non-Shifters
Shifters Non-Shifters
Confirming No All Confirming No All
Feedback Feedback Shifters Feedback Feedback Non-shifters
(n = 148) (n = 144) (n = 292) (n = 55) (n = 64) (n = 119)
M (SD) M (SD) M (SD) M (SD) M (SD) M (SD)
30 feet: All participants 75.37 (18.83) 64.01 (18.92) 69.81 (19.65) 79.92 (22.29) 78.25 (19.87) 78.99 (20.79)
Accurate 76.42 (17.44) 60.53 (17.94) 68.47 (19.22) 80.43 (21.79) 84.75 (9.95) 82.73 (16.06)
Inaccurate 75.02 (19.41) 65.24 (19.27) 70.26 (19.87) 79.71 (23.15) 75.88 (22.14) 77.55 (22.37)
13 feet: All participants 77.36 (19.44) 64.57 (21.03) 71.01 (21.17) 84.35 (16.78) 73.03 (21.32) 78.43 (19.97)
Accurate 78.93 (23.71) 68.47 (20.35) 72.91 (22.11) 90.70 (14.11) 81.15 (15.77) 85.08 (15.64)
Inaccurate 76.98 (18.49) 63.14 (21.29) 70.44 (20.95) 79.12 (17.37) 61.43 (23.32) 71.13 (21.83)
Total: All participants 76.34 (19.09) 64.29 (19.92) 70.39 (20.39) 82.41 (19.31) 75.47 (20.66) 78.68 (20.26)
Accurate 77.48 (20.03) 64.50 (19.35) 70.54 (20.59) 87.28 (17.22) 82.18 (14.27) 84.36 (15.64)
Inaccurate 76.01 (18.89) 64.21 (20.21) 70.35 (20.37) 79.41 (20.15) 70.26 (23.39) 74.71 (22.20)
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Two experiments were conducted to examine whether misidentifying an innocent suspect with a unique feature (i.e., facial tattoo) would impact witnesses’ recollections of the culprit so that they would mistakenly describe the culprit as having that same feature, when in reality, he did not. In both experiments, participants viewed a video of a simulated carjacking in which the perpetrator’s face was visible from a close or far distance, and were then led to misidentify an innocent suspect from a suggestive culprit-absent photo-array. The innocent suspect either had a tattoo photoshopped onto his face, or, like the culprit, had no visible facial markings. After making the identification, half the participants received confirming post-identification feedback. Finally, participant-witnesses were asked to describe the culprit in their own words. In both experiments, over a third of the participant-witnesses in the tattoo condition mistakenly described the culprit as having had a tattoo on his face. In Experiment 2, remember/know judgements indicated that participant/witnesses in the tattoo condition were also more likely to report remembering that the culprit had a face tattoo, rather than simply knowing this to be true. Also, as predicted, in both experiments, confirming feedback significantly boosted erroneous reports of recalling the tattoo.
Article
Full-text available
Objective: The Executive Committee of the American Psychology-Law Society (Division 41 of the American Psychological Association) appointed a subcommittee to update the influential 1998 scientific review paper on guidelines for eyewitness identification procedures. Method: This was a collaborative effort by six senior eyewitness researchers, who all participated in the writing process. Feedback from members of AP-LS and the legal communities was solicited over an 18-month period. Results: The results yielded nine recommendations for planning, designing, and conducting eyewitness identification procedures. Four of the recommendations were from the 1998 article and concerned the selection of lineup fillers, prelineup instructions to witnesses, the use of double-blind procedures, and collection of a confidence statement. The additional five recommendations concern the need for law enforcement to conduct a prelineup interview of the witness, the need for evidence-based suspicion before conducting an identification procedure, video-recording of the entire procedure, avoiding repeated identification attempts with the same witness and same suspect, and avoiding the use of showups when possible and improving how showups are conducted when they are necessary. Conclusions: The reliability and integrity of eyewitness identification evidence is highly dependent on the procedures used by law enforcement for collecting and preserving the eyewitness evidence. These nine recommendations can advance the reliability and integrity of the evidence. (PsycINFO Database Record (c) 2020 APA, all rights reserved).
Article
Full-text available
We examined how giving eyewitnesses a weak recognition experience impacts their identification decisions. In 2 experiments we forced a weak recognition experience for lineups by impairing either encoding or retrieval conditions. In Experiment 1 (n = 245), undergraduate participants were randomly assigned to watch either a clear or a degraded culprit video and then viewed either a culprit-present or culprit-removed lineup identification procedure. In Experiment 2 (n = 227), all participants watched the same clear culprit video but were then randomly assigned to either view a clear or noise-degraded lineup procedure. Half of the participants viewed a culprit-present lineup procedure and the remaining participants viewed a culprit-removed lineup procedure. Not surprisingly, degrading either encoding or retrieval conditions led to a sharp drop in culprit identifications. Critically, and as predicted, degrading either encoding or retrieval conditions also led to a sharp increase in the identification of innocent persons. These results suggest that when a lineup procedure gives a witness a weak match-to-memory experience, the witness will lower her criterion for making an affirmative identification decision. This pattern of results is troubling because it suggests that witnesses who encounter lineups that do not include the culprit might have a tendency to use a lower criterion for identification than do witnesses who encounter lineups that actually include the culprit. (PsycINFO Database Record (c) 2019 APA, all rights reserved).
Article
Full-text available
Two experiments were conducted to see if asking witnesses to take another look at the lineup after they voiced their identification decisions would alter their choices, and if confirming feedback could then be used to solidify the selections they shifted to. Participants watched a simulated crime and were asked to identify the culprit from a photographic lineup. After voicing their identification decisions, participants were prompted to reexamine the lineup. Half of the participants then received confirming feedback for their decisions, regardless of whether they shifted to a new picture or not. Later on, a different experimenter escorted participants to a second room and administered the same lineup again. In Experiment 1 (N = 432), biased instructions were used to encourage choosing, and when participants were prompted to reexamine the lineup, 70% changed their identification decisions and selected a different picture. When that new selection was reinforced with feedback and participants were given a second opportunity to identify the culprit at a later time, 72% selected the picture they shifted to as the culprit. Participants who made their decisions more quickly were less likely to shift, but accuracy did not predict shifting. This general pattern of findings was replicated using unbiased instructions in Experiment 2 (N = 237). Results suggest that prompting witnesses to reexamine the lineup can often lead witnesses to change their identification decisions, and when the altered choice is reinforced, they will often stay with that influenced decision over time, asserting it with a high degree of confidence.
Article
Full-text available
Many have recommended that lineups be conducted by administrators who do not know which lineup member is the suspect (i.e., a double-blind administration). Single-blind lineup administration, in which the administrator knows which lineup member is the suspect, increases the rate at which witnesses identify suspects, increasing the likelihood that both innocent and guilty suspects are identified. Although the increase in correct identifications of the guilty may appear desirable, in fact, this increase in correct identifications is the result of impermissible suggestion on the part of the administrator. In addition to these effects on witness choices, single-blind administration influences witness confidence through an administrator’s feedback to witnesses about their choices, reducing the correlation between witness confidence and accuracy. Finally, single-blind administration influences police reports of the witness’s identification behavior, with the same witness behavior resulting in different outcomes for suspects depending upon whether the administrator knew which lineup member was the suspect. Administrators who know which lineup member is the suspect in an identification procedure emit behaviors that increase the likelihood that witnesses will choose the suspect, primarily by causing witnesses who would have chosen a filler (known innocent member of the lineup who is not the suspect) to choose the suspect. To avoid impermissible suggestion, photo arrays and lineups should be administered using double-blind procedures.
Article
Full-text available
The U.S. legal system increasingly accepts the idea that the confidence expressed by an eyewitness who identified a suspect from a lineup provides little information as to the accuracy of that identification. There was a time when this pessimistic assessment was entirely reasonable because of the questionable eyewitness-identification procedures that police commonly employed. However, after more than 30 years of eyewitness-identification research, our understanding of how to properly conduct a lineup has evolved considerably, and the time seems ripe to ask how eyewitness confidence informs accuracy under more pristine testing conditions (e.g., initial, uncontaminated memory tests using fair lineups, with no lineup administrator influence, and with an immediate confidence statement). Under those conditions, mock-crime studies and police department field studies have consistently shown that, for adults, (a) confidence and accuracy are strongly related and (b) high-confidence suspect identifications are remarkably accurate. However, when certain non-pristine testing conditions prevail (e.g., when unfair lineups are used), the accuracy of even a high-confidence suspect ID is seriously compromised. Unfortunately, some jurisdictions have not yet made reforms that would create pristine testing conditions and, hence, our conclusions about the reliability of high-confidence identifications cannot yet be applied to those jurisdictions. However, understanding the information value of eyewitness confidence under pristine testing conditions can help the criminal justice system to simultaneously achieve both of its main objectives: to exonerate the innocent (by better appreciating that initial, low-confidence suspect identifications are error prone) and to convict the guilty (by better appreciating that initial, high-confidence suspect identifications are surprisingly accurate under proper testing conditions).
Article
Full-text available
One of the most recommended procedures proposed by eyewitness experts is the use of double-blind lineups, in which the administrator does not know the identity of the suspect in the lineup. But despite the near universality of this recommendation, there is surprisingly little empirical research to support the claim that nonblind administration inflates false identifications. What little research has been conducted has shown conflicting findings with regard to the conditions under which nonblind administration affects false identifications, as well as its effects on witness confidence. The current study attempts to elucidate this effect. Student-participants (n = 312) were randomly assigned to play the role of either a lineup administrator (who were either told the identity of the suspect in the lineup or not) or a mock crime witness. Following unbiased instructions, administrators presented either a target-present or target-absent sequential lineup to the witness while being surreptitiously videorecorded. Nonblind administration significantly inflated false, but not correct, identifications, and significantly inflated witness confidence in those false identifications. Video recordings indicated that nonblind administrators were significantly more likely than blind administrators to smile (a) while the witness was viewing a photograph of the suspect, and (b) after a suspect identification. Results provide stronger support for the use of blind lineup administration by broadening the conditions under which nonblind administration is shown to inflate false identifications. Possible reconciliations for conflicting findings in the literature are discussed. (PsycINFO Database Record
Article
Full-text available
Eyewitnesses’ retrospective reports of certainty, view, attention, and other judgments constitute central variables used by courts to assess the credibility of eyewitness identification evidence. Recently, important state Supreme Court decisions (e.g., New Jersey v. Henderson, 2011; Oregon v. Lawson, 2012) have relied on psychological research regarding the post-identification feedback effect to revamp their assumptions about when witness retrospective self-reports can be trusted. The post-identification feedback effect, originally demonstrated by Wells and Bradfield (1998), refers to the way in which witness self-reports are distorted by feedback to the witnesses that suggests that their identifications were accurate or mistaken. We present a meta-analysis of the post-identification effect involving 7,000 participants from the United States, Canada, Europe, and Australia. The results show that confirming feedback robustly inflates eyewitnesses’ retrospective judgments across experimental manipulations and laboratory settings with large effect sizes. We describe the policy implications of the feedback effect with regard to the need for double-blind lineup procedures. Moreover, we propose that testimony-relevant witness judgments should be collected and documented, preferably with videotape, before feedback can occur. We use moderator analyses to examine the current explanation of the feedback effect and delineate new research questions that could help develop a more complete theoretical understanding of the processes giving rise to the effect. (PsycINFO Database Record (c) 2014 APA, all rights reserved)
Article
Lineup administrators were trained to respond to witnesses in such a way as to redirect them from making non-identifications or foil identification responses toward making identifications of the suspect. Compared to a no-influence control condition, suspect identification rates in the influence condition increased substantially and proportionally for guilty and innocent suspects. Administrators steered witnesses more specifically toward the suspect when the suspect was guilty than when the suspect was innocent. Post-identification confidence for correct identifications of the guilty suspect did not differ significantly across the influence and no-influence groups. However, post-identification confidence for false identifications of the innocent suspect was significantly lower for the influence group than for the no-influence group because witnesses who were influenced to make false identifications tended to be those who were less confident prior to the lineup, and also because those witnesses became less confident from pre- to post-identification.