ArticlePDF Available

Perceptions of Movement Patterns and Concealment Detection in Naive Observers and Law Enforcement Officers: A Lens Model Analysis

Authors:

Abstract

This research investigates whether police officers can reliably use behavioral cues to determine whether a person is conceal- ing an object. Using a Lens Model framework, we performed a mega-analysis of three experiments. In each study, officers and laypersons judged whether people were concealing an object and reported “articulable behaviors” they used to perform this task. Although participants were able to articulate behaviors that they believed were helpful, results showed that these behaviors were not related to whether the person was actually concealing. Officers and laypersons were equally poor at judging whether someone was concealing or not. Current officer training on the use of nonverbal behaviors to determine who is concealing a dangerous object may be ineffective, and a reconsideration of training is warranted. In light of the findings, requiring officers to provide “articulable behaviors” in Fourth Amendment cases may not provide a sufficient safeguard against unreasonable searches of civilians.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
1
Perceptions of Movement Patterns and Concealment Detection in Naïve Observers and Law
Enforcement Officers: A Lens Model Analysis
Dawn M. Sweet1, Adele Quigley-McBride2, Christian A. Meissner3, & Katharine Ringstad4
1 Department of Psychology & Communication, University of Idaho
2 Wilson Center for Science and Justice, Duke University School of Law
3 Department of Psychology, Iowa State University
4 Drake University
In Press at Criminal Justice & Behavior
Author Note
This research was funded by a grant from the Motorola Solutions Foundation.
Dawn M. Sweet https://orcid.org/0000-0002-3825-540X
Adele Quigley-McBride https://orcid.org/0000-0001-6940-8707
Christian A. Meissner https://orcid.org/0000-0002-6094-5167
Correspondence concerning this article should be addressed to: Dr. Dawn M. Sweet, Dept of
Psychology & Communication, University of Idaho MS 3043, Moscow, ID 83844-3043. Email:
sweet@uidaho.edu
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
2
Abstract
This research investigates whether police officers can reliably use behavioral cues to determine
whether a person is concealing an object. Using a Lens Model framework, we performed a
mega-analysis of three experiments. In each study, officers and laypersons judged whether
people were concealing an object and reported “articulable behaviors” they used to perform this
task. Although participants were able to articulate behaviors that they believed were helpful,
results showed that these behaviors were not related to whether the person was actually
concealing. Officers and laypersons were equally poor at judging whether someone was
concealing or not. Current officer training on the use of nonverbal behaviors to determine who is
concealing a dangerous object may be ineffective and a reconsideration of training is warranted.
In light of the findings, requiring officers to provide “articulable behaviors” in Fourth
Amendment cases may not provide a sufficient safeguard against unreasonable searches of
civilians.
Keywords: Fourth Amendment, Lens Model; nonverbal behavior; judgment; police training;
decision-making.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
3
Perceptions of Movement Patterns and Concealment Detection in Naïve Observers and law
Enforcement Officers: A Lens Model Analysis
Without visual confirmation of a weapon, law enforcement officers often use behavior-
based cues to determine if an individual is concealing a dangerous or illegal object. An officer’s
ability to accurately detect who is concealing a firearm and to articulate why they believe that to
be the case can have consequences under the Fourth Amendment of the U.S. Constitution, which
asserts a citizen’s right to be free from “unreasonable search and seizure.” In Terry v. Ohio
(1968), the Court determined that, before performing a search, officers must have a reasonable
suspicion that a person has committed, is committing, or is about to commit a crime, as well as a
reasonable belief that the person may be “armed and dangerous.” The exact meaning of
“reasonable suspicion” is unclear, but officers are expected to reference behaviors, movements,
observations, and other “articulable behaviors” or “articulable suspicions” to justify their
decision to perform a search. In other words, for a search to be considered reasonable under the
Fourth Amendment, officers must be able to explain their decision to search someone for a
dangerous object, such as a firearm, and their explanation must include specific, articulable
behavioral evidence before their justification can be considered valid.
Officers have a particularly challenging task when patrolling communities. They must
remain alert so that they can watch for suspicious behavior and signs of criminal activity while
serving their primary goal, namely protecting the public. Unlike other public safety services such
as aviation security, officers do not generally have access to supportive technologies that are
very effective and accurate with respect to identifying concealed prohibited objects. In other
words, those working in aviation security are not required to rely solely on their experience and
observation skills to identify dangerous concealed object because they can use metal detectors,
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
4
full body scanners, and x-ray machines to facilitate the identification of concealed prohibited
items. Although extremely useful, these kinds of technological aides cannot be integrated into an
officer’s normal routine in a practical way (for example, Chen, et al., 2005; Sheen, et al., 2001).
Instead, when officers decide to stop and search someone on the street without a warrant,
they must rely on the observable and articulable behaviors, movements, and other information
about a suspicious person that can be determined from afar. It is these subjective behavioral
judgments that form the basis of the “reasonable suspicion” rule and result in reasonable or
unreasonable contact with a citizen. To perform this task well, officers rely upon their perception
and interpretation of behavioral cues derived from their experience and any training they may
have received. Thus, research addressing the ability of officers to detect concealment in contexts
that trigger Fourth Amendment issues could help establish whether police officers can reliably
discriminate the movement patterns of those who are concealing a prohibited object from those
who are not, and which behaviors are most diagnostic of concealment.
As is the case with many laws, the legal requirements associated with Terry stops,
instances in which officers stop someone and perform a warrant-less search based on something
they observed about the person, are not based on scientific evidence. Officers report drawing on
their training and experience with concealment cases in situations that invoke the Terry rule. For
instance, in US v. Briggs (2013), the defendant “repeatedly looked over his shoulder at the
officers…grabbed at the waistlines of his pants” and, as the officer moved closer, sped up. Based
on their training and experience, the officer felt this was the type of behavior exhibited by
someone concealing a weapon at their waistline. The court agreed that the stop was justified, but
this decision was based on a “common sense,” “totality of the circumstances” legal approach, not
empirical evidence demonstrating that such behaviors are, in fact, reliably and consistently
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
5
associated with weapon concealment. Further, the Supreme Court expressly recognized the lack
of “empirical studies dealing with inferences drawn from suspsious behavior, and … cannot
reasonably demand scientific certainty when none exists” (Judge Rehnquist; Illinois v. Wardlow,
2000).
As such, whether officers can accurately and reliably detect when an individual is
concealing a dangerous object remains an empirical uncertainty. There are currently no empirical
data suggesting that police officers are any better at detecting a concealed weapon than
laypersons (e.g., Sweet et al., 2017) and there is no experimental, scientific evidence indicating
that any articulable behavior patterns are diagnostic of this type of deception. As should be the
goal for all work evaluating the tasks that officers perform (e.g., determining if an individual is
lying, Bond & DePaulo, 2006), there is a need for evidence-based assessments of judgments in
law enforcement settings, rather than relying solely on officers’ experience, gut instinct, or
anecdotal evidence.
With this goal in mind, the current research examines three datasets that investigate
perceptions and interpretations of behavioral cues believed to be associated with attempts to
conceal a firearm, an unstable device, and an innocuous object, and whether these behavioral
cues are, in fact, associated with individuals who are concealing versus those who are not. In
doing so, we seek to identify the predictive value of various gait, arm, and head behaviors that
have been subjectively reported by observers and objectively coded by researchers and that
might serve as diagnostic indicators of object concealment. Subjective judgments in these studies
were provided by both officers and laypersons, permitting an assessment of any differences in
behavioral observation and discrimination ability between these groups.
Support for Detecting Concealment with Articulable Behaviors
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
6
As already noted, there exists no empirical evidence of any reliable, nonverbal cues that
officers can use to identify suspicious movement patterns that are indicative of object
concealment (Sweet et al., 2017), particularly concealment of a prohibited object. Decisions
made by CCTV operators and laypersons have been examined to investigate the extent to which
people can use their perceptions of affective states (i.e., current emotional experience) exhibited
by observed others to determine if people viewed on CCTV footage were concealing a firearm, a
bottle, or nothing in the hand that was occluded from the video camera (Blechko, et al., 2008).
The study found that judgments about affective states were not reliable indicators of concealment
and there were no significant differences in performance on the detection task between CCTV
operators and laypersons or based on the observer’s level of experience and training.
Sweet and colleagues (2017) investigated whether officers would perform better than
untrained laypersons at the task of identifying individuals who may be concealing a firearm or an
unstable device. Consistently, across three studies with two different concealed objects, officers’
performance was not significantly better than untrained laypersons, suggesting that their training
and experience was not as useful as is typically believed. Moreover, when making judgments
about whether a person had a concealed weapon, officers with more years of experience seemed
to have a greater tendency to judge a person as concealing, though there was no significant
association between years of experience and accuracy. In fact, years of experience was
negatively correlated with officer accuracy—officers with more years of experience made fewer
correct “concealing” and “not concealing” judgments than less experienced officers. Thus, more
training and years of experience might result in a bias to see evidence of concealing.
Previous research also shows that people are not generally successful at concealing
information by altering their gait (e.g., concealing your gender or the weight of an object when
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
7
lifting it; e.g., Runeson & Frykholm, 1983). When people observe the gait movements of others,
they are adept at inferring something about the person’s gender (e.g., Pollick, et al., 2005), age
(e.g., Montepare & Zebrowitz-McArthur, 1988), and victim status (e.g., Wheeler, et al., 2009).
However, people are less adept when making judgments about whether a person is concealing
something – any type of object.
When people consciously attempt to deceive, their behavior is said to be marked by a
sense of unusual deliberateness (DePaulo, et al., 2003). That is, their behavior does not appear
“typical” or “natural” in the particular context, or some of the kinematic details of a genuine or
typical behavior pattern are missing (e.g., Runeson & Frykholm, 1983). Although not skilled at
determining whether someone is concealing something, observers do appear to be sensitive to
whether movement patterns are typical or atypical (Richardson & Johnston, 2005). Why might
we expect movement patterns to vary when someone is concealing an object? Gait and other
movement patterns may differ because “typical” or “natural” movement patterns are not
sustainable due to the nature or size of the object being concealed. When an individual knows
that they are trying to conceal something, this knowledge could also lead to a change in their
movements.
The findings offered by Sweet et al. (2017) and Blechko et al. (2008) suggest there is
little reason to believe that officers, even those trained or experienced in such detection tasks,
can render accurate judgments about whether people are concealing a potentially dangerous
object. However, these studies do not speak to whether there are any reliable nonverbal
behaviors associated with the concealment of objects—only that officers cannot reliably
distinguish between people who are and are not concealing. If reliable behavioral cues—or
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
8
“articulable behaviors” under the 4th Amendment law—do exist, research is needed to establish
what these are and the relevant boundary conditions under which they operate.
Officer Training Regarding Articulable Behaviors
Research and training efforts relevant to nonverbal behavior and law enforcement have
primarily focused on detecting deception in investigative interview settings (e.g., Vrij, 2008) or
detecting suspicious or criminal behavior in security screening contexts (e.g., Burgoon et al.,
2009). While officers are trained to recognize movement patterns associated with concealment,
this training is typically grounded in anecdotal evidence (e.g., Meehan & Strange, 2021;
Greenbaum, 2014). While publications exist that provide suggestions about which behavioral
cues may indicate whether a person is concealing, these recommendations are not based on
empirical data and must be interpreted with caution (Meehan & Strange, 2021).
Consider, for example, the Bureau of Alcohol, Tobacco, Firearms, and Explosives (ATF)
“Project Safe Neighborhoods Enforcement Training” (https://www.atf.gov/firearms/project-safe-
neighborhoods-enforcement-training-psn-3-day-program), which provides sworn officers with a
training course that includes identifying the “characteristics of an armed gunmen” (Track 3,
Street Enforcement). Again, this training only provides officers with anecdotal evidence of
movement patterns associated with weapon concealment; however, as stated above, courts tend
to find that this type of anecdotal evidence (or “articulable behaviors”), to be sufficient evidence
that officers can reliably and accurately determine who is concealing a firearm and who is not
(e.g., The Commonwealth of Massachusetts vs. Evelyn).
There exists one non-empirical study by Meehan and Strange (2015) in which the
researchers identified behaviors commonly believed by officers to be associated with concealing
weapons or drugs. During the first phase, the authors identified articulable behaviors used by
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
9
subject matter experts to determine when individuals were concealing a weapon or other
contraband by interviewing those experts and attending police training sessions. The behaviors
identified using these non-empirical methods included an atypically short stride, restricted arm
swing, scanning the environment, or constantly adjusting clothing.
For the second phase, the authors asked pairs of officers to role play scenarios that
intended to mimic a concealment context, with one role player acting as the observer and the
other pretending to be an individual who was concealing an object. The findings from this role-
play were used to create a Universal Interdiction Framework (e.g., Meehan, Strange, & McClary,
2015). The authors recognized that these behaviors were not identified using empirical research
and recommended a “systematic evaluation of these assertions … to determine the extent to
which these behaviors occur in an operational environment” (p. 12). Meehan, Strange, and
Garinther (2021) confirmed these conclusions with an additional set of qualitative interviews
from subject matter experts, identifying similar, behavioral indicators. Yet, Meehan and
colleagues (2021) again state that these data are not empirical and identify a need for “validating
and evaluating the utility of these indicators for law enforcement” (p. 476).
The Present Studies
The existing peer-reviewed literature provides no assurance that there are articulable
behaviors that can be used to determine when an individual is (or is not) concealing a prohibited
object. As such, our goal was to empirically investigate whether officers are able to identify
articulable behavioral cues that, when present, consistently indicate the individual is likely to be
concealing a prohibited object.
To do this, the present studies investigate perceptions and interpretations of behavioral
cues believed to be associated with attempts to conceal a firearm, an unstable device, and an
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
10
innocuous object. We aim to determine whether these articulable behaviors were, in fact, present
more frequently among individuals who were concealing compared with those who were not,
and whether police officers or lay persons report using these behavioral cues. While officers do
not appear to render more accurate concealment judgments than laypersons, it is possible that
officers attend to different articulable behaviors than laypersons because of their training and
experience. Further, it may be possible to objectively code and empirically identify behaviors
that do reliably discriminate between those who are concealing and those who are not.
In the current paper, we extend prior work by conducting a novel experiment in which
law enforcement and laypersons observe a series of individuals who are randomly assigned to
conceal or not conceal a red handkerchief in their pocket. This procedure was used to model the
concealment of a smaller object that a person might wish to hide from law enforcement, such as
a small bag of illicit drugs. Because the object in this experiment had little or no bulk, we
reasoned that the results could be best explained by the incentive and motivation to conceal the
object. Observers were tasked with providing a concealment judgment and then reporting the
nonverbal behaviors and movement patterns that supported their judgments.
We combine these new data described above with additional unpublished behavioral data
from two experiments previously described by Sweet et al. (2017). In Experiment 1 of Sweet et
al., observers judged whether a person of interest was concealing a concealed weapon (Glock
model 19, 9mm) with the firing pin removed. In Experiment 3 of Sweet et al., one of two
individuals in each stimulus video was concealing an untethered water bucket in a backpack
while walking through a crowded sidewalk. The bucket of water was used to mimic an unstable
device or live bomb.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
11
The combination of behavioral data from these three studies allows us to assess whether
articulable behaviors, noted by officers and laypersons in support of their concealment
judgments, are associated with trials in which an object was being concealed (or not). For each
experiment, we analyze the movement patterns that our participants proffered were indicative of
concealing (or not). We also assess objective data from trained coders who observed the same
recordings. Finally, we present the performance data from Experiment 1 comparing the detection
performance of officers and laypersons, and we contextualize this with the performance data
previously published by Sweet et al. (2017). Our goal was to examine whether officers and
laypersons report using articulable behaviors that can accurately discriminate between
individuals who are concealing an object and those who are not, regardless of the object in
question or the context (e.g., non-crowded versus crowded, outdoors versus indoors).
Method
The current study combines data from three experiments using a mega-analytic approach:
Experiment 1 involves new data collected and reported here, as well as previously unpublished
data from two studies conducted by Sweet et al. (2017, Experiments 1 & 3). For each
experimental session, participants viewed multiple videos and were asked to make judgments
about whether a person was concealing an object or not. Individuals who concealed objects (or
not) in each experiment did not appear in the videos used in other experiments presented here, so
each study had a unique set of individuals that participants observed.
The focus of the current paper is the data provided by participants when they were asked
to provide reasons for their decisions. Both officers and lay participants were told to report any
articulable behaviors that they observed and were used to inform their concealment decision.
This method allowed us to examine several key relationships between behaviors that objective
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
12
coders reported (objectively coded behaviors) and those that the officers and lay participants
reported (articulable behaviors). We also assessed whether these behavioral observations were
associated with trials in which the individual was actually concealing an object (ground truth)
and participants’ subjective judgments about whether they believed the individual in the video
was concealing an object (conceal judgments). Below we describe the method of each
experiment, after which we detail the behavioral coding that is unique and novel to the current
study, as well as the theoretical framework (Brunswikian Lens Model; see Brunswik, 1952;
1956) and analytic approach (mega-analysis; see, for example, Curran & Hussong, 2009;
Eisenhower, 2021) that was used to evaluate the data across experiments. All data and R code
used to produce these results, as well as additional details about the coding protocol, can be
found on Open Science Framework (https://osf.io/jptkb/) and in the Supplementary Materials.
Experiment 1: Detection of an Innocuous Object
Participants. Study participants included 50 law enforcement officers from a
Midwestern state in the United States (25 women) and 70 students (40 women) from a university
in the Midwestern United States. Officers ranged in age from 24 to 62 (Mofficer-age = 38.00; SD
officer-age = 7.60) and had a mean of 12.38 (SDofficer-experience = 9.68) years of experience. Officers
were recruited using snowball sampling and were paid $50 per hour to participate. Student
participants ranged in age from 18 to 43 years (Mstudent-age = 29.15; SDstudent-age = 3.25) and were
recruited through the Psychology Department’s participant pool in return for course credit. This
study and all studies discussed were approved by the Institutional Review Board.
Study Design. A repeated measures within-subjects experimental design was used in
which participants viewed nine videos that showed either a man walking through a room or a
woman walking through a room (7 men and 2 women in total). Each person either concealed or
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
13
did not conceal an innocuous object. Each person was video-recorded walking through the room
only once.
Stimulus Materials. Stimulus materials included nine video clips with an average length
of 23.77 seconds. The nine videos included seven men and two women. Each person was
recorded only once and filmed concealing a handkerchief or concealing nothing as they walked
through a room. Each video includes two angles: The first angle was the person walking toward
the camera and the second angle was the person recorded from the side. These two angles were
edited into a single video for viewing. A uniformed officer was also in the room making real-
time judgments as each person walked through the room. The seven men and two women were
told that they would earn a monetary award if they were able to convince the officer that they
were not concealing an object. Each person was given $20 before their turn to walk through the
room but were told that, if the officer correctly judged them to be concealing (or not concealing)
the object, they would have to forfeit the $20. In the end, each person was allowed to keep the
$20 and received a $10 bonus.
Procedure. Study participants were instructed that they would view several videos of
someone walking through a room, and that their task was to decide whether the person was
concealing an object on their person. They were then asked to estimate their certainty in the
judgement using a Likert-type scale ranging from -3 to +3. Participants were also instructed to
record which nonverbal behaviors and movement patterns they observed and used to inform their
decision for each trial. Participants were not told what the object was, nor were they given any
information about the number of persons who were concealing.
Sweet et al. (2017, Experiment 1): Detection of a Concealed Weapon
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
14
Participants. Participants were 51 law enforcement officers (6 women) and 56 students
(37 women) from a university in the midwestern U.S. Officers ranged in age from 27 to 64 years
(Mofficer-age = 40.53; SDofficer-age = 8.83) and had a mean of 14.53 years of experience (SDofficer-
experience = 9.40). Officers were recruited via snowball sampling and compensated $50 per hour for
their time. Student participants ranged in age from 18 – 24 years (Mstudent-age = 19.12; SDstudent-age
= 1.35). Students were recruited through the psychology department’s participant pool and
compensated with credits toward their research requirement. This study and all studies discussed
were approved by the Institutional Review Board.
Study Design. A repeated measures within-subjects experimental design was used in
which participants viewed eight videos that showed a man walking into a secure facility, namely
a local courthouse. Each man concealed and did not conceal an object in an equal number of
trials, and therein participants viewed four concealment and four no concealment trials.
Stimulus Materials. Two different men (of the same age, race, and height) were used in
the creation of stimulus materials, though each video presents only one individual. Videos
captured each man exiting a building, crossing the street, and walking into the courthouse
(traversing approximately 100 yards). Each man performed multiple trials in which they were
randomly assigned to conceal (or not conceal) a Glock Model 19, 9-mm handgun in the
waistband of their jeans. Videos averaged 47.6 seconds in duration.
A “high-stakes paradigm” was used (Frank & Ekman, 1997) wherein, for each trial,
individuals were incentivized with a monetary reward for which they could earn a portion of the
reward, earn nothing, or lose a portion of an earned reward. Such paradigms are commonly used
when researchers attempt to mimic real-world scenarios where there are rewards and penalties.
To increase cognitive load and to support the high-stakes paradigm, individuals were also
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
15
informed that uniformed sheriff’s deputies were monitoring visitors as they entered the building
and would make judgments regarding whether they believed the target was concealing the
firearm. The men were instructed to walk normally and try not to raise suspicion. They were told
they would be rewarded if they were able to deceive the deputies. After all the trials were
concluded, each man received the maximum reward of $125.
Procedure. Study participants were told that they would be viewing several videos of a
man exiting a building and walking into a courthouse, and that on each trial the person may or
may not be concealing a weapon. Participants were not told how many videos they would see.
Participants were instructed to judge whether the man was concealing a weapon (Yes / No) and
to indicate how certain they were in their decision using a scale labeled in 10% increments from
50% to 100%. Given that chance rate of accuracy for a task like this would be 50%, on average,
the use of this “half scale” better conceptual sense. Furthermore, confidence-accuracy calibration
is well modeled by half-range (i.e., 50-100%) confidence scales (Weber & Brewer, 2003).
Participants were also instructed to indicate which nonverbal behaviors and movement patterns
they observed and then used to inform their decision.
Sweet et al. (2017, Experiment 3): Detection of an Unstable Device
Participants. Study participants were 55 law enforcement officers (21 women) and 49
students (27 women) from a university in the Midwestern U.S. Officers ranged in age from 25 –
64 years (Mage = 40.00; SD = 9.16) and had a mean of 14.42 years of experience (SD = 8.31).
Officers were recruited via snowball sampling and compensated $50/hr for their time. Student
participants ranged in age from 18 – 26 years (Mage = 19.59; SD = 1.57). Students were recruited
through the psychology department’s participant pool and compensated with credits toward their
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
16
research requirement. This study and all studies discussed were approved by the Institutional
Review Board.
Study Design. A repeated measures experimental design was used in which participants
viewed nine videos that showed two men carrying backpacks and navigating their way through a
crowded sidewalk. Participants were tasked with identifying which of the two men was
concealing an unstable device in their backpacks (a two-alternative forced choice task). One
person from each pair was randomly assigned to conceal the device.
Stimulus Videos. In each stimulus video, participants viewed two individuals with
backpacks traversing a crowded sidewalk (with an average play time of 82.2 sec). Each of the
two-man pairs was viewed only once. One member of each team was randomly assigned to
conceal a 2-quart bucket of water in his backpack. The two-man teams were instructed to walk
normally, without raising suspicion. The 2-quart bucket of water simulated an unstable device
(e.g., an improvised explosive device). A high-stakes paradigm (Frank & Ekman, 1997) was
used again to incentivize performance, and following the final trial everyone was provided the
full reward.
Procedure. Study participants were told that they would watch several videos of three-
man teams walking through a crowd, and that one member of the team may be concealing an
“unstable device” in his backpack. Participants were instructed that they must decide which
subject was concealing the device and therein designate the appropriate person (Target A or B)
as concealing. They were then asked to estimate their certainty in the judgement using a Likert-
type scale ranging from -3 to +3. Participants were also instructed to record which nonverbal
behaviors and movement patterns they observed and used to inform their decision for each trial.
Measures of Detection Performance
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
17
In Experiment 1, the performance of participants in detecting concealment vs. no
concealment were evaluated using signal detection estimates of discrimination accuracy (d’) and
response criterion (c; see Stanislaw & Todorov, 1999). Discrimination accuracy (d’) is a
composite measure used to compare standardized proportions of correct conceal judgments
(called hits) and incorrect conceal judgments (called false alarms). Positive d’ values occur when
a participant’s hit rate is higher than their false alarm rate, while negative values indicate more
false alarms than hits. Therefore, higher d’ values were expected when participants were better at
distinguishing between individuals who were concealing and those who were not. Response bias
(c) was used to assess a participant’s general willingness to make a “concealing” (rather than a
“not concealing”) judgment. Positive values indicate that the participant was more conservative
(i.e., less likely to respond “conceal”), while negative values indicate more liberal responding
(i.e, more likely to respond “conceal”). Measures of detection performance from Sweet et al.’s
(2017) Experiments 1 and 3 are not analyzed here; instead, aggregate effects from this study are
briefly discussed for comparison purposes.
Behavioral Coding Across Experiments
Consistent with behavioral research practices (Bakeman & Quera, 2011), a coding
scheme was developed. For each video in each experiment, research assistants entered all the
qualitive responses (e.g., officer and layperson responses to “what informed your decision?”)
into an Excel sheet. Once all the responses were entered, two research assistants worked
independently to group similar responses. For example, responses that referenced head
movements were grouped together; responses that mentioned arm movements were grouped
together and so on. Once all responses were grouped, responses within each group were further
classified (e.g., head turn, head tilt, arm restricted). This served as the basis for an objective
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
18
coding system. A preliminary analysis was performed, and it was determined that head, arm, and
gait were the most often reported behaviors observed.
The lead author and a research assistant, who were blind to condition, objectively coded
the head, arm, and gait behaviors for each video used in this set of experiments. Objective coding
involved recording frequencies for each behavior from the coding scheme developed from the
participants’ responses. For example, the lead author and research assistant recorded the number
of head turns, tilts, restricted, freely moving arm swings, movements, arm movement, and
unnatural gait and typical gait. The interrater reliabilities associated with this set of coding can be
found in Table S1 and the interrater reliabilities for participants’ self-report can be found in
Table S2 (in the Supplementary Materials). This coding procedure was followed for all three
experiments reported here.
Conceptual Framework
Brunswik (1952, 1956) proposed a conceptual framework of human perception situated
in probabilities functionalism. The framework proposes that uncertainty in the environment leads
people to render probabilistic judgments regarding what they perceive based upon available cues.
Referred to as the Lens Model, Brunswik’s framework considers both the diagnostic value of
cues in the environment for predicting a criterion state (referred to as validity coefficients; here,
articulable behaviors) and the extent to which judges may rely upon such cues to render
judgments about a given state (referred to as utilization coefficients; here, whether the individual
is concealing). The Lens Model has been used to capture judgments related to clinical inference
(Hammond et al., 1964), interpersonal perception (Hammond et al., 1966), personality
attributions (DeGroot & Gooty, 2009), and deception detection (Hartwig & Bond, 2011).
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
19
Thus, the Lens Model proposed here (see Figure 1) examines the extent to which
articulable behaviors offered by observers were useful predictors of both ground truth (actual
concealment vs. no concealment) and the subjective judgments of concealment offered by
observers. In addition, we assessed whether objectively coded behaviors were predictive of both
ground truth and observers’ subjective judgments. Separate models were estimated for each cue
set (articulable behaviors and objectively coded behaviors) and outcome variable (judgments of
concealment and ground truth about concealment) using the mega-analysis approach in which
data across the three experiments were aggregated to estimate effects (described below).
[Figure 1 here]
Analytic Approach
The current study makes use of a statistical technique referred to as a mega-analysis
(Curran & Hussong, 2009; Eisenhauer, 2020). The approach is conceptually similar to a meta-
analysis, but rather than analyzing the results of similar studies to estimate an overall effect, a
mega-analysis pools the raw data from exact or conceptual replications of studies to arrive at an
aggregate estimate of effects (Cooper & Patall, 2009). The approach uses multi-level regression
analyses (a nested model) to combine data from multiple experiments and obtain estimates the
statistical association of interest, while controlling for any differences between studies (e.g.,
sample size, stimuli, etc.; cf. Mantua et al., 2021).
Mega-analyses are appropriate when the variables were collected and measured in the
same way and the participant population is expected to be relatively homogeneous between
studies. Here, we have three conceptual replications of the same paradigm, with participants
drawn from similar law enforcement and student samples. Hence, our data is ideal for a fixed
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
20
effects mega-analysis approach (in that we are not expecting the independently collected samples
to significantly differ).
We conducted a mega-analysis for each of the four “paths” within the proposed Lens
Model (Figure 1) to assess the overall predictive value of each independent variable
(“objectively coded” versus “articulable” behaviors) for both ground truth and subjective
judgments of concealment. The mega-analysis was run using a multilevel logistic regression
analysis (given our binary dependent variables) which controlled for the variation in responses
due to experiment (“experiment” Level 2 variable) and participant (“participant” Level 2
variable). Whether the participant was an officer, or a layperson was also included (“group
Level 1 variable), and interactions between this variable and the articulable behaviors were run
for each model and compared to the main effects model.
For each behavior type and Lens Model path, we ran a null model (no predictors), a
control model (group only), a main effects model (group and all articulable behaviors), and a full
model (all predictors and all interactions between Group and each behavior score). Model fit was
compared by running Goodness-of-Fit Chi-Square analyses. If the interaction model was a better
fit than the main effects model, this was the model reported. If not, the main effects model was
reported (and this model was always a better fit than the null or control model). Occasionally, an
interaction model will be a better fit according to the Chi-Square test, but the interaction was
non-significant. Because we are still interested in the main effect of Group (officers versus
laypersons) and articulated or objectively coded behaviors on conceal judgments and ground
truth, in these cases the model was reduced further by removing the non-significant interaction.
In cases like this, the main effects model would then be reported and interpreted. Refer to the R
code posted on OSF (https://osf.io/jptkb/) for further details.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
21
Results
How Accurate were Participants’ Conceal Judgments in Experiment 1?
While the focus of the current paper centers on a Lens Model analysis of behavioral cues
that might predict ground truth and subjective concealment judgments, we first discuss the
measures of detection performance for participants in Experiment 1 (innocuous object study).
Average discrimination accuracy (d’) and response criterion (c) values for each participant group
can be found in Table 1, along with inferential statistics for the between-group comparisons.
Performance data were analyzed using the “lm” function (in the “stats” package in R; R Core
Team, 2013). We calculated Bayes Factors (using the “lmBF” function, “BayesFactor” package;
Morey et al., 2018) and Cohen’s d effect sizes to assess differences between officers and
laypersons. Participants in Experiment 1 demonstrated little ability to distinguish between
individuals concealing a handkerchief and those who were not (d′). Officers were nominally less
accurate than laypersons (M = 0.19 and 0.51, SD = 0.94 and 1.05, respectively); however, this
difference was not statistically significant (d = -0.32, 95% CI [-0.68, 0.05], BF01 = 0.72). The
observed effect size in this experiment was somewhat larger than the average weighted effect
size reported by Sweet et al. (2017; d = -0.08, 95% CI [-0.30, 0.14]), though both studies suggest
that officers perform slightly worse than lay observers.
With respect to response criterion (c), officers were significantly more liberal in
responding than laypersons (M = -0.04 and 0.47, SD = 0.64 and 0.74, respectively), suggesting
that officers were more willing to render conceal judgments (d = -0.73, 95% CI [-1.11, -0.36],
BF01 = 177.05). This finding is quite consistent with investigative biases in which officers
demonstrate a tendency to perceive deception even when it is not present (see Meissner &
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
22
Kassin, 2002), and appears somewhat larger than the average weighted effect size previously
observed by Sweet et al. (2017; d = -0.20, 95% CI [-0.47, 0.07]).
Is there Correspondence between Objectively Coded and Articulable Behaviors?
To what extent did participants notice and report the same behaviors that were reported
by our objective coders? Table 1 provides ICC values for each experiment, group, and articulable
behavior type. Overall, ICCs indicated low correspondence between coded and articulable
behaviors. Further, no systematic differences were observed in the correspondence between
coded and articulable behaviors reported by officers versus laypersons.
ICC values for head behaviors indicated poor correspondence between the coded head
behaviors and the head movements reported by participants (ICChead = .236, 95% CI [.089,
.812]). When assessed separately, no differences were observed in corresponded between
perceived and objectively coded behaviors for officers (ICCofficer = .252, 95% CI [.082, .930]) vs.
laypersons (ICCcontrol = .223, 95% CI [.082, .801]). ICC values for arm behaviors indicated
similar patterns – low correspondence between the arm behaviors reported by participants and
the behaviors observed by the objective coder (ICCarm = 0.236, 95% CI [0.089, 0.812]). Officers
demonstrated slightly lower correspondence between the objective coder’s observations and their
perceptions of head behaviors (ICCpolice = .204, 95% CI [.088, .610]) than did laypersons
(ICCcontrol = .260, 95% CI [.110, .746]), though not statistically different. Finally, ICC values for
gait behaviors suggests slightly higher, but still low, correspondence than was found for head and
arm behaviors. That is, participants and objective coders reported similar levels of gait behaviors
in individuals a small portion of the time (ICCgait = .336, 95% CI [.163, .753]). Offiers
demonstrated approximately the same level of correspondence between gait behaviors they
reported and the objective coder’s behaviors (ICCofficers = .374, 95% CI [.187, .784]) as
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
23
laypersons (ICCcontrol = .397, 95% CI [.202, .799]). So, behaviors recorded by objective coders
are occasionally reported by participants too, but the correspondence is low in general, for both
officers and laypersons.
[Table 1 here]
Do Officers and Laypersons Differ in the Articulable Behaviors They Report?
Three regressions were run on the data across all three experiments, with articulable
behaviors (a unique model for head, arm, and gait behaviors) as the outcome variable. The group
variable was included as a predictor and these analyses were nested within Participant and
Experiment. There was a significant effect of group on perceptions of head behaviors, indicating
that laypersons tended to see more abnormal gait behaviors across experiments than police (Bhead
= .15, p < .001). For articulated arm behaviors, police tended to report significantly more arm
behaviors than laypersons (Barm = -.17, p < .001). Finally, laypersons reported significantly more
gait behaviors in general than officers (Bgait = .14, p < .001). Overall, these analyses suggest that
police notice and report more atypical arm movements, while laypersons report more unusual
head and gait behaviors. Table 2 provides the average number of typical / atypical articulable
behaviors for officers and laypersons.
[Table 2 here]
Do Articulable Behaviors Predict Subjective Conceal Judgments vs. Ground Truth?
To examine the relation between participants’ perceptions of articulable head, arm, and
gait behaviors and subjective judgments about whether an individual was concealing vs. ground
truth, separate multilevel logistic regressions were run with subjective conceal judgments or
ground truth as the outcome variables (0 = “not concealing”, 1 = “concealing”). The interaction
model was a better fit than the main effects model when predicting conceal judgments (X2[3, N =
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
24
298] = 16.57, p < .001) and ground truth (X2[3, N = 298] = 22.32, p < .001). Results reported in
this section also appear in the Figure 2 Lens Model.
[Figure 2 here]
Articulable Head Behaviors. There was no significant interaction between group and
articulable head behaviors when predicting judgments about concealment (Binteraction = .13, p =
.308,
𝜃
interaction = 1.14) or ground truth (Binteraction = -.15, p = .266,
𝜃
interaction = 0.86). So, this
interaction was removed from the model to assess the main effect of articulable head behaviors.
This reduced model was not significantly different from the full interaction model for the conceal
judgements model (X2[1, N = 298] = 1.04, p = .308) or the ground truth model (X2[1, N = 298] =
4.40, p = .111). There was a significant main effect of head behaviors when predicting ground
truth (Bhead = .23, p < .001,
𝜃
head = 1.25), but not when predicting conceal judgments (Bhead = -
.08, p = .221,
𝜃
head = 0.92). Thus, more articulable, atypical head behaviors were associated with
a higher likelihood that an individual was concealing a weapon, though this was a fairly small
effect1. Conversely, articulable head behaviors were not associated with conceal judgments.
Articulable Arm Behaviors. There was a significant interaction between group and
perceptions of arm behaviors when predicting conceal judgments (Binteraction = -.20, p = .022,
𝜃
interaction = 0.82) and ground truth (Binteraction = .36, p < .001,
𝜃
interaction = 1.42). The nature of
these interactions differed, though. When analyzing conceal judgments, officers reported
atypical, articulable arm movements more frequently when made a conceal judgment compared
with judgments that an individual was not concealing. Although the pattern in these data were
similar for laypersons, the difference in the number of atypical, articulable arm movements for
concealing and not concealing judgments was smaller and not significantly different. When
examining ground truth, officers reported seeing more atypical arm behaviors for individuals
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
25
who were actually not concealing than for those that were concealing an object. In comparison,
there was no significant difference in the number of atypical, articulable behaviors for laypersons
whether the individual was actually concealing or not.
Articulable Gait Behaviors. There was no significant interaction between group and
articulable gait behaviors when predicting ground truth, but this interaction was significant for
conceal judgments (Binteraction = .24, p < .001,
𝜃
interaction = 1.27). Officers and laypersons reported
approximately the same number of atypical, articulable gait movements when they thought the
individual was not concealing, but laypersons reported more atypical, articulable gait behaviors
than officers for conceal judgments. The main effect of perceived gait behaviors on ground truth
revealed that more atypical gait behaviors reported by participants were more likely to be
associated with individuals who were actually concealing (Bgait = .29, p < .001,
𝜃
gait = 1.34),
though laypersons reported slightly more atypical gait behaviors for concealing individuals than
did officers (non-significant; Binteraction = -.12, p = .068,
𝜃
interaction = 0.89).
Do Objectively Coded Behaviors Predict Subjective Conceal Judgments vs. Ground Truth?
To examine the relation between coded head, arm, and gait behaviors and ground truth, a
logistic regression was run with ground truth as the outcome variable (0 = “not concealing”, 1 =
“concealing”). Because each individual in each video served as the individual for these analyses,
the sample size was only 35. Specifically, Experiment 1 had nine videos, Sweet et al. (2017)
Experiment 1 included eight videos, and Sweet et al. (2017) Experiment 3 featured nine videos
with two individuals per video (18 total observed individuals). As a result, these models were
underpowered, though they were not significantly different from a perfect model fit (Hosmer and
Lemeshow Goodness of Fit: X2[8, N = 35] = 10.19, p = .252). Results reported in this section
appear in the Figure 3 Lens Model.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
26
[Figure 3 here]
Objectively Coded Behaviors. Objectively coded head movements were non-significant
predictors of ground truth, and the odds ratio indicated that increases objectively coded atypical
head behaviors were not associated with an increase or decrease in the likelihood that an
individual was concealing an object (Bhead = -.01, p = .683,
𝜃
head = 0.99). Thus, the number of
coded atypical head behaviors did not predict whether an individual was actually concealing.
Analyses indicated that coded arm movements were a nonsignificant predictor of ground truth,
and the effect size indicated that objective arm movements were not associated with an increase
in the likelihood that an individual was concealing (Barm < .001, p = .915,
𝜃
arm = 1.00).
Therefore, objectively coded, atypical arm behaviors could not help sort between individuals that
were concealing and were not concealing. Objectively coded gait behaviors were not indicative
of ground truth (Bgait < .001, p = .884,
𝜃
gait = 1.00). That is, increases in atypical gait behaviors
reported by our objective coders were not associated with whether the individual was concealing.
General Discussion
Using a mega-analysis of data from three experiments, we investigated the behaviors that
officers and laypersons believed were associated with concealing three different objects, namely
an innocuous object (Experiment 1 from the current study), a gun (from Sweet et al., 2017,
Experiment 1), and an unstable device (from Sweet et al., 2017, Experiment 3). Using a Lens
Model framework (Brunswick, 1952; 1956), we observed little correspondence between
participants’ perceptions of articulable behaviors and the objective coding of such behaviors,
suggesting that participants may be seeing atypical behaviors that are not actually present.
Moreover, the behaviors articulated by officers, who have received training and who have more
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
27
experience judging whether someone is concealing than laypersons, were no more consistent
with objective coders’ observations than the reports from laypersons.
Officers often articulated more behaviors than laypersons, but not because they were able
to identify more truly diagnostic behaviors or render more accurate conceal judgments than
laypersons. Atypical head and gait behaviors perceived and reported by participants were weakly
associated with ground truth, but these effects were small and, even if participants articulated
these behaviors, they did not lead to participants making more accurate conceal judgments.
Perceptions of arm movements in this context were particularly problematic, such that while
atypical, articulable arm movements were often reported when participants justified their conceal
judgment, these movements were negatively related to whether the individual was actually
concealing. That is, observations of atypical arm movements actually predicted when individuals
were not concealing, but participants thought they were indicative of concealing and reported
using them to make their judgments.
Even though participants were able to detect and articulate specific behaviors to justify
their concealment judgments, these articulated behaviors neither align with what objective coders
reported nor do they appear to help officers or laypersons accurately discriminate concealment.
The failure to align with objective coding of the behaviors may reflect a tendency toward
confirmation bias, a phenomenon where individuals interpret ambiguous cues in a way that
supports their existing beliefs (Nickerson, 1998). Officers may be particularly susceptible to
confirmation bias as a function of the training they receive, which is based on anecdotal
evidence. The biases encouraged by this anecdotal training is further reinforced by their
experience on the job. We found that police were more likely to render conceal judgments than
laypersons, indicating a greater willingness to see and report that a person was concealing
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
28
(evidenced by a more liberal response criterion). This suggests that police are more likely to
interpret an ambiguous behavior or movement as a clue that someone is concealing. But, as
stated here, these interpretations are based upon patterns of behavior that have not been
validated. It is also possible that officers’ tendency to believe that someone is concealing could
lead them to perceive patterns of behavior that confirm their belief that someone might be
concealing. Indeed, there was evidence of this in these current data (e.g., officers tended to
articulate more behaviors).
The current findings suggest that there are very few reliable behavioral cues or movement
patterns that accurately discriminate between individuals who are concealing and those who are
not, a finding which contradicts the existing non-empirical work (Meehan & Strange, 2021;
2015). The type of object that the individual is concealing and its associated features (e.g., bulk,
weight, size) made little difference—participants struggled to identify truly diagnostic behaviors
whether the object was a firearm, an unstable device, or a handkerchief. We were able to control
for these between study differences to conclude that, regardless of the object, participants were
unable to accurately use observable behaviors to determine which individuals were concealing.
In fact, officers’ performance was slightly worse than that of laypersons. Thus, their additional
training and on-the-job experience did not help officers identify truly diagnostic behaviors or
accurately determine who was concealing.
The Lens Model also permitted an examination of the differences in beliefs about which
movements are diagnostic of concealing among officers and naïve participants, while controlling
for the differences between the detection contexts and objects using the mega-analysis. It appears
that officers and laypersons have formed beliefs about the kinds of behaviors that someone will
exhibit if they are concealing an object; however, there were no consistent differences in patterns
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
29
between participant groups, across experiments, or across behavior types. Thus, not only do
these beliefs appear inconsistent and not well-defined within these groups, but they also failed to
prove diagnostic of whether an individual was actually concealing.
Taken together, these studies demonstrate the inherent difficulty associated with
discriminating movement patterns of those who are concealing a weapon, an unstable device, or
an innocuous object, and those who are concealing nothing. Thus, previous research suggesting
that gait is a useful cue for discriminating personal characteristics such as sex, age, and
vulnerability (Barclay et al., 1978; Gunns et al., 2002; Montepare & Zebrowitz-McArthur, 1988)
cannot be extended to gait-related movements of someone who is attempting to conceal
something (e.g., Runeson & Frykholm, 1983). Across all three experiments, gait patterns
observed by objective coders could not discriminate between a concealing individual and non-
concealing individual and, although there was some evidence that articulable head and arm
behaviors were more diagnostic, none of these patterns were reliable across studies.
Implications for Policy, Practice, and Training
With respect to the original, applied purpose of these experiments, it is critical that any
articulable behavioral cues that officers are taught to use in a field scenario are consistently
observable across individuals, contexts, and objects, and that any differences in articulable
behaviors across such conditions are, at a minimum, clearly defined. Behavioral indicators of
concealing are used in the context of invoking “reasonable suspicion” to justify an officer
infringing on the right of citizens to be free from “unreasonable search and seizure.” Suspicion
triggered by articulable behaviors should only be considered “reasonable” if there are consistent
patterns of behaviors linked with the concealment of a dangerous object. But our studies suggest
that officers cannot tell when someone is (or is not) concealing. Police training, policy, and
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
30
practice should be grounded in empirical science rather than anecdotal evidence (e.g.,
Greenbaum, 2014).
We believe there are two key implications of this research. First, the studies presented
here demonstrate the difficulty of a task that is frequently required of officers, namely to
accurately discriminate and articulate behaviors believed to be associated with concealment. In
the current studies, there were no head, arm, and gait behavior articulated by participants or
objective coders that were reliably associated with concealment. Nevertheless, we were able to
develop a systematic understanding of what officers are attending to when they render
concealment decisions and what behaviors they believe are linked with concealing objects.
Specifically, officers tended to focus on arm movements more than laypersons, which may be a
result of their training.
Second, law enforcement training is typically grounded in anecdotal evidence, but the
current findings suggest that this evidence does not support accurate judgments about whether
someone is concealing. Future research is needed to define articulable behaviors or observation
strategies that might be truly helpful for determining if a person is concealing. If there are any
such reliable cues, this could form the basis for research-based training approaches. If no
empirical evidence of observable cues can be found, then the field must abandon this ineffective
strategy and search for novel approaches that can assist with this task. In sum, these findings
challenge the common belief that there are certain movements associated with attempting to
conceal an object and that officers can acquire the ability to detect these behaviors.
Limitations and Future Directions
As mentioned in the Results, our analysis of objectively coded behaviors had a relatively
small sample size. Specifically, these analyses were drawn from the experimental stimuli
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
31
resulting in a “sample” of only 35 individuals (across stimulus videos). The mega-analysis
approach was, in part, used to circumvent this issue by pooling these data across experiments
while controlling for experimental context and individual-level variance. Nevertheless, this
portion of the Lens Model framework remains underpowered and is deserving of further study.
Such power issues represent a challenge whenever behavioral characteristics are coded across
stimuli—a resource-intensive process. Given the novelty of our analysis, we encourage future
research with more robust stimulus samples to replicate and extend our understanding of the
predictive utility of objective behavioral cues to concealment.
Another limitation of the current work is that it is difficult to mimic the conditions
present in a real-world concealment situation. In fact, creating an experiment that activates a
limbic system response is most challenging and could violate many ethical guidelines that
researchers must follow. Although we incentivized our individuals in appropriate ways,
generating realistic emotional and motivational responses with serious consequences (e.g., the
possibility of being arrested or put in jail) was not possible. Although we believe our incentive
paradigm was effective in facilitating the motivation for individuals to perform their task well,
including the potential for performance anxiety, it was unlikely to produce any real fear or
psychological stress that would be associated with concealing an object in the presence of an
officer.
Similar concerns arise regarding our officer participant group. While their task was clear
and the materials were similar in important ways to a relevant and realistic situation, our
experiments were not be able to elicit the same motivational factors that might be experienced in
a real-time judgment situation. The extent to which these aspects of an observers’ experiences
are important for adequately detecting the concealment an object is a potential future direction.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
32
For instance, a more immersive scenario involving real crime footage of people concealing or
not (e.g., security camera footage) may provide an enhanced context for such judgments.
However, such materials can be difficult to source and there are likely to be issues with the
quality of footage. Inducing cognitive load by having officers perform the experimental tasks
under time constraints or while dividing their attention may better mirror real-world parameters.
Virtual reality-based experiments that offer a more immersive experience may also produce the
type of emotional and motivational responses found in the real world, but also permit researchers
to experimentally control the environment and to manipulate ground truth (Riva et al., 2007).
Another consideration for future work concerns the absence of variation in
environmental, contextual, and personal factors in the current studies. Our target individuals
were not representative of the broader U.S. population in terms of race, ethnicity, or gender, as
they were primarily White men from the Midwestern U.S. Furthermore, the stimulus materials
were filmed restricted environmental settings that were conducive to video recording. Hence, the
variation experienced by officers in their day-to-day activities was not fully captured, though the
officers who participated in this study were from the Midwest, where the population is primarily
Caucasian (91.8% according to recent census data) and the crime rate is low. Future studies
should assess officers who work in more diverse areas and should vary the features of the
individuals and contexts associated with the materials accordingly.
Conclusion
The current study provides an important set of empirical findings that seek to understand
officers’ perceptions of what behaviors and movement patterns are associated with concealment.
Across three studies, we examined officers’ perceptions of behaviors believed to be associated
with a concealed object, and how these perceptions compare to the perceptions of laypersons or
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
33
an objective coder. In general, officers were somewhat less accurate when judging whether a
person was concealing an object, but also more willing to render a concealment judgment than
laypersons. Further, the articulated behaviors cited by our participants were not reliable
indicators of whether someone was actually concealing an object. Finally, the behaviors coded
by objective observers was not associated with either the behaviors articulated by officers and
laypersons or with ground truth.
Officers are routinely called upon to assess the behavior of others and make decisions
about whether a person is concealing a weapon or dangerous object. Their ability to accurately
perform this task and justify it appropriately with articulable behaviors can have serious
implications under the Fourth Amendment of the U.S. Constitution. Our findings question the
ability of officers to render such judgments and suggest a need for additional research to examine
whether there are any objective behaviors that can predict whether someone is concealing an
object. The current research also calls for the development of empirically supported training that
arms police officers with more diagnostic approaches to detecting the concealment of a weapon,
device, or object.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
34
References
Bakeman, R., & Quera, V. (2011). Sequential analysis and observational methods for the
behavioral sciences. Cambridge University Press.
Barclay, C. D., Cutting, J.E., & Kozlowski, L. T. (1978). Temporal and spatial factors in gait
perception that influence gender recognition. Perception & Psychophysics, 23(2), 145-
152. https://doi.org/10.3758/BF03208295
Blechko, A., Darker, I., & Gale, A. (2008, October). Skills in detecting gun carrying from
CCTV. In 2008 42nd Annual IEEE Intl Carnahan Conference on Security Tech (pp. 265-
271). IEEE.
Bond Jr, C. F., & DePaulo, B. M. (2006). Accuracy of deception judgments. Personality &
Social Psychology Review, 10(3), 214-234.
https://doi.org/10.1207/s15327957pspr1003_2
Bureau of Alcohol, Tobacco, Firearms and Explosives (2019, February 28). Project Safe
Neighborhoods Enforcement Training. https://www.atf.gov/firearms/project-safe-
neighborhoods-enforcement-training-psn-3-day-program
Burgoon, J. K., Twitchell, D. P., Jensen, M. L., Meservy, T. O., Adkins, M., Kruse, J…. &
Nunamaker, J. F. (2009). Detecting concealment of intent in transportation screening: A
proof of concept. IEEE Transactions on Intelligent Transportation Systems, 10(1), 103-
112. https://doi.org/10.1109/TITS.2008.2011700
Brunswik, E. (1956). Perception and the representative design of psychological experiments.
University of California Press.
Brunswik, E. (1952). The conceptual framework of psychology. Chicago, IL: UC Press.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
35
Chen, H. M., Lee, S., Rao, R. M., Slamani, M. A., & Varshney, P. K. (2005). Imaging for
concealed weapon detection: A tutorial overview of development in imaging sensors and
processing. IEEE Signal Processing Magazine, 22(2), 52-61.
https://doi.org/10.1109/MSP.2005.1406480
Commonwealth v. Evelyn, 485 Mass. 691 (2020).
Cooper, H., & Patall, E. A. (2009). The relative benefits of meta-analysis conducted with
individual participant data versus aggregated data. Psychological Methods, 14(2), 165–
176. https://doi.org/10.1037/a0015565
Curran, P. J., & Hussong, A. M. (2009). Integrative data analysis: the simultaneous analysis of
multiple data sets. Psychological methods, 14(2), 81-100.
https://doi.org/10.1037/a0015914
DeGroot, T., & Gooty, J. (2009). Can nonverbal cues be used to make meaningful personality
attributions in employment interviews? Journal of Business and Psychology, 24, 179-
192. https://doi.org/10.1007/s10869-009-9098-0
DePaulo, B. M., Lindsay, J. J., Malone, B. E., Muhlenbruck, L., Charlton, K., & Cooper, H.
(2003). Cues to deception. Psychological Bulletin, 129(1), 74–
118. https://doi.org/10.1037/0033-2909.129.1.74
Eisenhauer, J. G. (2021). Meta‐analysis and mega‐analysis: A simple introduction. Teaching
Statistics, 43(1), 21-27. https://onlinelibrary.wiley.com/doi/epdf/10.1111/test.12242
Frank, M. G., & Ekman, P. (1997). The ability to detect deceit generalizes across different types
of high-stake lies. Journal of Personality and Social Psychology, 72(6), 1429–1439.
https://doi.org/10.1037/0022-3514.72.6.1429
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
36
Greenbaum, D. (2014). Know how to spot the signs of a hidden handgun. Retrieved from
http://lifehacker.com/know-how-to-spot-the-signs-of-a-hidden-handgun-1650351312
Gunns, R. E., Johnston, L., & Hudson, S. M. (2002). Victim selection and kinematics: A point-
light investigation of vulnerability to attack. Journal of Nonverbal Behavior, 26(3), 129-
158. https://doi.org/10.1023/A:1020744915533
Hammond, K. R., Hursch, C. J., & Todd, F. J. (1964). Analyzing the components of clinical
inference. Psychological Review, 71, 438-456. https://doi.org/10.1037/h0040736
Hammond, K. R., Wilkins, M. M., & Todd, F. J. (1966). A research paradigm for the study of
interpersonal learning. Psychological Bulletin, 65, 221-232.
https://doi.org/10.1037/h0023103
Hartwig, M., & Bond, C. F. (2011). Why do lie-catchers fail? A Lens Model meta-analysis of
human lie judgments. Psychological Bulletin, 137, 643-659.
https://doi.org/10.1037/a0023589
Illinois v. Wardlow, 528 U.S. 119, 120 S. Ct. 673, 145 L. Ed. 2d 570 (2000).
Mantua, J., Bessey, A. F., Mickelson, C. A., Choynowski, J. J., Noble, J. J., Burke, T. M.,
McKeon, A. B., & Sowden, W. J. (2021). Sleep and high-risk behavior in military service
members: A mega-analysis of four diverse U.S. Army units. Sleep, 44(4), zsaa221.
https://doi.org/10.1093/sleep/zsaa221
Meehan, N. C., Strange, C., & Garinther, A. (2021). It’s the walk, not the talk: Behavioral
indicators of concealed and unholstered firearms carrying. The Police Journal: Theory,
Practice and Principles, 94(4), 462-480. https://doi.org/10.1177/0032258X20960777
Meehan, N. C., & Strange, C. (2015). Behavioral indicators of legal and illegal gun
carrying (No. NRL/MR/5508--15-9597). Naval Research Lab: Washington, DC.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
37
Meehan, N. C., Strange, C., & McClary, M. (2015). Behavioral Indicators during a Police
Interdiction. Naval Research Lab, Washington DC, Information Technology Division
Montepare, J. M., & Zebrowitz-McArthur, L. (1988). Impressions of people created by age-
related qualities of their gaits. Journal of Personality and Social Psychology, 55(4), 547.
https://doi.org/10.1037/0022-3514.55.4.547
Morey, R. D., Rouder, J.N., Jamil, T., Urbanek, S., Forner, K., & Ly, A. (2018). Package ‘Bayes
Factor’: Computation of Bayes Factors for Common Designs. Retrieved from
https://cran.r-project.org/web/packages/BayesFactor/BayesFactor.pdf
Nickerson, R.S., (1998). Confirmation bias. A ubiquitous phenomenon in many guises. Review
of General Psychology, 2, 175-220. https://doi.org/10.1037/1089-2680.2.2.175
Pollick, F. E., Kay, J. W., Heim, K., & Stringer, R. (2005). Gender recognition from point-light
walkers. Journal of Experimental Psychology: Human Perception and Performance,
31(6), 1247–1265. https://doi.org/10.1037/0096-1523.31.6.1247
Richardson, M. J., & Johnston, L. (2005). Person recognition from dynamic events: The
kinematic specification of individual identity in walking style. Journal of Nonverbal
Behavior, 29(1), 25-44. https://doi.org/10.1007/s10919-004-0888-9
Riva, G., Mantovani, F., Capideville, C. S., Preziosa, A., Morganti, F., Villani, D., Gaggioli, A.,
Botella, C., and Alcañiz, M. (2007). Affective interactions using virtual reality: The link
between presence and emotions. CyberPsychology & Behavior, 10(1), 45-56.
http://doi.org/10.1089/cpb.2006.9993
Runeson, S., & Frykholm, G. (1983). Kinematic specification of dynamics as an informational
basis for person-and-action perception: expectation, gender recognition, and deceptive
intention. Journal of Experimental Psychology: General, 112(4), 585-615.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
38
https://psycnet.apa.org/doi/10.1037/0096-3445.112.4.585
Sheen, D. M., McMakin, M. L., & Hall, T. E. (2001). Three-dimensional millimeter-wave
imaging for concealed weapon detection. IEEE Transactions on Microwave Theory and
Techniques, 49(9), 1581-1592. https://doi.org/10.1109/22.942570
Stanislaw, H., & Todorov, N. (1999). Calculation of signal detection theory measures. Behavior
Research Methods, Instruments, & Computers, 31(1), 137-149.
https://doi.org/10.3758/BF03207704
Sweet, D. M., Meissner, C. A., & Atkinson, D. J. (2017). Assessing law enforcement
performance in behavior-based threat detection tasks involving a concealed weapon or
device. Law and Human Behavior, 41(5), 411 – 421. https://doi.org/10.1037/lhb0000243
Team, R. C. (2013). R: A language and environment for statistical computing. http://www.R-
project.org/
Terry v. Ohio 329 U.S. 1 (1968)
United States v. Briggs, 720 F.3d 1281 (10th Cir. 2013)
Vrij, A. (2008). Detecting lies and deceit: Pitfalls and opportunities. John Wiley & Sons.
Weber, N., & Brewer, N. (2003). The effect of judgment type and confidence scale on
confidence-accuracy calibration in face recognition. Journal of Applied Psychology, 88(3),
490-499. https://doi.org/10.1037/0021-9010.88.3.490
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
39
Endnotes
1
The magnitude of an odds ratio effect is assessed by looking at how close the value is to
“1”, and the direction is indicated by whether the value is below “1” or above “1”. These odds
ratios (1.25 and 0.92) are both close to 1, indicating smaller effects, in opposite directions. An
odds ratio of 1.25 is further away from 1 than 0.92 and indicates that atypical head behaviors
were associated with individuals who were actually concealing. On the other hand, an odds ratio
of 0.92 suggests that head behaviors have a very small association with conceal judgments, and
atypical head behaviors tend to be reported alongside judgments that the individual was not
concealing an object.
Table 1
ICC Values Indicating the Amount of Correspondence Between Variables in the Lens Models.
Sample
Conceal Judgments and
Ground Truth
Objectively Coded and
Articulable Head Behaviors
Objectively Coded and
Articulable Gait Behaviors
All Data
0.017,
[0.003, 0.948]
0.236,
[0.089, 0.812]
0.336,
[0.163, 0.753]
Experiment 1
0.017,
[0.002, 0.951]
0.107,
[0.029, 0.829]
0.579,
[0.328, 0.920]
Sweet et al. (2017,
Experiment 1)
0.030,
[0.004, 0.972]
0.037,
[-0.030, 0.571]
-0.003,
[-0.008, 0.079]
Sweet et al. (2017,
Experiment 3)
0.010,
[0.001, 0.921]
0.006,
[-0.001, 0.265]
0.116,
[0.042, 0.537]
Officers
0.006,
[0.000, 0.884]
0.252,
[0.082, 0.93]
0.374,
[0.187, 0.784]
Laypersons
0.033,
[0.006, 0.973]
0.159,
[0.045, 0.885]
0.891,
[0.722, 0.991]
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Notes. ICCs are reported with 95% CI in square brackets.
Table 2
Descriptive Statistics for Articulable Behaviors as a Function of Group and Experiment.
Sample
Articulable
Behavior
Participant Group
Officers
Laypersons
Experiment 1
Head
0.002 (0.817)
0.444 (0.685)
Arm
1.225 (0.916)
0.778 (0.786)
Gait
1.002 (1.249)
1.778 (0.786)
Sweet et al. (2017,
Experiment 1)
Head
< 0.001 (0.070)
0.040 (0.238)
Arm
0.120 (0.347)
0.038 (0.191)
Gait
0.029 (0.395)
0.094 (0.518)
Sweet et al. (2017,
Experiment 3)
Head
-0.057 (0.622)
< 0.001 (0.472)
Arm
1.058 (1.355)
0.612 (0.892)
Gait
2.280 (0.804)
2.445 (1.014)
Notes. Values represent the mean number of articulable behaviors for each sample and group.
Positive values indicate greater reports of atypical behaviors, while negative values indicate
greater reports of typical movements. Standard deviations are provided in parentheses.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Table S1
Interrater Reliability for Objective Observations of Subjects’ Head, Arm, and Gait Behaviors.
Table S2
Interrater Reliability for Coding of Participant’s Reports of Head, Arm, and Gait Behaviors.
Experiment
Behavior
Kappa
Cronbach’s Alpha
ICC
Experiment 1
Head
0.593
0.996
0.989
Arms
0.724
0.983
0.970
Gait
0.857
0.999
0.998
Experiment 2
Head
0.940
1.00
0.999
Arms
0.580
1.00
0.999
Gait
1.00
1.00
1.00
Experiment 3
Head
0.614
0.991
0.984
Arms
0.723
0.991
0.980
Gait
1.00
1.00
1.00
Experiment
Behavior
Kappa
Cronbach’s Alpha
ICC
Experiment 1
Head
0.724
0.998
0.996
Arms
1.000
1.00
1.00
Gait
0.857
0.998
0.996
Experiment 2
Head
0.940
1.000
1.000
Arms
1.000
1.000
1.000
Gait
1.000
1.000
1.000
Experiment 3
Head
1.000
1.000
1.000
Arms
0.859
0.995
0.990
Gait
1.000
1.000
1.000
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Figure 1
Lens Model Framework Assessing the Extent to which Objectively Coded and Articulable
Behaviors can Predict Ground Truth and Conceal Judgments
Ground Truth: Individual
is actually concealing an
object or not
Conceal Judgment:
Individual is judged to be
concealing or not
Objectively coded
behaviors the individual’s
performed in the videos
Articulable behaviors the
participants reported observing
in the videos of individuals
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Figure 2
Lens Model of Articulable Head, Arm, and Gait Behaviors Predicting Ground Truth and
Subjective Conceal Judgments
Note. Inferential statistics and intraclass correlations are reported. Dashed lines indicate non-
significant associations, while solid lines indicate relations that are significant (p < .05).
𝜃=#0.93
𝜃= 1.25
𝜃= 1.55
𝜃= 1.29
𝜃= 0.63
𝜃
=
1.36
Ground Truth
0 = not concealing
1 = concealing
Conceal
Judgments
0 = not concealing
1 = concealing
Articulable Head
Behaviors
Articulable Arm
Behaviors
Articulable Gait
Behaviors
With interaction with Group: 𝜃 = 1.34
Without interaction with Group: 𝜃 = 1.47
R2 = 0.08
R2 = 0.13
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Figure 3
Lens Model of Objectively Coded Head, Arm, and Gait Behaviors Predicting Ground Truth and
Subjective Conceal Judgments
Note. Inferential statistics and intraclass correlations are reported. Dashed lines indicate non-
significant associations, while solid lines indicate relations that are significant (p < .05). *While
this result is significant (p = 0.042), this is essentially a null effect.
𝜃
=
1.05
𝜃= 1.00
𝜃= 1.00*
𝜃= 0.99
𝜃
= 1.00
𝜃= 1.00
Ground Truth
0 = not concealing
1 = concealing
Conceal
Judgments
0 = not concealing
1 = concealing
Objectively Coded Head
Behaviors
Objectively Coded Arm
Behaviors
Objectively Coded Gait
Behaviors
With interaction with Group: 𝜃 = 1.34
Without interaction with Group: 𝜃 = 1.47
R2 = 0.02
R2 = 0.01
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Supplementary Material
Coding procedures used for each of the three experiments are described below.
Experiment 1: Coding Procedures. There were two different approaches to coding
subject’s gait patterns across the eight videos. First, for each video, research assistants entered
participants’ qualitative responses into an Excel workbook. Once all the responses were entered,
two research assistants worked independently to group similar responses. For example, all
responses that mentioned the head were grouped together; all responses that mentioned the arms
were grouped together, and so on.
Second, similar responses were grouped, within category (e.g., head, arm, torso, etc.) the
responses were further classified. For example, the ‘head’ category was refined to include head
neutral, turned, down, and bobbing. Refer to OSF for the coding sheet and full list of categories.
Third, the lead author and a research assistant independently coded subjects’ behavior
across each of the eight videos. In the section that follows, each approach to coding is described.
Coding participant responses. To facilitate analysis, research assistants entered
participants’ qualitative responses for each of the eight videos into an Excel workbook;
responses for each video were organized by Participant ID number and a worksheet was created
for each of the eight videos. The research assistants were blind to which videos were in the
concealing and non-concealing condition during the coding process. After all of the participants’
responses were entered into Excel, the next step was to organize similar responses within each
video.
There were 429 responses grouped into eight broad categories: body part, gaze, gait,
demeanor, overt signs, body touch, no signs, and unsure. Analyses revealed head, arm, and gait
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
movements were the most useful behaviors. Within each of these three categories, responses
were then further refined and categorized creating multiple levels of codes within each broader
category (refer to OSF). For example, participants noted the specific manner in which a
particular body part was moving such as a head turn, asymmetrical arm swing, or stiff gait.
Head. Head movements were considered any movements of the head away from the
neutral position. Neutral position was defined as the Target’s chin being parallel with the ground
and no noticeable deviations along the x, y, or z axes. The number of head movements for each
Target was recorded. Informed by the subjective responses, head movements were categorized as
head neutral, head, turn, dead down, or head bobbing. A head turn was recorded when the
Target’s chin broke the sagittal plane and the tip of the chin moved to the left or right of the
suprasternal notch. A head tilt down was recorded when the Target’s chin dipped below the X
axis toward the Target’s chest. Head bobbing was recorded when the Target’s head moved in a
continuous rhythmic pattern.
Arms. An arm swing began when the arm moved forward with the movement of the
opposing leg. The number of arm swings for each Target was recorded. Informed by the
subjective responses, arm movements were categorized as arm expected, arm disrupted, arm
restricted, arm large, or adrenaline shake. Arms expected was recorded when the Target’s arms
were perceived to move in a typical or expected pattern. Arms restricted was recorded when the
Target’s arm(s) was not perceived to be moving freely or typically. Arm large was recorded was
the Target’s arm movements were perceived to be larger than what would be expected that
person’s size. An adrenaline shake was recorded when the Target’s arm shook.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Gait (Legs). A step began when the Target’s leg swung forward and the heel struck the ground
and ended when the whole foot was on the ground followed by the forward motion of the back
leg. The number of strides for each Target was recorded. Informed by the subjective responses,
arm movements were categorized as gait expected, gait unnatural / stiff, gait controlled, gait
small, or gait long. Gait expected was recorded when the Target’s gait pattern was perceived as
typical or expected pattern. Gait unnatural / stiff was recorded when the Target’s gait pattern was
perceived as moving in an atypical or rigid manner. Gait controlled was recorded when the
Target’s gait pattern was perceived to be deliberate. Gait small was recorded when the gait
pattern was perceived as being too small for the size of the Target. Gait large was recorded when
the gait pattern was perceived as being too large for the size of the Target. Interrater reliability is
reported in Table 1.
Experiment 2 Coding Procedure. There were two different approaches to coding the subjects’
gait patterns across the nine videos. First, for each video, research assistants entered participants’
qualitative responses into an Excel workbook and a subjective coding system was developed.
Second, the lead author and a research assistant independently recorded behavior across each of
the eight videos. These procedures are described in detail above. Interrater reliability is reported
in Table 1.
Experiment 3 Coding Procedure. There were two different approaches to coding the subjects’
gait patterns across the nine videos. First, for each video, research assistants entered participants’
qualitative responses into an Excel workbook and a subjective coding system was developed.
Second, the lead author objectively and a research assistant independently coded the subjects’
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
behavior across each of the 10 videos. These procedures are described in detail in Experiment 1
above. Interrater reliability is reported in Table 1.
DETECTING CONCEALED OBJECTS IN LEGAL CONTEXTS
Author bio-sketches
Dawn Sweet is an Assistant Professor at the University of Idaho. She earned her PhD in
Communication at Rutgers University. Her research focuses on nonverbal behavior, movement
patterns, bias, decision making, and person perception in the contexts of deception detection,
object concealment, and use of force.
Adele Quigley-McBride has a Ph.D. in Psychology from Iowa State University and is currently a
Research Fellow at the Wilson Center for Science and Justice at Duke Law. Her research focuses
on memory, decision-making, and judgment processes in different legal contexts, including
eyewitness identification, jury-decision making, plea bargaining, and forensic testing procedures.
Christian Meissner is Professor of Psychology at Iowa State University. He holds a Ph.D. in
Cognitive & Behavioral Science from Florida State University (2001) and conducts empirical
studies in applied cognition, including the role of memory, attention, perception, and decision
processes in real world tasks.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
The authors investigated whether accuracy in identifying deception from demeanor in high-stake lies is specific to those lies or generalizes to other high-stake lies. In Experiment 1, 48 observers judged whether 2 different groups of men were telling lies about a mock theft (crime scenario) or about their opinion (opinion scenario). The authors found that observers’ accuracy in judging deception in the crime scenario was positively correlated with their accuracy in judging deception in the opinion scenario. Experiment 2 replicated the results of Experiment 1, as well as P. Ekman and M. O'Sullivan's (1991) finding of a positive correlation between the ability to detect deceit and the ability to identify micromomentary facial expressions of emotion. These results show that the ability to detect high-stake lies generalizes across high-stake situations and is most likely due to the presence of emotional clues that betray deception in high-stake lies.
Article
Full-text available
Across three experiments, we assessed the ability of law enforcement officers and naïve controls to detect the concealment of a weapon or device. Study 1 used a classic signal detection paradigm in which participants were asked to assess whether a target was concealing a neutered 9mm handgun. Study 2 involved a compound signal detection paradigm in which participants assessed whether or not one of several individuals was concealing an unstable device in their backpack. Study 3 moved to a two-alternative forced choice paradigm in which participants evaluated which of two targets was concealing an unstable device in his backpack. Across all three experiments we consistently found no significant differences in detection performance between law enforcement and naïve controls, although participants did perform above chance levels when response bias was free to vary. Furthermore, officers’ years of experience was associated with a bias towards perceiving concealment. Given the frequency with which officers are asked to assess the concealment of weapons or devices, and therein to identify threats, our findings suggest the need for additional research to explore a variety of factors (e.g., context, race of target, operational experience, etc.) likely related to performance on such tasks.
Article
Full-text available
Confirmation bias, as the term is typically used in the psychological literature, connotes the seeking or interpreting of evidence in ways that are partial to existing beliefs, expectations, or a hypothesis in hand. The author reviews evidence of such a bias in a variety of guises and gives examples of its operation in several practical contexts. Possible explanations are considered, and the question of its utility or disutility is discussed.
Article
Experimental sleep restriction and deprivation lead to risky decision-making. Further, in naturalistic settings, short sleep duration and poor sleep quality have been linked to real-world high-risk behaviors (HRB), such as reckless driving or substance use. Military populations, in general, tend to sleep less and have poorer sleep quality than non-military populations due to a number of occupational, cultural, and psychosocial factors (e.g., continuous operations, stress, trauma). Consequently, it is possible that insufficient sleep in this population is linked to HRB. To investigate this question, we combined data from four diverse United States Army samples and conducted a mega-analysis by aggregating raw, individual-level data (n = 2296, age 24.7 ± 5.3). A negative binomial regression and a logistic regression were used to determine whether subjective sleep quality (Pittsburgh Sleep Quality Index [PSQI], Insomnia Severity Index [ISI] and duration [hours]) predicted instances of military-specific HRB and the commission of any HRB (yes/no), respectively. Poor sleep quality slightly elevated the risk for committing HRBs (PSQI Exp(B): 1.12 and ISI Exp(B): 1.07), and longer duration reduced the risk for HRBs to a greater extent (Exp(B): 0.78), even when controlling for a number of relevant demographic factors. Longer sleep duration also predicted a decreased risk for commission of any HRB behaviors (Exp(B): 0.71). These findings demonstrate that sleep quality and duration (the latter factor, in particular) could be targets for reducing excessive HRB in military populations. These findings could therefore lead to unit-wide or even military-wide policy changes regarding sleep and HRB.
Article
Statistical methods are increasingly being used to integrate findings from the ever‐expanding universe of empirical research. Meta‐analysis encompasses various techniques for synthesizing summary statistics, and mega‐analysis pools raw data across studies. This paper offers an introduction to meta‐analysis and mega‐analysis that complements the study of analysis of variance (ANOVA). After a brief conceptual discussion, we provide simple numerical examples.
Article
This article identifies a set of behavioral indicators associated with the carrying of concealed and unholstered handguns. Using qualitative data collected from interviews and focus groups with veteran law enforcement, we outline a variety of cues that, when used appropriately, can help authorities identify people who may be concealing handguns. This work provides a systematic means of assisting the police in identifying and safely interdicting persons who might pose a threat to police or the public. The cues described here contribute to a lexicon that may also serve law enforcement communication, training, and research.
Chapter
People are generally poor at detecting deceit when observing someone’s behaviour or listening to their speech. In this chapter I will discuss the major factors (pitfalls) that lead to failures in catching liars: the sixteen reasons I will present are clustered into three categories: (i) a lack of motivation to detect lies; (ii) difficulties associated with lie detection; and (iii) common errors made by lie detectors. Discussing pitfalls provides insight into how lie detectors can improve their performance (for example, by recognising common biases and avoiding common judgment errors). The second section of this chapter discusses 11 ways (opportunities) to improve lie detection skills. Within this section, I first provide five recommendations for avoiding common errors in detecting lies. Next, I discuss recent lie detection research that introduces novel interview styles aimed at eliciting and enhancing verbal and nonverbal differences between liars and truth tellers. The recommendations are relevant in various settings, from the individual level (e.g., “Is my partner really working late?”) to the societal level (e.g., “Can we trust this suspect when he claims that he is not the serial rapist the police are searching for?”).
Article
Behavioral scientists – including those in psychology, infant and child development, education, animal behavior, marketing, and usability studies – use many methods to measure behavior. Systematic observation is used to study relatively natural, spontaneous behavior as it unfolds sequentially in time. This book emphasizes digital means to record and code such behavior; while observational methods do not require them, they work better with them. Key topics include devising coding schemes, training observers, and assessing reliability, as well as recording, representing, and analyzing observational data. In clear and straightforward language, this book provides a thorough grounding in observational methods along with considerable practical advice. It describes standard conventions for sequential data and details how to perform sequential analysis with a computer program developed by the authors. The book is rich with examples of coding schemes and different approaches to sequential analysis, including both statistical and graphical means.