Conference PaperPDF Available

Trust and Deception in Mediated Communication.

Authors:

Abstract

Guided by interpersonal deception theory and the principle of interactivity, this investigation examined whether communication modalities differentially affect the extent to which group members develop trust or are vulnerable to manipulation and deceit, based on the degree of interactivity the modalities afford. According to the principle of interactivity, involvement and mutuality should increase as one move from text to audio and audiovisual (AV) modalities, to face-to-face (FtF) communication. Under nondeceptive circumstances, greater interactivity should elicit corresponding increases in trust and credibility; under deceptive circumstances, it should produce greater truth biases and inaccurate detection of deceit. This effect should be partly mitigated in text and audio modalities due to the presence of diagnostic deception indicators Pairs were assigned to a truthful or deceptive condition in one of three mediated conditions, or in a face-to-face condition. In the deceptive condition, one member of each pair was enlisted to deceive during the interaction. Following discussion, participants rated their communicative behavior and the credibility of the truthful or deceptive actor. Truth bias and accuracy in judging deceptive information was calculated. Results are compared to previous findings from face-to-face deception. Implications for collaborative technologies are advanced.
Trust and Deception in Mediated Communication
Judee K. Burgoon
Center for the Management of Information, University of Arizona, Tucson, AZ 85721
Gates Matthew Stoner
Program in Integrated Medicine, University of Arizona
Joseph A. Bonito
Department of Communication, University of Arizona
Norah E. Dunbar
Department of Communication Studies, California State University, Long Beach
Abstract
Guided by interpersonal deception theory and the
principle of interactivity, this investigation examined
whether communication modalities differentially
affect the extent to which group members develop
trust or are vulnerable to manipulation and deceit,
based on the degree of interactivity the modalities
afford. According to the principle of interactivity,
involvement and mutuality should increase as one
moves from text, to audio and audiovisual (AV)
modalities, to face-to-face (FtF) communication.
Under nondeceptive circumstances, greater
interactivity should elicit corresponding increases in
trust and credibility; under deceptive circumstances,
it should produce greater truth biases and inaccurate
detection of deceit. This effect should be partly
mitigated in text and audio modalities due to the
presence of diagnostic deception indicators Pairs
were assigned to a truthful or deceptive condition in
one of three mediated conditions, or in a face-to-face
condition. In the deceptive condition, one member of
each pair was enlisted to deceive during the
interaction. Following discussion, participants rated
their communicative behavior and the credibility of
the truthful or deceptive actor. Truth bias and
accuracy in judging deceptive information was
calculated. Results are compared to previous
findings from face-to-face deception. Implications for
collaborative technologies are advanced.
1. Introduction
Successful collaboration and teamwork
often depend on the manner in which participants
exchange and process information [1]. Although
many reasons are given for the failure of teams or
groups to pool and process their
information resources [see 2, 3, one often overlooked
factor is that members may have reasons to withhold
or distort information], such as when group members
wish to conceal their lack of knowledge, have hidden
agendas, possess information they do not wish to
share with others, and have other vested interests that
result in introducing false, faulty, or misleading
information. Under such circumstances, widely held
presumptions about the trustworthiness of group
members and the truthfulness of their communication
[see 4, 5] are no longer valid. Research evidence
from the interpersonal realm reveals that as much as
one-third of daily conversations include some form
of deception, broadly construed to include concealed,
evasive, ambiguous, or exaggerated information as
well as outright lies [6, 7]. Yet deception often goes
undetected in face-to-face (FtF) interchanges, in part
because as interpersonal relationships grow, the
belief in others' truthfulness also grows [8].
Deception, then, stands as a threat to successful
collaborative work.
What happens to trust and deception when
such work is technologically mediated becomes an
interesting and potentially paradoxical one. Several
investigations have shown that distributed work
yields weaker interpersonal relationships and less
trust than does co-located and face-to-face work
[e.g., 9, 10]. Paradoxically, a diminution in trust may
cause potential targets of deception to become more
skeptical and thus to detect deception more
accurately. Alternatively, social identity
deindividuation theory predicts that distributed
collaborations promote group identity [11]. If
individuals develop faith in the group, and lack other
individuating information to assist them in making
veracity judgments, there is the risk that deception
detection will be impaired, making the group more
vulnerable to manipulation and performance
impairment. These effects may be moderated by the
modality under which communication occurs. The
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 1
current investigation was undertaken to address not
only the question of whether different
communication modalities affect trust and users’
abilities to discern deception but also how qualitative
differences in communication patterns across
modalities mediate these effects. Our investigation
was guided by interpersonal deception theory and its
corollary principle of interactivity, both of which we
review next.
2. Interactivity in Interpersonal
Deception
Interpersonal deception theory [IDT, 6] was
initially developed as a propositional framework to
account for the nature and success of deceptive
interchanges by applying principles of interpersonal
communication to the domain of deception.
Encompassing a wide range of related
communication and psychological principles, the
theory illuminates interrelationships among input
(preinteractional), process (interactional), and output
(postinteractional) variables. From its inception, a
central premise of IDT has been that interactive
deception is unlike noninteractive deception. The
principle of interactivity holds that interaction
processes and outcomes are systematically influenced
by the degree of interactivity that is afforded and
transpires [12]. Within the context of deception in
interpersonal encounters, interactivity should affect
senders’ deception displays, ongoing credibility, and
receivers’ accuracy in detecting deception. For
example, being involved in an interaction is
fundamentally different from merely witnessing it,
and that difference often manifests itself as varying
levels of trust and accuracy in detecting deceit [13].
Most extant deception research has been
noninteractive: deceivers do not talk to their intended
targets. Instead, they lie to a third party or record
their lies for later viewing, hearing, or reading by the
intended target who is to judge message veracity.
Because interactive deception is a dynamic rather
than static event, it allows deceivers to modify and
adapt their communicative performance to changing
circumstances. Compared to noninteractive
deception, interactive contexts should give deceivers
more opportunities to deliberately monitor, control,
modify, and repair the content of their messages, the
nonverbal behaviors accompanying those messages,
and their overall demeanor. Moreover, underlying
FtF interaction is a presumption of truthfulness that
creates a truth bias, in which people err in the
direction of perceiving another's communication as
truthful rather than deceptive [e.g., 8, 13, 14, 15-18].
This bias is especially common as people become
more familiar with one another, a perception that is
also fostered by interactivity. The truth bias, coupled
with strategic and adaptive communication by
senders, should result in higher levels of trust,
attributions of sender credibility, and poorer
recognition of sender deceit by receivers.
3. The Principle of Interactivity
Within the context of mediated
communication, the term “interactive” has referenced
a wide array of phenomena but typically refers to the
structural features, or affordances, that are “built
into” a given communication interface or modality
and that enable (“afford”) multiple, tightly
interwoven conversational exchanges and a sense-
making process that references current and prior
messages. For example, technologies that allow
synchronous (same-time), participatory, and
contingent interaction are said to be interactive; those
that interpose time delays between message
transmissions, that place users in an observer or
eavesdropper role, or that fail to provide responses
that are directly related to prior transmissions are said
to be noninteractive. This definition, though quite
serviceable for many purposes, only scratches the
surface of the complex nature of interactivity.
Interactivity actually comprises a constellation of
properties, one or more of which may be the
operative feature leading to the effects ascribed to the
general term “interactivity.” One such property may
be the extent to which interactants have multiple
sensory modalities through which to exchange verbal
and nonverbal cues and hence, to create a tighter
interpersonal web of related messages. As
communication technologies move from FtF to
mediated formats, and from mediated formats with
full access to visual, auditory, tactile, and other
sensory information to ones in which some
modalities are absent, communicators have fewer
sensory channels that are engaged and available to
“interact” with one another. This can create, in the
short-run at least, a chain reaction of dampened
involvement, weakened interactional coordination
and synchrony, reduced sense of mutuality, and
diminished exchange of individuating social
information. Structural affordances, then, have the
potential to amplify or reduce interactivity by virtue
of affecting the qualitative properties of
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 2
communication contained therein. As for the
qualitative properties associated with the interaction
process itself, those that appear to be the sine qua
non of interactive communication include
involvement (the degree to which users feel
cognitively, affectively, and behaviorally engaged in
the interaction), mutuality (the extent to which users
perceive and create relational connection,
interdependence, coordination, and understanding
with one another), and individuation (the degree of
distinctive, personal knowledge of another that sets
that person apart from others). Interactivity, then, can
be judged by the extent to which the communication
process is involving, evinces behavioral patterns and
perceptions of mutuality, and entails exchange of
individuating information.
Interactivity in itself is neither virtue nor
vice. The principle of interactivity merely postulates
that the degree of interdependent, contingent,
participative and synchronous interaction afforded by
a communication interface and/or experienced by
users will systematically and substantially affect
communication processes and outcomes, including
social judgments such as trust, honesty, and other
facets of credibility. In the case of truthful
interchanges, higher interactivity may be associated
with higher trust and perceived credibility.
Conversely, in the case of deceptive interchanges,
higher interactivity may be associated with higher
truth biases and failure to detect deception. We
consider these two respective circumstances in more
detail next.
4. Interactivity When Deception is Absent
It is of course impossible to claim that a
given interaction is devoid of deceit, in light of the
substantial estimates of how much lying, fibbing,
exaggerating, concealing, equivocating, and the like
take place in routine conversation [19, 20].
Nevertheless, we can speak of circumstances under
which deceit is neither highly probable nor actively
introduced. Under these “normal” conditions, the
question is the extent to which communication
modality moderates interactivity and ensuing trust
and truth estimates. We have already noted that the
paradigm case of interactivity is one characterized by
high involvement and mutuality (among other
qualities). The more participants become engaged in
the interaction, establish feelings of connection and
similarity, share a sense of the other as “present” and
receptive, and feel understood, the more likely they
will create the kind of coordinated, synchronized
interaction that, at is most effortless, evokes a mutual
flow of consciousness [21]. Empirical evidence
confirms this: higher degrees of involvement and
mutuality have been shown to create more favorable
perceptions of one's partner or group [22].
Consequently, communication interfaces that create
more involvement and perceived mutuality should
foster more trust and attributions of truthfulness
between participants.
Research by several investigators [e.g., 12,
23] has shown that trust is higher under FtF than
CMC, and should produce more favorable
attributions about another’s sincerity and honesty.
Among CMC modalities, there is evidence [24, 25]
that the auditory channel has unique advantages
relative to text in synchronizing and pacing
interaction in a way that sustains involvement and
facilitates comprehension, revealing receiver states
that serve as back-channel feedback about receiver
understanding and reactions, and through turn-taking
mechanisms, creating a coherently threaded
discourse. Because both audio and AV modalities
include oral speech, the organizing and pacing
properties of the voice may confer benefits on both
of these CMC modalities, although there is some
reason to believe that removing visual distractions
may actual promote more personalized,
“hyperpersonal” communication than when visual
cues are present [26]. Thus, under CMC conditions,
when vocal cues are available, trust and attributions
of truthfulness should be higher than when they are
absent.
5. Interactivity When Deception is
Present
The very features assumed to promote
interactivity also promote truth bias. Mutuality and
involvement contribute to leniency and truth biases in
making judgments of another’s credibility, and with
truth bias comes lowered accuracy in detecting deceit
because telltale indicators of deceit are overlooked,
ignored, or discounted in favor of believing the
person is telling the truth. So, all else being equal,
conditions that create the highest mutuality should be
the worst for detecting deceit. In FtF contexts,
deceivers have been shown to deliberately and
successfully modify their performances over time, in
part because they respond to any observed skepticism
on the part of receivers by working harder to appear
normal, engaged, and pleasant. If mediated
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 3
communication attenuates not only the sense of
mutuality and level of involvement but also the total
available amount of feedback, then deceivers may not
accrue the same advantages present in FtF. Thus,
those mediated conditions that foster highest
mutuality should be the same ones that create the
highest trust and truth biases and the least accurate
detection of deception.
But, all else is not equal. Additional factors
are also at work. It has been argued, and
demonstrated empirically, that deception is often a
more difficult task than telling the truth [6]. It
requires more mental “heavy lifting,” hence is
referred to as creating more cognitive “load” than
truthtelling because of the demands of creating
plausible and coherent verbal messages while also
monitoring and managing accompanying nonverbal
behavior and weaving it all into a congruent whole.
To the extent that the act of deceit creates more
cognitive load for message senders, deceivers should
experience more challenges in putting together a
credible deceptive performance than a truthful one.
The question, then, is whether the degree of
interactivity afforded by different communication
modalities eases or exacerbates this load.
FtF might be thought to have highest load
because of the number of verbal and nonverbal
channels and features needing to be managed. But the
same demands apply to message receivers as well as
senders, and deceivers have been shown to be rather
adept at managing deceit under these conditions. For
deceivers, as the number of channels to manage
decreases, the management tasks should also
decrease, allowing them to turn their attention to
more careful management of those that remain.
Additionally, fewer channels means fewer
opportunities for channel discrepancies, which have
been shown to be more common under deception
than truth and which can tip off receivers to the
presence of deceit. Moreover, when communicating
via text, senders also have opportunities to plan and
edit their messages before transmitting. This brief
time lapse between sender and receiver messages–a
slight bit of asynchronicity, if you will-- relative to
the immediate conversational turn-switching required
under FtF, audioconferencing and videoconferencing
modalities gives senders a further advantage and
receivers a further disadvantage. Deception detection,
then, might be the least accurate under text
conditions for receivers untrained in the subtle,
inadvertent deception indicators available in text.
Finally, although senders do attempt to control the
verbal and nonverbal features of their deceptive
communication, they are commonly less prone to
monitor and successfully manage their voice than
their face and body, so the audio channel leaks
indicators to deceit, some of which elicit suspicion
from receivers [27, 28]. Thus, among mediated
conditions, receivers may be most successful at
detecting deceit when using audio-based formats
(such as cell phones and audio-conferencing) and
least successful when using text formats (such as
email and instant messaging).
6. Hypotheses and Research Questions
In the experiment to be reported, participants
completed a decision-making task either FtF or under
one of three distributed, mediated conditions: text,
audio, or audiovisual (AV). In addition, dyads were
randomly assigned to either a deception or truthful
condition. In the former, one of the participants was
instructed to provide deception information to his or
her partner, whereas participants in the truthful
condition were simply allowed to act normally. If the
principle of interactivity holds, and visual and
auditory nonverbal cues provide for richer
interactions characterized by high involvement and
mutuality, then conditions should be most favorable
for high degrees of involvement and mutuality as one
moves from text to audio and AV to FtF modalities.
Because the special properties of oral
communication–e.g., interactional synchrony,
pressure to tightly link conversational turns, greater
message comprehension, and rapid exchange of
information--may be sufficient to overcome any
deficits from losing visual cues, we left as a research
question whether the AV modality affords more
interactivity than the audio-only condition. Thus, we
hypothesized as follows:
H1: Involvement and mutuality are greatest under
FtF communication, followed by AV and audio forms
of mediated communication, and lastly, text.
RQ: How do the audio and AV modalities compare
on involvement and mutuality?
IDT predicts that the introduction of
deception should adversely impact senders’
performance during the early stages of the
interaction, leading to behavioral patterns of reduced
involvement and deviations from normal behavioral
patterns that can trigger suspicions. Conceivably,
receivers attuning to such behavioral abnormalities
may feel less mutuality. However, IDT holds that
such difficulties should be transitory, as deceivers
adjust to receiver feedback. In the current
experiment, deception was also only present during
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 4
the first task. Thus, Hypothesis 2 sought to assess the
adverse impact of deception on interactivity and the
extent to which it would persist or disappear after the
first task by positing a deception by time interaction:
H2: Relative to truthtellers, deceivers (a) exhibit
lower involvement and mutuality initially in
interactions and (b) converge toward truthtellers’
levels of involvement later in interactions.
Given our stance that modalities merely
create favorable or unfavorable conditions for
involvement and mutuality to materialize and that it
is these communication properties of interactivity
that are truly responsible for resultant outcomes; and,
given our previous findings that higher involvement
and mutuality are associated with more favorable
social judgments, a natural corollary is as follows:
H3: Trust and truth estimates are positively
correlated with involvement and mutuality.
It follows, then, that under nondeceptive
conditions, trust and truth estimates should increase
as one moves from text to audio and AV to FtF
modalities. However, adding deception to the mix
may modify this rank-ordering because the
diagnosticity of deception indicators available in the
text and audio channels should partially offset the
interactivity effects. Thus, under deceptive
conditions, trust, truth estimates, and truth bias
should be greatest in the FtF condition followed by
the AV and then the text and audio conditions. This
deception by modality interaction is reflected in
Hypotheses 3 and 4. Under nondeceptive conditions,
it makes little sense to speak of truth biases, because
we would expect all estimates to be in the upper end
of the spectrum based on a combination of the actual
veracity of the messages and the truth bias [29]. But,
in the deceptive condition, it becomes useful to know
the extent to which estimates are actually above the
midpoint of a truth estimate scale and thus constitute
absolute truth biases. Consequently, H4 was worded
in terms of truth bias.
H4: Under nondeceptive conditions, trust and truth
estimates are (a) higher under FtF than mediated
communication and (b) higher under AV and audio
than under text communication.
H5: Under deceptive conditions, trust and truth
biases are (a) higher under FtF than mediated
communication and (b) higher under AV than audio
and text communication.
7. Method
7.1 Sample
Participants (N=128) were undergraduate
students, recruited from a mass-lecture
communication course at a large southwestern
university, who received extra credit for their
participation. Participants were paired to form 64
same-sex dyads.
7.2 Independent Variables
The experiment was a 4 (Modality) x 2
(Truth/Deception) design with cells balanced by
gender. The four modalities consisted of (1) FtF, (2)
text, (3) audio, and (4) AV communication. In the
FtF condition, participants were seated in caster
armchairs facing one another, with separate computer
terminals available nearby for each to complete
written materials. Participants in the mediated
conditions reported to separate locations in the same
building and interacted via the Microsoft NetMeeting
program. In the text condition, they were seated at
computer terminals in separate rooms and
communicated via the chat window. In the video
condition, in place of the chat window, two small
windows presented the participants with a view of
their partner as well as the image of themselves being
transmitted. The audio signal was transmitted via
computer as well. In the audio condition, the video
link and chat window were disabled. In mediated
conditions, participants were unaware of the
proximity of their partner (i.e., that he or she was
located in the same building).
Truth/deception was manipulated by
randomly assigning participants to the truth or
deception conditions. Within the deception
conditions, one member of each dyad was randomly
assigned the role of deceiver (designated as Person
A) and given special instructions. The other (Person
B) received no special instructions.
7.3 Procedures
Participants, randomly assigned the role of
Person A or Person B, reported to separate rooms in
the Communication Research Lab (CRL). The CRL
consists of four interaction rooms which are
networked via TCP/IP and Ethernet. Each room
contains a Pentium computer, Connectix video
camera, and video-conferencing software. After
completing consent forms and the Social Skills
Inventory [30] in their respective intake rooms,
Persons A who were randomly assigned to the
deception condition received their deception
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 5
instructions. Persons A and B in the FtF condition
were then brought together in one of the interaction
rooms. Those in the mediated conditions remained in
their respective locations.
All participants next received instructions
for conducting a “get-acquainted” social task in
which they were to discuss, in order, a series of
topics printed on index cards. Topics
included....Discussions lasted approximately 10
minutes. At the conclusion of this discussion,
participants completed a Web-based questionnaire.
They then proceeded to conduct a second, decision-
making task which consisted of an interactive
management case study on employee motivation
[31]. No deception occurred during this task. At the
conclusion of this task, participants again completed
a Web-based questionnaire. They were then
scheduled to conduct a third task a week later (results
of which are to be reported elsewhere), after which
they were debriefed and thanked.
7.4 Pre-Interaction and Post-Interaction
Measures
All measures used were modified from
previous IDT and interactivity studies. To measure
interaction involvement, participants rated perceived
involvement with four Likert-format items taken
from Burgoon and Hale’s [32] Relational
Communication Scale (coefficient alpha reliability =
.82 and .87 after the first and second tasks,
respectively). To capture the range of possible
perceptions that might correspond to mutuality,
participants also rated partners on four
receptivity/similarity items from the RCS
(reliabilities = .80 and .81) and three items from
McCroskey, Hamilton, and Weiner’s [33] homophily
scale (reliabilities = .88 and .92). To these measures
were added Aron, Aron, and Smollan’s [34] pictorial
instrument, which uses seven increasingly
overlapping circles to depict degrees of perceived
connectedness, and three items from Cahn and
Shulman’s [35] Feelings of Understanding/-
Misunderstanding Scale (reliability = .66 and .79).
Involvement and mutuality were measured
immediately after the social get-acquainted task and
again after the decision-making task.
Assessments of the partner trust, completed
after the get-acquainted task, utilized items taken
from credibility measures developed by McCroskey,
Hamilton, and Weiner [33] and Wheeless and Grotz
[36]. The specific attributes of truthful, trustworthy,
high character, and very credible obtained coefficient
alpha reliability of .75. Estimates of partner
truthfulness during the get-acquainted discussion
were obtained at the conclusion of the session’s tasks
so as not to create any reactivity during the decision-
making task (during which no deception took place).
Participants rated on a 0 (not at all truthful) to 10
(completely truthful) scale their partner’s honesty on
each of the four topics.
8. Results
8.1 Hypotheses 1and 2
Communication process measures are
typically highly intercorrelated and more
parsimoniously understood as a set of interactivity
indicators. Thus, we initially tested the H1 and H2
variables as a set to determine if the measures
collectively showed modality and deception effects.
Analyses were conducted on ratings taken
immediately following the get-acquainted task and
also at the end of the session, after completing the
decision-making task. The latter measures assess
persistence over time. The initial analysis was a 4
(modality) x 2 (deception) x 2 (time: first and second
administrations of the measures following each task)
x 2 (role: Person A or B) x 5 (dependent measures)
repeated measures analysis of variance, with the last
three factors repeated. (Dependent measures were
included as a factor in the statistical analysis to
control for multicollinearity and to pinpoint where
differences existed among measures.) Where there
were violations of compound symmetry conditions
(e.g., a significant Mauchly’s test of sphericity for
measure), we used Huynh-Feldt adjusted degrees of
freedom in the averaged univariate analyses. The
five dependent measures included involvement and
the four mutuality measures of connectedness,
feeling understood, homophily, and receptivity.
In preface, the initial omnibus analyses
produced several significant within-subjects effects
that place the hypothesized effects in context. First,
there was a significant effect for time, F(1,94) =
17.26, p <.001, partial Ș² = .16, showing that,
regardless of condition, involvement and mutuality
increased over time. A significant measure by time
interaction, F(3.40,319.48) =13.47, p<.001, partial Ș²
= .12, revealed differential changes across the two
tasks for the different measures. A significant
measure by modality interaction, F(8.67.272) = 3.36,
p =.001, partial Ș² = .10, indicated that the modality
effects differed by measure. Finally, there was a
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 6
deception by measure interaction, F(3.18,305.18) =
9.71, p <.001, partial Ș² = .09, and deception main
effects after the get-acquainted task, F(5,92) = 2.76,
p=.023, partial Ș² = .13, and after the decision-
making task, F(5,94) = 5.30, p<.001, partial Ș² = .22.
The latter, in addition to supporting H2, argued for
conducting the contrast tests for H1 separately within
truth and deception conditions. The focused
contrasts that corresponded to H1 were then
conducted on the individual measures averaged
across the two time periods. The 3 df contrasts
followed a modified Helmert contrast scheme: the
first compared FtF to the mediated conditions, the
second compared audio and AV to text, and the third
compared AV to audio. We used a p < .05, one-
tailed, criterion for the first two directional
comparisons (which is the same as p < .10, two-
tailed) and a p < .05, two-tailed, criterion for the
third, nondirectional one.
Means shown in Table 1 reveal that the first
contrast was significant for involvement and feeling
understood within the deception condition; it was not
within the truth condition. The second contrast,
comparing audio and AV to text, was significant for
the same two measures under deception on both the
averaged measures and measures collected
immediately after the get-acquainted task, and also
on connectedness under truth (first task only). The
third contrast, comparing audio to AV, was
significant for all measures except understanding
under deception and for homophily and involvement
on the averaged measures under truth plus on
receptivity after task one. (Figures 1a and 1b show
the effects of modality on each of the measures,
averaged across time.)
Table 1. Means and standard deviations for all
measures, averaged across tasks.
MEASURE Mode Truth Deception
Mean SD mean SD
Connectedness Text 2.93 .98 4.29 1.12
Audio 4.00 1.31 4.04 1.16
AV 4.15 1.13 4.81 1.32
FtF 3.63 1.23 3.92 1.12
Receptivity Text 5.49 1.15 5.50 1.00
Audio 5.88 .89 5.44 .56
AV 5.44 .83 5.80 .94
FtF 5.79 .84 5.66 .79
Homophily Text 4.44 1.13 4.28 1.38
Audio 4.38 1.02 3.80 1.13
AV 4.66 .92 4.48 1.57
FtF 4.97 1.24 4.23 1.02
Understanding Text 5.43 1.06 5.07 1.12
Audio 5.77 .82 5.43 .62
AV 5.52 .46 5.61 .95
FtF 5.78 .91 5.71 1.03
Involvement Text 5.39 .99 4.76 1.15
Audio 5.59 .83 4.87 .70
AV 5.33 .57 5.40 .89
FtF 5.56 1.08 5.33 1.08
Trust Text 5.36 1.07 4.98 1.18
Audio 5.89 1.04 5.11 1.11
AV 5.56 .78 5.72 .64
FtF 5.39 .96 5.48 1.02
Truth Estimate Text 7.18 3.04 8.33 1.91
Audio 8.52 1.81 7.07 2.33
AV 9.08 .88 8.13 1.44
FtF 8.06 2.15 7.85 2.08
Figure 1. Effects of modality on involvement and
mutuality under truth (a) and deception (b).
b. Under Deception
MODALITY
FtFAudiovisualAudioText
Mean Ratings
6.0
5.5
5.0
4.5
4.0
3.5
MEASURES
Involvement
Connectedness
Receptivity
Homophily
Feeling Understood
Involvement and feeling understood were
higher under FtF than under mediated conditions and
under audio and AV than text when deception was
present. Moving from FtF to leaner modalities
a. Under Truth
MODALITY
FtFAudiovisualAudioText
Mean Ratings
6.0
5.5
5.0
4.5
4.0
3.5
3.0
MEASURES
Involvement
Connectedness
Receptivity
Homophily
Feling Understood
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 7
produced corresponding declines in the interactive
qualities of involvement and feeling understood
when one partner had engaged in deception,
indicating that availability of nonverbal cues was
relevant in establishing a sense of involvement and
mutuality under such conditions. Interestingly, those
effects persisted beyond the deceptive task to a
nondeceptive one. Although the patterns were quite
similar under truth (as evident from the figures),
these same effects did not reach statistical
significance. Only feelings of connectedness differed,
being higher under AV and audio than text. Thus,
Hypothesis 1 received far more support under
deception than truth.
H2 predicted that deception would initially
dampen involvement and mutuality but that
deception and truth would converge by the end of the
second task. The hypothesized interaction did not
obtain. Instead, deception reduced involvement and
mutuality relative to truth but its effects continued to
persist beyond the initial task, despite the fact that
deception was no longer present. The impact was
most applicable to involvement, F(1,96) = 4.30,
p=.041, partial Ș² = .04, after the get-acquainted task,
and to involvement, F(1,98) = 3.85, p=.05, partial Ș²
= .04, connectedness, F(1,98) = 7.81, p=.006, partial
Ș² = .07, and homophily, F(1,98) = 2.72, p=.10,
partial Ș² = .03, after the decision-making task (see
Table 1 for means averaged across the two tasks). In
sum, H2 was partially supported: Deception lowered
interactivity in terms of involvement, felt similarity,
and felt understanding. Contrary to the hypothesis,
these effects continued past the deceptive task, with
the exception that the changes across questions in the
effects of deception under text did demonstrate some
temporal adjustments in favor of deceivers becoming
less detectable as the interaction progressed.
The research question asked whether audio
or AV modalities would elicit more interactivity. The
focused contrast tests comparing audio to AV
indicated that under truth, involvement and
homophily were higher under audio than AV yet
under deception, all measures except understanding
showed higher ratings under AV than audio.
8.2 Hypothesis 3
Hypothesis 3 predicted that involvement and
mutuality measures would be positively correlated
with trust and with truth estimates. Correlations were
computed between trust and the involvement and
mutuality measures, which were taken following
each task, and with truth estimates, which were taken
at the conclusion of the session (so as not to
introduce suspicion during the case study task). They
were calculated separately for Persons A (who
included deceivers) and Persons B. This hypothesis
was fully supported for both criterion measures and
for both roles. Trust and truth estimates were higher,
the more participants felt involvement and mutuality
with their partners. Feelings of understanding and
receptivity showed especially strong relationships to
trust and truth bias. (Table of contrast tests available
from authors upon request).
8.3 Hypotheses 4 and 5
Hypotheses 4 and 5 were tested on Person B
data only, inasmuch as half of those in the Person A
role were deceivers. Although there was no modality
by deception interaction on either trust or truth, there
was some support for the hypothesized relationships
in the simple effect tests. H4b (truth condition)
received support in that the AV and audio conditions
elicited higher truth estimates than the text condition,
t(49) = 2.28, p = .01, one-tailed. Contrary to H4a, FtF
did not elicit higher truth ratings. Related to H5b
(deception condition), trust and truth estimates did
not differ between FtF and mediated conditions or
between text and the other two mediated conditions.
However, the AV condition elicited more trust than
the audio condition, t(55) = 1.75, p = .04, one-tailed.
Moreover, simple effect tests within modality
between truth and deception revealed that trust was
much higher under truth than deception when
participants interacted within the audio modality,
t(28) = 2.00, p = .03, one-tailed. Truth estimates were
also much higher under truth than deception within
audio, t(24) = 1.75, p = .04, one-tailed, and within
AV, t(29) = 2.11, p = .02, one-tailed, indicating that
believability suffered under these modalities when
deception was present. These patterns are shown in
Figures 2a and 2b. Finally, the text condition
produced a question by deception interaction, F(3,75)
= 3.36, p = .02, such that the last two questions were
actually judged as more truthful under deception than
under truth when participants interacted via text (see
Figure 2c). In other words, deceivers were more
successful over the course of time in creating
believable text messages than were truthful
participants. In sum, modality did influence
believability. Under nondeceptive circumstances,
participants were best served in
appearing trustworthy and truthful when using audio
or AV modalities. However, under deceptive
circumstances, trust dropped under the audio
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 8
modality, and truth estimates were lower under both
audio and AV, indicating that these modalities were
more revelatory of 0deception than were the other
conditions. And, deceivers were actually more
believable under text as questioning progressed,
indicating that this was the least "leaky" modality for
revealing deceit and therefore the one that was best
for evading detection.
Figure 2. Effects of Deception and Modality on (a) Trust
and (b) Truth Estimates across Modalities and (c) across
Questions within Text.
a. Trust
MODALITY
FtFAudiovisualAudioText
Mean Trust Ratings
6.0
5.8
5.6
5.4
5.2
5.0
4.8
TRUTH/DECEPTION
Deception
Truth
b. Truth Estimates
MODALITY
FtFAudiovisualAudioText
Mean Truth Ratings
10.0
9.5
9.0
8.5
8.0
7.5
7.0
6.5
6.0
5.5
5.0
TRUTH/DECEPTION
Deception
Truth
c. Truth Estimates across
Questions within Text
QUESTION
4321
Mean Truth Estimates
10.0
9.5
9.0
8.5
8.0
7.5
7.0
6.5
6.0
5.5
5.0
TRUTH/DECEPTION
Deception
Truth
9. Discussion
Ever since researchers have been comparing
FtF to mediated forms of communication, there have
been disputes about whether FtF communication is a
necessary and/or sufficient condition for creating
trust, group morale, and effective group performance.
In many cases, in straight FtF to text-based CMC
comparisons, CMC has fallen short, leading to the
conclusion that some form of FtF, either as the sole
means of communication, or as a prerequisite to
CMC, is needed for groups to establish common
ground, solidarity, and trust [see, e.g., 23].
The current investigation disputes these
claims. Clearly, participants in this investigation were
able to establish trust and mutuality without meeting
face-to-face. In fact, across all conditions,
involvement and mutuality increased over time
indicating that even in mediated formats, as people
get to know one another and become accustomed to
interacting with each other, they can share a
connection that influences their perceptions of one
another. Further, a close examination of the means in
Table 1 reveals that participants in mediated formats
established greater involvement, mutuality, and trust
than those in the FtF modality under some
conditions. This stands in contrast to earlier claims
that to create trust and mutuality, individuals must
meet FtF.
Tests of the interaction of modality and
deception produced some of the most informative
results regarding the conditions most likely to
engender trust and the conditions most likely to
impair deception detection. Under "truthful"
conditions, the two mediated modalities that include
oral speech--audio and AV-- generated the most
trust; and the three that include both auditory and
visual nonverbal information--audio, AV, and FtF--
elicited highest estimates of truth. The audio
modality was especially facilitative of trust
perceptions. Text suffered by comparison. On
average, trust and truth estimates were the lowest
under text communication. This modality, then, is not
the optimal choice when the objective is to achieve
trust among interactants. When text is the only
choice, one must be cognizant that trust is more
fragile than under other modalities and may require
extra measures to bolster such trust. Of course, as the
interactivity results show, the key is in creating
involvement and mutuality among participants.
When deception was introduced, it also
generally had a dampening effect on judgments.
Even so, truth bias was still evident in that all truth
estimates were above the midpoint of the range.
conditions. Thus, the tendency for CMC users to
believe fellow users persists, even though users seem
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 9
to recognize at some less-than-conscious level that
something is amiss. The modality under which truth
bias was least acute and where trust also plummeted
was the audio channel. This means that the audio
channel is a "leaky" channel, one in which CMC
users are most likely to detect deceit. Comparatively,
the condition in which audio cues are absent, i.e.,
text, was the least likely to detect deceit. Even though
trust was also at the lowest level in this condition,
deceivers were more believable than truthtellers
under this modality, especially on later questions,
making it perhaps the most ripe for manipulation and
misuse. Also, consistent with the "seeing is
believing" visual bias, the channels in which visual
information was available (AV and FtF) offered
relatively poor discrimination between truth and
deception, suggesting that visual information may
serve as a distraction or source of misdirection when
one user is deceiving others. Therefore, more is less;
the more visual information available, the less
accurate the detection of deceit.
This investigation challenges some
traditional beliefs in the CMC literature. First, it
demonstrates that FtF communication is not essential
for establishing trust or mutuality and that mediated
formats can be used successfully for group tasks or
establishing interpersonal relationships. In fact, in
many cases, CMC modalities are on a par with or
even superior to FtF for these goals. Second, this
research demonstrates that FtF is neither the ideal
venue for detecting deceit nor for perpetrating it.
Results show that FtF conditions did not elicit higher
truth ratings than CMC, indicating that the truth bias
is active even in mediated formats. The FtF truth and
deception conditions also did not differ from one
another on trust and truth estimates, indicating that
FtF interaction would be the effective modality for
discriminating truth from deception. By comparison,
the audio condition appears to be the ideal condition
for detecting deceit in that interactants had the most
trust for truthful senders and the least trust for
deceptive ones under this modality. Third, the
variability across the modalities in differentiating
truthfulness and trust indicates that future
investigators cannot simply compare FtF messages to
CMC (often tested only as text). We must separate
out different forms of mediation, and must not
combine different formats or compare different
formats across studies under the global heading of
CMC. A high-priority research issue must be
establishing what aspects of various modalities
contribute to interactivity, rather than assuming that
FtF must be replicated in CMC.
10. References
[1] V. B. Hinsz, R. S. Tindale, and D. A. Vollrath, "The
emerging conceptualization of groups as information
processors," Psychological Bulletin, vol. 121, pp. 43-64,
1997.
[2] G. Stasser, D. D. Stewart, and G. M. Wittenbaum,
"Expert roles and information exchange during discussion:
The importance of knowing who knows what," Journal of
Experimental Social Psychology, vol. 31, pp. 244-265,
1995.
[3] G. Stasser, S. I. Vaughan, and D. D. Stewart, "Pooling
unshared information: The benefits of knowing how access
to information is distributed among group members,"
Organizational Behavior and Human Decision Processes,
vol. 82, pp. 102-116, 2000.
[4] H. P. Grice, "Logic and conversation," in Syntax and
semantics 3: Speech acts, P. C. Morgan, Ed. New York:
Academic Press, 1975, pp. 41-58.
[5] R. J. Lewicki, D. J. McAllister, and R. J. Bies, "Trust
and distrust: New relationships and realities," Academy of
Management Review, vol. 23, pp. 438-458, 1998.
[6] D. B. Buller, J. K. Burgoon, A. Buslig, and J. Roiger,
"Testing interpersonal deception theory: The language of
interpersonal deception," Communication Theory, vol. 6,
pp. 268-289, 1996.
[7] P. Ekman, "Why don't we catch liars?," Social
Research, vol. 63, pp. 801-818, 1996.
[8] S. A. McCornack and M. R. Parks, "Deception
detection and relationship development: The other side of
trust," in Communication yearbook, vol. 9, M. L.
McLaughlin, Ed. Beverly Hills CA: Sage, 1986, pp. 377-
389.
[9] J. K. Burgoon, J. A. Bonito, and K. Kam,
"Communication and trust under face-to-face and mediated
conditions: Implications for leading from a distance," in
Leadership at a distance, S. Weisband and L. Atwater,
Eds. Mahwah, NJ: Erlbaum, 2002.
[10] M. A. Griffin, M. G. Patterson, and M. A. West, "Job
satisfaction and teamwork: The role of supervisor support,"
Journal of Organizational Behavior, vol. 22, pp. 537-550,
2001.
[11] L. C. Tidwell and J. B. Walther, "Computer-mediated
communication effects on disclosure, impressions, and
interpersonal evaluations: Getting to know one another a
bit at a time," Human Communication Research, vol. 28,
pp. 317-348, 2002.
[12] J. K. Burgoon, D. B. Buller, and K. Floyd, "Does
participation affect deception success? A test of the
interactivity principle," Human Communication Research,
vol. 27, pp. 503-534, 2001.
[13] N. E. Dunbar, A. Ramirez, Jr., and J. K. Burgoon,
"Interactive deception: Effects of participation on
participant-receiver and observer judgments.,"
Communication Reports, in press.
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 10
[14] D. B. Buller, K. D. Strzyzewski, and F. G. Hunsaker,
"Interpersonal deception: II. The inferiority of
conversational participants as deception detectors,"
Communication Monographs, vol. 58, pp. 25-40, 1991.
[15] H. H. Clark and E. V. Clark, Psychology and
language: An introduction to psycholinguistics. New York:
Harcourt Brace Jovanovich, 1977.
[16] P. J. Kalbfleisch, "Deceit, distrust and the social
milieu: Application of deception research in a troubled
world," Journal of Applied Communication Research, vol.
20, pp. 308-334, 1992.
[17] M. O'Sullivan, P. Ekman, and W. V. Friesen, "The
effect of comparisons on detecting deceit," Journal of
Nonverbal Behavior Special Issue: Deception, vol. 12, pp.
203-215, 1988.
[18] M. Zuckerman, S. A. Fischer, R. W. Osmun, B. A.
Winkler, and et al., "Anchoring in lie detection revisited,"
Journal of Nonverbal Behavior, vol. 11, pp. 4-12, 1987.
[19] B. M. DePaulo, D. A. Kashy, S. E. Kirkendol, M. M.
Wyer, and J. A. Epstein, "Lying in everyday life," Journal
of Personality & Social Psychology, vol. 70, pp. 979-995,
1996.
[20] R. E. Turner, C. Edgley, and G. Olmstead,
"Information control in conversations: Honesty is not
always the best policy," Kansas Journal of Speech, vol. 11,
pp. 69-89, 1975.
[21] M. Csikszentmihalyi and I. S. Csikszentmihalyi,
Optimal experience: Psychological studies of flow in
consciousness. New York: Cambridge University Press,
1988.
[22] J. K. Burgoon, J. A. Bonito, B. Bengtsson, A.
Ramirez, and N. Dunbar, "Testing the Interactivity Model:
Communication Processes, Partner Assessments, and the
Quality of Collaborative Work," Journal of Management
and Information Systems, vol. 16, pp. 33-56, 2000.
[23] E. Rocco, "Trust breaks down in electronic contexts
but can be repaired by some initial face-to-face contact,"
presented at CHI, Los Angeles, CA, 1998.
[24] C. Jensen, S. D. Farnham, S. M. Drucker, and P.
Kollock, "The effect of communication modality on
cooperation in online environments," presented at
Proceedings of CHI 2000, Hague, Amsterdam, 2000.
[25] G. M. Stoner, "Decision-making via mediated
communication: Effects of mediation and time pressure," in
Department of Communication. Tucson, AZ: University of
Arizona, 2001.
[26] J. B. Walther, "Computer-mediated communication:
Impersonal, interpersonal, and hyperpersonal interaction,"
Communication Research, vol. 23, pp. 3-43, 1996.
[27] J. K. Burgoon and D. B. Buller, "Interpersonal
deception: III. Effects of deceit on perceived
communication and nonverbal behavior dynamics,"
Journal of Nonverbal Behavior, vol. 18, pp. 155-184, 1994.
[28] J. K. Burgoon, D. B. Buller, and L. Dillman,
"Interpersonal deception: IV. Effects of suspicion on
perceived communication and nonverbal behavior
dynamics," Human Communication Research, vol. 22, pp.
163-196, 1995.
[29] H. S. Park and T. R. Levine, "A probability model of
accuracy in deception detection experiments,"
Communication Monographs, vol. 68, pp. 201-210, 2001.
[30] R. E. Riggio, "Assessment of basic social skills,"
Journal of Personality and Social Psychology, vol. 51, pp.
649-660, 1986.
[31] D. J. C. Moberg, D. F., Interactive Cases in
Organizational Behavior, 2 ed. New York: HarperCollins,
1995.
[32] J. K. Burgoon and J. L. Hale, "Validation and
measurement of the fundamental themes of relational
communication," Communication Monographs, vol. 54, pp.
19-41, 1987.
[33] J. C. McCroskey, P. R. Hamilton, and A. M. Weiner,
"The effect of interaction behavior on source credibility,
homophily, and interpersonal attraction," Human
Communication Research, vol. 1, pp. 42-52, 1974.
[34] A. Aron, E. N. Aron, and D. Smollan, "Inclusion of
other in the self scale and the structure of interpersonal
closeness," Journal of Personality and Social Psychology,
vol. 63, pp. 596-612, 1992.
[35] D. D. Cahn and G. M. Shulman, "The perceived
understanding instrument," Communication Research
Reports, vol. 1, pp. 122-125, 1984.
[36] L. R. Wheeless and J. Grotz, "The measurement of
trust and its relationship to self-disclosure," Human
Communication Research, vol. 3, pp. 250-257, 1977.
Portions of this research were supported by funding from the U. S.
Army Research Institute (Contract #DASW01-98-K-009). The
views, opinions, and/or findings in this report are those of the
authors and should not be construed as an official Department of
the Army position, policy, or decision. The authors wish to
acknowledge the assistance of Artemio Ramirez, Jr. and Karadeen
Kam in preparing and conducting aspects of this investigation.
Proceedings of the 36th Hawaii International Conference on System Sciences - 2003
0-7695-1874-5/03 $17.00 (C) 2003 IEEE 11
... In the literature, the consensus is that English deceptive texts exhibit some common linguistic characteristics. Deceptive texts are longer than those that are non-deceptive, showing less lexical diversity (i.e., fewer repeated words) and lower syntactic complexity [9,10]. Mobility verbs, such as 'go', 'walk' or 'move', were also found to be more frequent in deceptive texts, according to [2,11,12]. ...
... First, deceptive texts were found to be longer than truthful ones [9,10]. In Table 1, we already described the characteristics of the corpus in terms of the document length average. ...
... However, note that 7 documents out of the 35 fraudster documents are longer than the non-fraudster average, and that 18 out of the 60 non-fraudster documents are shorter than the non-fraudster length average. Second, deceptive texts were found to have lower lexical diversity or richness than truthful ones [9,10]. The Type-Token Ratio (TTR) is the usual measure to assess lexical diversity: the closer the ratio is to 1, the greater the diversity. ...
Article
Full-text available
The goal of our research was to assess whether the observation about deceptive texts having a lower positive tone than truthful ones in terms of sentiment could become operative and be used for building a classifier in the particular case of fraudster’s letters written in Spanish. The data were the letters that CEOs address to company shareholders in their annual financial reports, and the task was to identify the letters of companies that committed financial misconduct or fraud. This case was challenging for two reasons: first, most of the research worked with spontaneous written or spoken texts, while these letters did not; second, most of the research in this area worked on English texts, while we validated the linguistic cues found as evidence of deception for Spanish texts. The results of our research confirm that an SVM trained with a bag-of-words model of frequent adjectives can achieve 81% accuracy because these adjectives bring the information about which positive or negative tone and which word combinations in a text turn out to be a characteristic of fraudster’s texts.
... Most studies in the field of engineering, focused on investigating trust, have been based on speech analysis [37], where it has often been contrasted with deception [15,38,39]. Some studies have also investigated textual/lexical features [40,41], as well as facial expressions [42]. ...
Article
Full-text available
This paper explores the automatic prediction of public trust in politicians through the use of speech, text, and visual modalities. It evaluates the effectiveness of each modality individually, and it investigates fusion approaches for integrating information from each modality for prediction using a multimodal setting. A database was created consisting of speech recordings, twitter messages, and images representing fifteen American politicians, and labeling was carried out per a publicly available ranking system. The data were distributed into three trust categories, i.e., the low-trust category, mid-trust category, and high-trust category. First, unimodal prediction using each of the three modalities individually was performed using the database; then, using the outputs of the unimodal predictions, a multimodal prediction was later performed. Unimodal prediction was performed by training three independent logistic regression (LR) classifiers, one each for speech, text, and images. The prediction vectors from the individual modalities were then concatenated before being used to train a multimodal decision-making LR classifier. We report that the best performing modality was speech, which achieved a classification accuracy of 92.81%, followed by the images, achieving an accuracy of 77.96%, whereas the best performing model for text-modality achieved a 72.26% accuracy. With the multimodal approach, the highest classification accuracy of 97.53% was obtained when all three modalities were used for trust prediction. Meanwhile, in a bimodal setup, the best performing combination was that combining the speech and image visual modalities by achieving an accuracy of 95.07%, followed by the speech and text combination, showing an accuracy of 94.40%, whereas the text and images visual modal combination resulted in an accuracy of 83.20%.
... In recent years, engineering education and research has witnessed a huge spike in academic misconducts including plagiarism and contract cheating. It was pointed out by several researchers ( [1], [2], [3], [4] etc.) that students who feel disconnected are more prone to engage in deceptive behaviour including academic dishonesty. Due to COVID19 global crisis, more and more classrooms are going online which only aggravates the situation. ...
... In recent years, engineering education and research has witnessed a huge spike in academic misconducts including plagiarism and contract cheating. It was pointed out by several researchers ( [1], [2], [3], [4] etc.) that students who feel disconnected are more prone to engage in deceptive behaviour including academic dishonesty. Due to COVID19 global crisis, more and more classrooms are going online which only aggravates the situation. ...
Conference Paper
Various types of academic misconducts including plagiarism have recently been a matter of concern for engineering educators and administrators. Apart from being a form of academic misconduct, plagiarism hinders the students from developing critical thinking and analysis abilities. If the issue is related with another newer form of academic misconduct known as contact cheating, where students can hire a third party to create their assignments, it poses serious threat to academic standard, quality of education and academic integrity. This paper analyses various forms of academic misconducts, what motivates students to commit academic misconduct and presents some innovative methods to detect and prevent various forms of academic misconducts including contract cheating in online engineering education. Many of these strategies are equally applicable to any field and mode of higher education.
... The findings from these studies have been mixed. While some have found direct effects (Burgoon, et al., 2003;Burgoon, et al., 2008;Dunbar, et al., 2014;Zhou & Zhang, 2007), others have found evidence of a mediated relationship (George et al, 2008;George et al, 2014;Hancock, et al., 2010;Rockmann & Northcraft, 2008). ...
Article
Full-text available
Information Systems researchers can now more easily access vast amounts of data on the World Wide Web to answer both familiar and new questions with more rigor, precision, and timeliness. The main goal of this tutorial is to explain how Information Systems researchers can automatically “scrape” data from the web using the R programming language. This article provides a conceptual overview of the Web Scraping process. The tutorial discussion is about two R packages useful for Web Scraping: “rvest” and “xml2”. Simple examples of web scraping involving these two packages are provided. This tutorial concludes with an example of a complex web scraping task involving retrieving data from Bayt.com - a leading employment website in the Middle East.
... It may be clear that a website telling someone their cancer can be cured by using cannabis is likely to end up with social impairment due to over-stimulation of the prefrontal cortex with dopamine (Bishop, 2011b;Bishop, 2012b;Thomas et al., 2006). Equally a vulnerable or disabled person might, depending on their condition, be more likely to fall for other pseudo-activism, thinking they are helping others especially if the number of people that have already used or shared the artefact is high (Burgoon, Stoner, Bonito, & Dunbar, 2003). This is known as social proof, as one is willing to agree to something if one's friends agree with it. ...
Chapter
Full-text available
The growth in Internet use is not only placing pressure on service providers to maintain adequate bandwidth but also the people who run the Websites that operate through them. Called systems operators, or sysops, these people face a number of different obligations arising out of the use of their computer-mediated communication platforms. Most notable are contracts, which nearly all Websites have, and in the case of e-commerce sites in the European Union, there are contractual terms they must have. This chapter sets out to investigate how the role contract law can both help and hinder sysops and their users. Sysop powers are limited by sysop prerogative, which is everything they can do which has not been taken away by statute or given away by contract. The chapter finds that there are a number of special considerations for sysops in how they use contracts in order that they are not open to obligations through disabled or vulnerable users being abused by others.
Chapter
The nature of a pandemic often seizes its capacity to adjust its duration. Whilst biological pandemics might ebb in a duration of a few years, cognitive pandemics like misinfodemic that attack the cognitive possessions of the human brain fail to converge at points to ebb. Consequently, it is natural that the world has been constantly in the grasp of misinformation, evolving from the form of simple linguistic deception manifested as lies to heavy-lettered words of misinformation and disinformation. As the academic research community attempts to circumnavigate this phenomenon from interdisciplinary perspectives, it becomes crucial to take a step back and put this phenomenon under the microscope of temporal analysis. This allows us to visualize that similar to any biological threat, the cue to understanding misinformation lies in its simple evolution from a linguistic lie to the state it occupies today. Psycho linguistically speaking, the cure of misinfodemic lies in circumnavigating the devil that lies in the fine print published as deceptive lies a few years ago to misinformation disseminated on social media platforms today. Our study fulfils the task and expands to present how these fine linguistic cues of deception, even today, play a significant role in drafting feature engineering augmented NLP models that occupy positions of the state of the art in ML-based task of fake news classification and identification.
Article
The current study reduced the time lag between performing a diagnostic assessment and identifying a critical finding in CT and MRI exams through improving radiographers’ abilities to identify those critical findings. Radiographers’ diagnostic assessments in CT and MRI exams were used to develop a mobile training application with the aim to improve radiographers’ awareness of critical findings. The current research used data analytics to examine radiographers’ interpretation of imaging studies from a privately owned medical group in Israel. During the project, the radiographers’ ability to identify critical findings improved. Implementation of the mobile training program yielded positive results where the knowledge gap was reduced and time to identify critical cases was decreased. Specifically, this study showed that radiographers can be trained in ways that enhance their involvement with radiologists to provide high quality services and improve treatment Ultimately, this gives patients higher quality of care and safer treatment.
Chapter
This paper discusses the challenges faced by Emirati students enrolled in a private university in the United Arab Emirates. In particular, the study focuses on the development of their academic writing skills. It uses data collected from composition papers written in a semester through the online platform. The data was collected from undergraduate engineering and information technology students and it reports findings of a study examining the effectiveness of promoting Arabic speakers’ development of critical and analytical thinking skills using a content with a content-based approach. The data is discussed in relation to development of academic language in an English course through application of skills in an essay format. This study attempts to highlight the value of content-based approach through instructional practices and designing of online authentic assessment tasks. Two groups of engineering and computer science students participated in three different written assessments in the study. Six essay topics were designed considering students linguistic and societal factors. Data were collected via online assessments. The findings revealed significant improvements in the students’ academic writing scores and considerable reduction in plagiarism cases was also found. The quantitative and qualitative examination show positive impact of active participation in the online learning platform. Based on these findings, instructional suggestions and caveats are provided.
Article
Full-text available
This piece was the first in history to posit the notion of "truth-bias," which has now become foundational within the field of deception. It also posits what has come to be known as The McCornack-Parks Model of Deception Detection; namely, that as relational intimacy increases, detection confidence increases, truth-bias increases, and detection accuracy decreases.
Article
Interpersonal deception theory (IDT) frames deception as a communication activity and examines deception within interactive contexts. One key element of the theory is the role of suspicion in prompting behavior changes. An experiment testing several suspicion-related hypotheses paired participants (half friends, half strangers) for interviews during which interviewees (EEs) lied or told the truth and interviewers (ERs) were induced to be (moderately or highly) suspicious (or not). Results confirmed that suspicion and deceit were perceived when present, suspicion was manifested through nonverbal behaviors but with different behavioral patterns for moderately versus highly suspicious ERs, and suspicion affected sender behavior. Relational familiarity moderated some behaviors. Results are discussed in terms of mutual influence processes and the dynamic nature of communication in interpersonal deception.
Article
This essay extends the recent work of Levine, Park, and McCornack (1999) on the veracity effect in deception detection. The probabilistic nature of a receiver's accuracy in detecting deception is explained, and a receiver's detection of deception is analyzed in terms of set theory and conditional probability. Detection accuracy is defined as intersections of sets, and formulas are presented for truth accuracy, lie accuracy, and total accuracy in deception detection experiments. In each case, accuracy is shown to be a function of the relevant conditional probability and the truth-lie base rate. These formulas are applied to the Levine et al. results, and the implications for deception research are discussed.
Article
This essay draws together deception detection research and explains its relevance to decision‐making in today's social milieu. Major areas of research focus are described including the differential effects of observational conditions, the correspondence between perceived and actual signs of deception, training observers to detect deception, and communicating with suspected deceivers. Communicator relationships, observer characteristics and observer professions are considered in relation to accurate deception detection, as is overall confidence in the ability to detect deception. Implications are drawn for legal, organizational, sales, medical and relational settings.
Article
This experiment compared the deception detection skills of conversational participants and observers. Participants were expected to have a more pronounced truth‐bias than observers. They were also predicted to be less accurate detecting deception than observers because of the cognitive and communication requirements of conversational management. Each of fifty observers viewed two videotaped interactions between an interviewer (i.e., conversational participant) and two interviewees (i.e., sources) who either told the truth or deceived. Conversational participants attributed more truth to sources than did observers and they were less accurate detectors than observers. When not told about the deception manipulation, observers relied on more accurate behavioral cues than participants. When informed about deception, participants relied on inaccurate facial cues, whereas observers relied on inaccurate vocal cues to judge deceit.