Professional Documents
Culture Documents
net/publication/292989338
CITATIONS READS
12 193
4 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Jaume Masip on 26 March 2016.
Deception detection research has shown that police officers are less truth-biased and make their veracity
judgments with greater confidence than do nonofficers. Here we examined nonofficers, novice officers,
and experienced officers’ response bias, confidence, and generalized communicative suspicion. In
Experiment 1, novice officers aligned with nonofficers in terms of both generalized communicative
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
suspicion scores and confidence, with both these groups scoring lower than experienced officers.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Generalized communicative suspicion scores and veracity judgments were not significantly related for
either sample. However, novice officers aligned with experienced officers in terms of judgments: both
police groups were lie-biased, whereas nonofficers were truth-biased. These findings suggest that unlike
experienced officers, who have embraced the police culture to a greater degree, novice officers are not
dispositionally suspicious (generalized communicative suspicion); however, they are able to mirror the
prototypical police behavior (deception judgments) in police-related contexts. Experiment 2 supported
these notions.
“We don’t interrogate innocent people.” This was the reply judgments than truth judgments (see reviews by Alonso, Masip, &
given by Joseph Buckley, president of John E. Reid and Associ- Garrido, 2009; Garrido & Masip, 1999; Meissner & Kassin, 2002).
ates, a U.S. police training firm, when someone asked him whether In his major exploration of the deception detection literature, Vrij
his interrogation methods could cause innocent suspects to confess (2008) concluded that
(Kassin & Gudjonsson, 2004, p. 36). Buckley said this because in
order for a suspect to be questioned with Reid’s confrontational [c]ompared to laypersons, professionals seem to be somewhat better
in detecting lies and somewhat worse in detecting truths. This is, at
interrogation techniques, the suspect must first have been judged
least in part, because professional lie catchers do not appear to show
guilty on the grounds of either his or her reactions to a preinter- a truth-bias. In fact, many studies have demonstrated that police
rogation interview (called Behavior Analysis Interview; see Inbau, officers show a lie-bias, judging the majority of the fragments they
Reid, Buckley, & Jayne, 2013), or some kind of evidence. How- were exposed to as being deceptive. Moreover, in every study where
ever, Buckley’s statement also reflects the view of many seasoned both professional lie catchers and laypersons have participated, the
law enforcement officers—the view that suspects are typically professional lie catchers were more inclined to judge a fragment as
guilty and lie during police questioning. In this study, we exam- deceptive than laypersons. (p. 164)
ined the sources of this bias.
C. F. Bond and DePaulo’s (2006) meta-analysis did not find a
lie bias among “experts” (i.e., people whose occupations expose
Investigator Bias Effect them to lies, such as police officers, judges, auditors, job inter-
Unlike the general public, which displays a truth bias when viewers or psychiatrists); however, between-study comparisons
judging someone else’s veracity (e.g., C. F. Bond & DePaulo, revealed that such experts were less biased toward truthfulness
2006; Levine, Park, & McCornack, 1999), police officers have judgments than nonexperts.
often been found to show a lie bias—that is, they make more lie Not only are officers less inclined to make truth judgments than
are nonofficers; they also see themselves as skilled lie- and truth-
detectors and make their veracity judgments with great confidence
(e.g., DePaulo & Pfeiffer, 1986; Garrido & Masip, 1999; Garrido,
This article was published Online First February 4, 2016. Masip, & Herrero, 2004; Meissner & Kassin, 2002). Meissner and
Jaume Masip, Hernán Alonso, Carmen Herrero, and Eugenio Garrido, Kassin (2002; see also Masip, 2014b) coined the term investigator
Department of Social Psychology and Anthropology, University of Sala- bias effect to refer to the officers’ tendency to make lie judgments
manca. (often with high confidence).
Hernán Alonso is now at the Pedagogical Medical Institute San Juan de
Dios, Valladolid, Spain; Eugenio Garrido is now retired. Behavior, Personality, and Situations
The authors are grateful to Professor Sigi Sporer for his help with the
signal detection theory formulas. There are some crucial notions in this paper that should not be
Correspondence concerning this article should be addressed to Jaume confused. These notions are lie bias, generalized communicative
Masip, Faculty of Psychology, University of Salamanca, Avenida de la suspicion, and police-related situations. Lie bias refers to behavior,
Merced, 109-131, 37005 Salamanca, Spain. E-mail: jmasip@usal.es generalized communicative suspicion is a personality variable, and
169
170 MASIP, ALONSO, HERRERO, AND GARRIDO
police-related situations are a type of situation. We think it is findings. These findings suggest confirmation bias was in opera-
important to clarify these notions. tion: officers appeared to enter the veracity assessment situation
First, calling someone a liar (or, for that matter, a truth teller) is with the belief the sender was going to lie, and they appeared to
a behavior. Repeatedly calling other people liars is performing the search for supportive evidence. But why would officers assume the
same behavior repeatedly. Thus, the lie bias pertains to the behav- sender was going to lie?
ioral domain. Masip et al. (2005) reasoned that the police are often involved
Behavior may be caused by personality, by situations, or by in competitive interactions (Robinson, 1996). In cooperative inter-
their interaction (e.g., Bandura, 1986). Concerning personality, actions all participants collaborate to attain a common goal. Con-
some people may be more distrustful and suspicious than others, versely, in competitive interactions different participants have
which may lead them more often to judge others’ statements as different goals and everyone is pursuing his or her own benefit at
deceptive. Levine and McCornack (1991) posited that individuals the expense of the others. The nature of policing, as well as the
systematically differ in terms of their generalized communicative police goals of maintaining order and enforcing the law lead the
suspicion (GCS) —that is, their “predisposition towards believing police to be frequently involved in competitive interactions. Over
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
that the messages produced by others are deceptive” (p. 328). the years, police involvement in situations like these may generate
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Levine and McCornack conceptualized this construct as a rela- chronic distrust and suspiciousness.
tively enduring personality characteristic and created a scale to More generally, distrust or suspiciousness may also reflect po-
measure it. They expected that generalized communicative suspi- lice culture. Pioneering work by Skolnick (1966), Westley (1970),
cion scores (personality) would predict deception judgments (be- and Rubinstein (1973) depicted the police as an isolated, cohesive,
havior). distrustful, secretive, and authoritarian professional group exposed
Also, some situations or contexts may be more likely to prompt to multiple dangers in their interactions with citizens. More re-
deception judgments than others. Imagine that on Friday morning, cently, some researchers have argued that police culture may be
a police officer is interviewing a suspect at the police station. The more diverse than what early researchers thought, and that the
next day, the same officer, who is off duty, is visiting a distant central characteristics of the culture may have changed somewhat
town with his or her family and asks a pedestrian for directions. It (see, e.g., O’Neill, Marks, & Singh, 2007; Paoline, 2003, 2014;
is reasonable to assume that the first situation is more likely to Scaramella et al., 2010; Yüksel & Tosun, 2015). However, Wad-
prompt deception judgments from the officer than the second dington (1999) maintains that the core elements of the police
situation. Interestingly, situations may affect not only behaviors culture remain essentially the same across time and nations.
but also personality (e.g., Bandura, 1986). For instance, over the Among these elements are (a) position of authority; (b) glorifica-
long run, repeated exposure to situations arousing suspicion may tion of violence, action and excitement (celebrating the “real”
turn the individual into a suspicious person. police job); (c) crime-fighting image; (d) social isolation; (e)
The main point of Experiment 1 was to examine (a) whether protective solidarity with fellow officers; and (f) “dirty work” and
officers (particularly the most experienced ones) displayed a stron- knowledge of the miseries of the “underworld” (see Waddington,
ger lie bias (behavior) than non-officers; (b) whether experienced 1999, for the rationale behind these and other characteristics of the
officers, in part because of their exposure to suspicion-arousing police culture, as well as their psychological benefits for officers).
situations, were more dispositionally (personality) suspicious (in It is important to note that distrust, cynicism, and suspicion have
terms of their generalized communicative suspicion scores) than also been associated with police culture:
novice officers and nonofficers; and (c) whether officers’ gener-
alized communicative suspicion (personality) could explain their An integral part of the police personality is cynicism: the notion that
all people are motivated by evil and selfishness. Police cynicism
tendency to make lie judgments (behavior). In Experiment 2, we
develops among many officers through the nature of police work. . . . By
examined the influence of the situation on novice officers’ decep-
constantly dealing with crime and the most unsavory aspects of social
tion judgments. In the following sections, we describe the back- life . . . [police officers’] faith in humanity seems to diminish.
ground of this study and make specific predictions. (McNamara, 1999, pp. 8 –9)
social isolation of officers engender suspicion and distrust directed mental overconfidence would be expected to be hallmarks of a
toward the general public. An essential component of suspicion is veteran officer’s personality.
the belief that other people have something to hide and will not be
fully honest, which might generate a lie bias.
Rationale for the Present Study
It is important to note here that these effects are more likely to
be apparent among experienced officers than among police re- One weakness in Masip et al.’s (2005) study is that even though
cruits. The cumulative effect of frequent involvement in compet- they measured participants’ generalized communicative suspicion,
itive interactions happens in the long run. Also, it is reasonable to they did not examine the relation between generalized communi-
assume that senior officers embrace the police culture—which cative suspicion scores and veracity judgments. Masip et al. as-
involves suspicion—to a greater degree than police recruits. In- sumed that higher scores lead to more lie judgments and fewer
deed, the police culture is transmitted from one generation to the truth judgments, but is this assumption correct? If it is, would the
next through socialization within the police force, first via formal more seasoned officers (who scored higher on the scale) thus make
training at the academy, and later on by more experienced peers at more lie judgments than their less experienced peers? The current
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
the workplace (McNamara, 1999; Paoline, 2003; Sato, 2003; study addressed these issues.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
On the other hand, if novice officers make as many deception the Spanish National Police Force, inspectors are of a higher rank
This document is copyrighted by the American Psychological Association or one of its allied publishers.
judgments as their more seasoned colleagues, then this would than plain officers. Experienced plain officers can apply to be
reveal that (a) despite novice officers being like nonofficers in promoted to inspectors after a number of years of service. Lay
terms of generalized communicative suspicion (because they have people can enter the police after passing a competitive examination
not yet undergone any lengthy socialization process within the to become either plain officers or inspectors. The requirements to
police force), they (b) try to act (behavior) as a prototypical “real” become an inspector are much more stringent than those to become
officer, that is, showing harshness, distrust, suspicion, and there- a plain officer. Some officers with not enough experience to be
fore making many deception judgments. promoted to inspector may also take this competitive examination
to have access to an inspector position (for more detail, see Masip
et al., 2005, Note 3 on p. 1052).
Experiment 1
We asked nonofficers, novice officers, and experienced officers Materials
to complete the Generalized Communicative Suspicion Scale. We used Masip et al.’s (2005) Spanish adaptation of Levine and
Later on, and with no explicit connection to the prior task, all McCornack’s (1991) Generalized Communicative Suspicion
participants judged the veracity of a number of videotaped truthful Scale. The scale contains 11 items, and Cronbach’s alpha for the
and deceptive statements. We predicted the following: Spanish adaptation is .82 (Masip et al., 2005).
In the present study, we used four videotapes previously used by
Hypothesis 1: Experienced officers would score significantly Masip and his colleagues (Masip, Garrido, & Herrero, 2006, 2009,
higher than both novice officers and nonofficers on the Gen- 2010; Street & Masip, 2015). The videotapes contain the state-
eralized Communicative Suspicion Scale. No difference was ments of 24 individuals who lied and told the truth (in counter-
expected between novice officers and nonofficers. balanced order) answering three questions about either an at-
tempted or successful theft they had seen on videotape. The 48
Hypothesis 2: Despite the mixed evidence concerning the resulting statements were arranged in four different videotapes,
relationship between generalized communicative suspicion each containing six liars and six truth tellers. All relevant vari-
scores and veracity judgments, we tentatively predicted that ables, such as senders’ gender, question order, and so forth were
higher scores would be significantly related to a greater ten- counter balanced, and the same person never appeared in the same
dency to make deception judgments. tape lying and telling the truth. Specifically, the videos were labeled A1,
A2, B1, and B2. Videos A1 and B1 contained the statements of the
Hypothesis 3: Police officers would make more deception same 12 senders, but those who lied in A1 told the truth in B1, and
judgments (and fewer truth judgments) than nonofficers. It those who told the truth in A1 lied in B1. Videos A2 and B2
remained an open question whether novice officers would contained the statements of the other 12 senders and, again, those
differ or not from experienced officers in terms of truth/ who lied in one of the videos told the truth in the other video (see
deception judgments. Masip et al., 2006).
academy are normally divided into smaller groups on the basis of ANOVAs on the percentage of judgments of deceptiveness and on
the initial letter of their family name. A different video was shown B” (to test Hypothesis 3), on the percentage of correct judgments
to each group. Groups did not differ in terms of gender or age for (accuracy; to test Hypothesis 4), and on confidence (to test Hy-
either sample. pothesis 5). Using ANOVAs is in keeping with the analytical
The videos were shown in a classroom at the School of Psy- approach normally used in deception research.
chology (nonofficers) or at the police academy (officers). Empty
seats were left between one participant and the next, and the Results
researcher asked participants to keep silent and do the task indi-
vidually. The researcher handed out an answer booklet to partici- Generalized Communicative Suspicion
pants. The instructions, which very closely resembled those used
by Masip et al. (2006), were printed on the front page and the Cronbach’s alpha of the Generalized Communicative Suspicion
researcher read them aloud. The participants learned that they were Scale for the current study samples was .84 for nonofficers and .85
about to watch 12 people (senders), each making a statement about for both novice officers and experienced officers. The ANOVA
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
a videotape that they (senders) had seen, and that the statement of revealed that the three samples differed significantly in their
This document is copyrighted by the American Psychological Association or one of its allied publishers.
each particular sender could be either truthful or deceptive. The scores, F(2, 213) ⫽ 4.69, p ⫽ .010, 2 ⫽ .042, 90% CI [.006, .089]
participants’ task was to determine whether each sender was lying (see Table 1). Post hoc Sheffé tests revealed that nonofficers (M ⫽
or telling the truth. The experimenter told the participants that after 34.94, SD ⫽ 9.68, SEM ⫽ 1.14) and novice officers (M ⫽ 33.90,
each statement he would pause the tape so they could provide their SD ⫽ 9.70, SEM ⫽ 1.14) did not differ significantly, p ⫽ .827;
answers by ticking the “Lying” or “Telling the truth” box for the however, experienced officers’ scores (M ⫽ 38.81, SD ⫽ 10.95,
particular sender in the response booklet and by indicating their SEM ⫽ 1.29) were significantly higher than were those of novice
confidence in each judgment on a 1 (not at all confident) to 7 officers (p ⫽ .016) and marginally higher than those of nonofficers
(absolutely confident) scale. After both tasks had been completed, (p ⫽ .075). Overall, the pattern of results supported Hypothesis 1.
the participants were debriefed. Remarkably, the descriptive statistics are strikingly similar to those
reported by Masip et al. (2005; for nonofficers, M ⫽ 35.57, SD ⫽
Overview of Analyses 9.36; for novice officers, M ⫽ 33.13, SD ⫽ 10.11; for experienced
officers, M ⫽ 39.42, SD ⫽ 11.09).
We conducted a one-way analysis of variance (ANOVA) to In line with Hypothesis 2, we expected generalized communi-
examine whether nonofficers, novice officers, and experienced cative suspicion to explain the percentage of judgments of decep-
officers differed in their generalized communicative suspicion tiveness, false alarms, and B”. The results of the linear regression
scores (Hypothesis 1). To test Hypothesis 2, we calculated linear analyses are displayed in Table 2. It is apparent that the predictive
regression analyses to see how well generalized communicative value of the generalized communicative suspicion scores was
suspicion scores could explain (a) the raw percentage of judgments essentially null. Hypothesis 2 was not supported by the data.
of deceptiveness made by participants, (b) B”, and (c) false alarms.
B” is a signal detection theory (SDT) measure of bias. According
Judgments
to SDT (e.g., Green & Swets, 1966; Swets, Dawes, & Monahan,
2000), a detection task depends on two distinct processes, namely, The Sample ⫻ Veracity ANOVA run on the percentage of
discrimination (accuracy) and decision criterion (response bias). judgments of deceptiveness showed that deceptive statements were
Unlike raw percentages, SDT makes it possible to evaluate each of judged as deceptive (M ⫽ 57.64, SD ⫽ 18.31, SEM ⫽ 1.25) more
these processes independently. A measure of response bias that is often than truthful statements (M ⫽ 46.91, SD ⫽ 21.16, SEM ⫽
independent of accuracy is B”. We calculated B” with Donaldson’s 1.44), F(1, 213) ⫽ 38.88, p ⬍ .001, p2 ⫽ .154, 90% CI [.086,
(1992) formula. An unbiased response would yield B” ⫽ 0; neg- .227]. What is more interesting is that the sample main effect was
ative values would reflect a lie bias and positive values a truth bias. also significant, F(2, 213) ⫽ 8.98, p ⬍ .001, p2 ⫽ .078, 90% CI
False alarms were the percentage of truths misjudged as lies. [.026, .135]. Post hoc Sheffé tests revealed that both experienced
For the other hypotheses, because the independent variables officers (M ⫽ 56.60, SD ⫽ 14.19, SEM ⫽ 1.67) and novice
were categorical and the dependent measures were quantitative, we officers (M ⫽ 53.70, SD ⫽ 13.55, SEM ⫽ 1.60) made more
ran 3 (sample: nonofficers vs. novice officers vs. experienced deception judgments than did students (M ⫽ 46.53, SD ⫽ 16.18,
officers) ⫻ 2 (veracity: truthful vs. deceptive statement) mixed SEM ⫽ 1.91); p ⬍ .001 and p ⫽ .015, respectively; however, the
Table 1
Means (Standard Deviations) for the Dependent Variables in Study 1
Generalized Communication Suspicion scores 34.94ab (9.68) 33.90a (9.70) 38.81b (10.95)
Percentage of judgments of deceptiveness 46.53a (16.18) 53.70b (13.55) 56.60b (14.19)
B” .14a (.56) ⫺.14b (.50) ⫺.22b (.48)
Accuracy 54.86a (12.80) 57.18a (12.22) 54.05a (12.89)
Confidence 4.17a (0.83) 4.51a (0.75) 4.99b (0.86)
Note. Within each row, scores with different subscripts differ at the p ⬍ .05 level.
174 MASIP, ALONSO, HERRERO, AND GARRIDO
Table 2 F[1, 123] ⫽ 4.58, p ⫽ .033, p2 ⫽ .021, 90% CI [.001, .063])
Outcomes of the Linear Regression Analyses to Explain the displayed greater accuracy in judging lies than truths, nonofficers
Percentages of Judgments of Deceptiveness, B”, and False were somewhat more accurate in judging truths (M ⫽ 58.33, SD ⫽
Alarms Based on Generalized Communication Suspicion Scores 22.38, SEM ⫽ 2.64) than lies (M ⫽ 51.39, SD ⫽ 18.71, SEM ⫽
2.21), F(1, 123) ⫽ 4.03, p ⫽ .046, p2 ⫽ .019, 90% CI [.0002,
Measure F df p R R2 B .0586]. Additional analyses revealed that experienced officers’
Percentage of judgments of accuracy was significantly above chance in judging lies, t(71) ⫽
deceptiveness 5.15, p ⬍ .001, d ⫽ 0.61, 95% CI [0.25, 0.96], but not in judging
Nonofficers 0.59 1, 70 .447 .09 .01 0.00 truths, t(71) ⫽ ⫺1.05, p ⫽ .300, d ⫽ ⫺0.12, 95% CI [⫺0.48,
Novice officers 0.26 1, 70 .613 .06 .00 0.00 0.23]. The same pattern emerged among novice officers, t(71) ⫽
Experienced officers 0.00 1, 70 .969 .01 .00 0.00
Across all samples 0.48 1, 214 .490 .05 .00 0.00 5.34, p ⬍ .001, d ⫽ 0.63, 95% CI [0.28, 0.98], and t(71) ⫽ 1.55,
B” p ⫽ .129, d ⫽ 0.18, 95% CI [⫺0.17, 0.53], respectively. However,
Nonofficers 0.67 1, 70 .416 .10 .01 ⫺0.01 nonofficers’ accuracy was significantly above chance in judging
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Novice officers 0.24 1, 70 .626 .06 .00 0.00 truths, t(71) ⫽ 3.16, p ⫽ .002, d ⫽ 0.37, 95% CI [0.02, 0.73], not
Experienced officers 0.01 1, 70 .941 .01 .00 0.00
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Accuracy Discussion
To test Hypothesis 4, we ran a Sample ⫻ Veracity mixed All our predictions except Hypothesis 2 were supported. In
ANOVA on percentage correct judgments (accuracy). The sample terms of generalized communicative suspicion scores, experienced
main effect was not significant, F(2, 213) ⫽ 1.19, p ⫽ .308, p2 ⫽ officers were the most suspicious of all participants. The fact that
.011, 90% CI [.000, .039]—accuracy did not differ significantly novice officers scored lower than experienced officers on the
across samples (see Table 1). A significant main effect of veracity Generalized Communicative Suspicion Scale suggests that the
revealed that lies (M ⫽ 57.64, SD ⫽ 18.32, SEM ⫽ 1.25) were increased suspicion and distrust of the police is not a result of
better detected than truths (M ⫽ 53.09, SD ⫽ 21.16, SEM ⫽ 1.44), police selection practices. Instead, what appears to increase suspi-
F(1, 213) ⫽ 5.19, p ⫽ .024, p2 ⫽ .024, 90% CI [.002, .067]. cion is lengthy police experience and socialization. Over the years,
However, this was moderated by sample, F(2, 213) ⫽ 8.98, p ⬍ officers experience many competitive interactions that might in-
.001, p2 ⫽ .078, 90% CI [.026, .135]. Whereas experienced crease their skepticism concerning other people’s truthfulness.
officers (Mlies ⫽ 60.65, SD ⫽ 17.54, SEM ⫽ 2.07; Mtruths ⫽ 47.45, Similarly, progressive socialization into the police culture may
SD ⫽ 20.68, SEM ⫽ 2.44; F[1, 123] ⫽ 14.54, p ⬍ .001, p2 ⫽ .064, lead police officers to become more distrustful.
90% CI [.021, .123]) and novice officers (Mlies ⫽ 60.88, SD ⫽ However, this increased suspicion did not explain the police’s
17.28, SEM ⫽ 2.04; Mtruths ⫽ 53.47, SD ⫽ 19.16, SEM ⫽ 2.26; lie bias. Even though it may seem reasonable to posit that police
OFFICERS’ GENERALIZED COMMUNICATIVE SUSPICION AND VERACITY JUDGMENTS 175
culture and its influence on personality have an impact on behav- 203). Pressure toward uniformity is great, and new recruits
ior, causing a lie bias among experienced officers, the present “actively seek to assimilate themselves into the organization at
findings do not support this notion. First of all, linear regression every available opportunity” (Sato, 2003, p. 291). Also, the
analyses revealed that generalized communicative suspicion could novice officers get in touch and interact with senior officers in
not explain lie judgments (measured either as raw percentages of this police-only environment. All of these circumstances help to
deception judgments or with more sophisticated SDT measures) explain why the self-identity of new recruits is already that of
for either group. Thus, the notion that officers’ lie bias is caused by a police officer. Novice officers perceive themselves as officers
their increased communicative suspicion is unsupported (see and seek to learn to act as officers.
also Hurst & Oswald, 2012). Second, experienced officers (who But the situation matters. In police-related situations, novice
had higher generalized communicative suspicion scores than officers may feel they are performing a professional task (e.g.,
novice officers) did not display a stronger lie bias than did running an interrogation or assessing veracity) and will play
novice officers. In fact, novice officers made as many deception their roles, acting the way they think others expect them to act
judgments as the experienced officers. Consequently, both po- (e.g., being inquisitive and questioning the truthfulness of the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
lice groups were more accurate in judging lies than truths. In suspects’ answers). In most respects, their behavior is very
This document is copyrighted by the American Psychological Association or one of its allied publishers.
contrast, nonofficers displayed a truth bias and were better at much like that of their experienced peers. However, differences
identifying truths than lies. may emerge between novice and experienced officers in more
It is thus apparent that the lie bias is not caused by a police ordinary, nonpolice-related situations. There, police recruits
personality but by other factors. The nature of the task (deception may feel no need to display prototypical officers’ behaviors.
detection) is relevant for police work. Situational demands placed In the context of the current experiment, judging the veracity of
upon the police’s behavior could be a good candidate to explain speakers describing a theft is related to police work. That is why
the current pattern of results. novice officers made as many deception judgments as experienced
Novice officers were aligned with experienced officers in terms officers. However, the Generalized Communicative Suspicion
of deception judgments (percentage of judgments of deceptiveness Scale is not a scale to measure communicative suspicion in police-
and B”) and with nonofficers in terms of generalized communica- related situations. Instead, it measures personality (see Levine &
tive suspicion scores and confidence (see Table 1). This shows that McCornack, 1991). It measures generalized communicative sus-
even though the novice officers had not yet experienced the slow picion, which is dispositional, not situational. This is one possible
changes that happen over a police career (suspicion and confi- explanation why experienced officers scored significantly higher
dence), in this police-related situation they were nevertheless ca- on this scale than did novice officers; the latter did not feel the
pable of behaving like “real officers,” displaying skeptical and need to play the role of an officer when filling in a scale about
distrustful behaviors prototypically attributed to the police. In general—rather than situation-specific and police-related— com-
other words, novice officers are not (personality) “real officers” municative suspicion.
yet, but they may act (behavior) as such in a police-related situa- This line of reasoning suggests that if we had used a scale
tion (see Goffman, 1959). designed specifically to tap suspicion in police-related situations
The reason for this may be that novice officers feel they are only instead of using the Generalized Communicative Suspicion
officers. To enter the police force, applicants have to spend a Scale, then novice officers would have scored as high as experi-
substantial amount of time (ranging from months to years) prepar- enced officers. We decided to conduct a follow up study to test this
ing to pass a harsh competitive examination that many fail. Passing suggestion.
the exam may create a sense of pride and achievement and may
also establish a psychological separation in the minds of those who
Experiment 2
pass between themselves, who are no longer ordinary citizens but
carefully selected “officers,” and those who failed. Once at the We created a new scale we named Interrogation Suspicion Scale
academy, formal socialization begins. The police academy is a (see Appendix). We built the scale by rewriting each item in
total institution, that is, “a place of residence and work where a Levine and McCornack’s (1991) Generalized Communicative Sus-
large number of like-situated individuals, cut off from the wider picion Scale so that the item specifically referred to a police
society for an appreciable period of time, together lead an en- interrogation situation (see Appendix). Therefore, whereas the
closed, formally administered round of life” (Goffman, 1961, p. Generalized Communicative Suspicion Scale items do not refer to
xiii). In a total institution, all aspects of life are tightly scheduled, any specific situation, the Interrogation Suspicion Scale specifi-
happen in the same place, under the same authority, and in the cally focused on a police-related situation.
company of other similar individuals, and all the enforced activi-
ties are designed to fulfill the aims of the institution (Goffman,
Participants and Procedure
1961; see also Davies, 1989). As in other total institutions, in the
police academy first “the attitudes, worldviews, and behavior We collected new data from 74 novice officers (Mage ⫽ 27
patterns of the individual are stripped away” (McNamara, 1999, years, SD ⫽ 1.97, SEM ⫽ 0.23; 56 with less than one year of
p. 3) to later provide the novice officer with “a new set of experience and none with more than three years of experience),
attitudes, values, and beliefs” (McNamara, 1999, p. 4). Police and 107 experienced officers (Mage ⫽ 45 years, SD ⫽ 3.89,
recruits are forced to adopt “uniformity in appearance, attitude, SEM ⫽ 0.38; Mexperience ⫽ 23 years, SD ⫽ 3.95 years, SEM ⫽
and behavior, as well as strict adherence to rules and proce- 0.38; experience range: 14 to 31 years). They all were at the
dures” (Paoline, 2003, p. 203) and “group cohesion and loyalty national police academy of Spain studying to become or to be
are stressed in a paramilitary environment” (Paoline, 2003, p. promoted to police inspectors. They completed the Interroga-
176 MASIP, ALONSO, HERRERO, AND GARRIDO
tion Suspicion Scale (13 items) in class and were debriefed experienced and novice officers completing the Interrogation Sus-
afterward. picion Scale (which measures suspicion in police-related situa-
tions only) would not differ and would all have high scores. To test
Results for these predictions, we ran an ANOVA to compare the interro-
gation suspicion scores of the current experienced and novice
Internal Consistency of the Interrogation officers with the generalized communicative suspicion scores of
experienced and novice officers of Experiment 1. More specifi-
Suspicion Scale
cally, we conducted a 2 (Scale) ⫻ 2 (Officers’ Experience: novice
We ran principal component factor analyses setting the program vs. experienced officers) independent groups ANOVA on gener-
to obtain a single factor solution. In shaping or adapting the alized communicative suspicion scores and interrogation suspicion
Generalized Communicative Suspicion Scale, both Levine and scores. Because the total number of items was the same (11) both
McCornack (1991) and Masip et al. (2005) adopted the criterion of in Masip et al.’s (2005) adaptation of the Generalized Communi-
retaining only those items with factor loadings of .40 or greater. To cative Suspicion Scale and in the current Interrogation Suspicion
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
remain in line with previous research, and also because of statis- Scale, the sum scores could be compared directly—that is, there
This document is copyrighted by the American Psychological Association or one of its allied publishers.
tical considerations, we adopted the same criterion: The greater the was no need to average across items.
loading, the more the variable reflects the factor. According to
Tabachnick and Fidell (2013), only variables with loadings of .32 Findings
and above should be interpreted. We also calculated Cronbach’s
alphas and removed those items that limited the scale reliability. The main effect of scale was significant, F(1, 321) ⫽ 276.94,
All these items happened to have factor loadings smaller than .40; p ⬍ .001, p2 ⫽ .463, 90% CI [.400, .517], with interrogation
thus, our two item-retention criteria converged. The process of suspicion scores (M ⫽ 53.37, SD ⫽ 8.01, SEM ⫽ 0.59) being
item deletion is detailed in Table 3 where, for each variation of the higher than generalized communicative suspicion scores (M ⫽
scale, the factor loading of each item is displayed along with the 36.35, SD ⫽ 10.60, SEM ⫽ 0.88). This reveals that officers’
scale reliability (alpha) if each item were removed. We can see that distrust is higher in situations specifically related to their profes-
an initial factor analysis revealed that Items 9 and 10 had factor sional role, such as police interrogations. A small but significant
loadings lower than required and scale reliability would increase if main effect for experience revealed that novice officers (M ⫽
they were removed (see Table 3). After removing these two items, 43.99, SD ⫽ 13.36, SEM ⫽ 1.11) scored lower than experienced
we obtained a unidimensional scale with 11 items and Cronbach’s officers (M ⫽ 47.32, SD ⫽ 11.62, SEM ⫽ 0.87), F(1, 321) ⫽ 4.08,
alpha ⫽ .73 (see Table 3). Alpha was remarkably similar to the p ⫽ .044, p2 ⫽ .013, 90% CI [.0002, .040]. However, as expected,
figures reported by Levine and McCornack (1991) in validating this effect was qualified by a significant Scale ⫻ Experience
the original Generalized Communicative Suspicion Scale (.71 in interaction, F(1, 321) ⫽ 7.60, p ⫽ .006, p2 ⫽ .023, 90% CI [.004,
Study 1 and .75 in Study 3). .057] (see Figure 1). Scheffé tests showed that although experi-
enced officers (M ⫽ 38.81, SD ⫽ 10.95, SEM ⫽ 1.29) scored
higher than did novice officers (M ⫽ 33.90, SD ⫽ 9.07, SEM ⫽
Data Analysis
1.14) on the Generalized Communicative Suspicion Scale (p ⫽
We expected that, unlike experienced and novice officers com- .001), scores of experienced (M ⫽ 53.06, SD ⫽ 8.00, SEM ⫽ 0.77)
pleting the Generalized Communicative Suspicion Scale (which and novice officers (M ⫽ 53.81, SD ⫽ 8.04, SEM ⫽ 0.93) did not
measures generalized suspicion), who differed in their scores, differ on the Interrogation Suspicion Scale (p ⫽ .585).
Table 3
Factor and Scale Reliability Analyses and Item Deletion Process for the Interrogation Suspicion Scale
Prior research has found that police officers place more confi-
dence in their veracity judgments than do nonofficers (e.g., De-
Paulo & Pfeiffer, 1986; Garrido et al., 2004; Meissner & Kassin,
2002). To our knowledge, the question of whether experienced and
novice officers’ confidence scores differ has remained untapped.
We tackled this issue here and found that experienced officers are
more confident in their veracity judgments than are both novice
officers and nonofficers.
This pattern of results provides interesting insights into novice
and experienced officers’ behaviors and personalities. Generalized
communicative suspicion and confidence reflect the influence of
police work and police socialization on the officers’ personality.
However, deception judgments are context-dependent behaviors.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
the available empirical evidence support these contentions? In their This has a number of practical implications. No matter whether a
meta-analysis, C. F. Bond and DePaulo (2006) found that in the suspect is interviewed by an experienced or a novice officer, the
relatively few studies where motivation had been manipulated likelihood of this suspect’s statement being disbelieved is high re-
(within-study comparisons), stronger motivation to be believed led to gardless of actual veracity. This may set into motion a process that
higher accuracy rates. However, when comparing across the many may produce false confessions and eventually wrongful convictions.
studies with different motivation levels (between-study comparisons), According to Leo and Drizin (2010), the lie bias—which may lead
C. F. Bond and DePaulo found that higher motivation did not increase officers to erroneously decide that an innocent person is guilty—may
accuracy; rather, it reduced truth judgments, particularly when judges lead the police in some countries to use an accusatorial interrogation
had access to visible cues. This evidence suggests that the lie bias to obtain a confession. Unlike information-gathering interview ap-
would be stronger in real life (high stakes) compared with the present proaches, accusatorial approaches increase the risk of false confes-
experiment (low stakes). It is uncertain whether the increased stakes sions (Meissner et al., 2014). Therefore, the innocent suspect may end
and sender’s motivation would affect the judgments of officers and up confessing. Police detectives can also influence the post admission
nonofficers differently. narrative (Leo & Drizin, 2010). Finally, confessions have a huge
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
In a series of studies with real-life, high stake truths and lies told influence over conviction rates (Kassin & Neumann, 1997). Thus, the
This document is copyrighted by the American Psychological Association or one of its allied publishers.
by offenders, Mann and Vrij (e.g., Vrij, Mann, Robbins, & Rob- initial misclassification of an innocent suspect caused by police offi-
inson, 2006; see Alonso et al., 2009, for a comprehensive review) cers’ lie bias may have negative far-reaching consequences.
did not find any lie bias among officers (no nonofficers partici- To solve this problem, the police should use systematic lie-
pated in these studies). More recently, both Carlucci et al. (2013) detection tools that work better than their intuitions and gut feelings.
and Wright Whelan et al. (2015) found a stronger lie bias among If the officers’ veracity judgments were the result of using structured,
officers than among nonofficers judging high stakes, real-life, systematic, and valid procedures, then they would be less biased.
crime-related truths and lies. Therefore, we do not think the Recently, psychologists started to develop and test a number of
officers’ lie bias we found is limited to low stake, laboratory interviewing approaches to detect deception by creating or increasing
situations. We do agree, however, that more ecologically valid observable differences between truth tellers and liars (see Vrij &
studies are necessary in the deception detection domain and ap- Granhag, 2012, for a review). For example, because often lying is
plaud Carlucci et al.’s (2013) and Wright Whelan et al.’s (2015)
cognitively more demanding than truth telling (Gombos, 2006;
efforts in this direction.
Hauch, Blandón-Gitlin, Masip, & Sporer, 2015; Walczyk, Igou,
Another limitation of this study is that the Interrogation Suspicion
Dixon, & Tcholakian, 2013), further increasing the interviewees’
Scale (Experiment 2) was administered at a later time than the
cognitive demand will be more detrimental to liars than truth tellers,
Generalized Communicative Suspicion Scale. This time delay could
such that behavioral signs of cognitive load will be displayed by liars
presumably have influenced the results such that the lack of a signif-
(Vrij et al., 2008). One way of increasing cognitive load during the
icant difference between novice and experienced officers in Experi-
interview is asking interviewees to describe the target event in the
ment 2 could be a result of the data collection timing rather than a
reverse order rather than in the normal chronological order (Vrij et al.,
result of using the Interrogation Suspicion Scale instead of the Gen-
2008). Walczyk et al. (2005, 2012) developed a promising interview
eralized Communicative Suspicion Scale. We believe this explanation
procedure to detect deception based on a cognitive model of decep-
is extremely unlikely to account for our results. First, there is no
theoretical reason why a few months’ time should obliterate the tion. Also, depleting the cognitive resources of suspects before the
difference between the scores of novice and experienced officers. interview negatively affects lying behavior and improves lie detection
Second, the generalized communicative suspicion difference between (Blandon-Gitlin, Echon, & Pineda, 2013). Withholding the incrimi-
groups made sense theoretically and replicated prior findings by Hurst nating information that the interviewer possesses until the end of the
and Oswald (2012) and Masip et al. (2005); this shows that the effect interview may result in liars providing statements that contradict the
does not depend on the moment of data collection. Third, the pattern available evidence (Hartwig, Granhag, & Luke, 2014). Finally, asking
of results found in Experiment 2 was in line with our hypothesis. unexpected questions to pairs of suspects when interviewing them
separately results in liars giving different answers to the same ques-
tions; conversely, truth tellers, who simply describe the real event
Conclusions and Practical Implications honestly, provide the same answers (Vrij et al., 2009).
The current findings suggest that, over time, experienced officers If experienced officers were given specific guidelines based on
become more dispositionally suspicious and confident, presumably these evidence-based lie-detection approaches and could be con-
because of their exposure to competitive interactions, their socializa- vinced to closely follow them, and if novice officers internalized the
tion and experience within the police force, and their embracement of notion that “the right thing to do” when questioning someone’s
police culture. Thus, socialization and exposure to certain situations credibility in police-related situations is to use systematic scientific
may cause changes in the personality of officers. However, in this approaches (rather than being skeptical or distrustful), then the police
study generalized communicative suspicion (personality) was unre- lie bias would presumably decrease and detection accuracy would
lated with lie bias (behavior). Instead, situational factors appeared to increase. The influence of veteran officers’ personality on their be-
have a greater impact. Indeed, novice officers, who were not dispo- havior would be circumvented because veteran officers judging ve-
sitionally suspicious, made as many deception judgments as experi- racity could follow a set of available systematic guidelines instead of
enced officers. In police-related situations, novice officers were as having to resort to their overall impressions or gut feelings. Concern-
suspicious as experienced officers. These findings suggest that novice ing novice officers, their tendency to act “as an officer” in police-
officers have not yet acquired the inner characteristics of their expe- related situations would be beneficial if this implied using systematic
rienced peers but are able to behave like them. procedures rather than behaving in a distrustful way.
OFFICERS’ GENERALIZED COMMUNICATIVE SUSPICION AND VERACITY JUDGMENTS 179
Unfortunately, the limits of these new interviewing approaches Fenn, E., Blandón-Gitlin, I., Pezdek, K., & Yoo, A. (2015, March). “I’m
to detect deception and their vulnerability to countermeasures are innocent, but look guilty”: Ethnicity matters during a cognitive load
still not well known. Therefore, at present, their widespread use approach interview. Paper presented at the meeting of the American
may be premature (see Blandón-Gitlin, Fenn, Masip, & Yoo, Psychology-Law Society, San Diego, CA.
2014). However, they are promising tools for the future. It is our Garrido, E., & Masip, J. (1999). How good are police officers at spotting
hope that they will contribute to eradicate the police’s lie bias and lies? Forensic Update, 58, 14 –21.
will help improve the justice system. Garrido, E., Masip, J., & Herrero, C. (2004). Police officers’ credibility
judgments: Accuracy and estimated ability. International Journal of Psy-
chology, 39, 254 –275. http://dx.doi.org/10.1080/00207590344000411
References Goffman, E. (1959). The presentation of self in everyday life. New York,
Aamodt, M., & Custer, H. (2006). Who can best catch a liar? A meta- NY: Anchor Books.
analysis of individual differences in detecting deception. Forensic Ex- Goffman, E. (1961). Asylums: Essays on the social situation of mental
aminer, 16, 6 –11. patients and other inmates. New York, NY: Anchor Books.
Alonso, H., Masip, J., & Garrido, E. (2009). La capacidad de los policías Gombos, V. A. (2006). The cognition of deception: The role of executive pro-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
para detectar mentiras [Police officers’ ability to detect lies]. Revista de cesses in producing lies. Genetic, Social, and General Psychology Monographs,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Masip, J. (2014b). Investigator bias. In T. Levine (Ed.), Encyclopedia of Scaramella, G. L., Cox, S. M., & McCamey, W. P. (2010). Introduction to
deception (Vol. 2, pp. 538 –540). Thousand Oaks, CA: Sage. http://dx policing. Thousand Oaks, CA: Sage.
.doi.org/10.4135/9781483306902.n203 Skolnick, J. H. (1966). Justice without trial: Law enforcement in demo-
Masip, J., Alonso, H., Garrido, E., & Antón, C. (2005). Generalized cratic society. New York, NY: Wiley.
communicative suspicion (GCS) among police officers: Accounting for Skolnick, J. H. (1994). Justice without trial: Law enforcement in demo-
the investigator bias effect. Journal of Applied Social Psychology, 35, cratic society (3rd ed.). New York, NY: Wiley.
1046 –1066. http://dx.doi.org/10.1111/j.1559-1816.2005.tb02159.x Street, C. N. H., & Masip, J. (2015). The source of the truth bias: Heuristic
Masip, J., Garrido, E., & Herrero, C. (2006). Observers’ decision moment processing? Scandinavian Journal of Psychology, 56, 254 –263. http://
in deception detection experiments: Its impact on judgment, accuracy, dx.doi.org/10.1111/sjop.12204
and confidence. International Journal of Psychology, 41, 304 –319. Swets, J. A., Dawes, R. M., & Monahan, J. (2000). Psychological science
http://dx.doi.org/10.1080/00207590500343612 can improve diagnostic decisions. Psychological Science in the Public
Masip, J., Garrido, E., & Herrero, C. (2009). Heuristic versus systematic Interest, 1, 1–26. http://dx.doi.org/10.1111/1529-1006.001
processing of information in detecting deception: Questioning the truth
Tabachnick, B. G., & Fidell, L. S. (2013). Using multivariate statistics (6th
bias. Psychological Reports, 105, 11–36. http://dx.doi.org/10.2466/PR0
ed.). Boston, MA: Pearson.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
.105.1.11-36
Van Maanen, J. E. (1975). Breaking in: A consideration of organizational
Masip, J., Garrido, E., & Herrero, C. (2010). Regression toward the mean
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Appendix
1. Everyone lies, the person who says that they don’t is the biggest 1. Every suspect lies during an interrogation, the suspect who says
liar of all that they don’t is the biggest liar of allⴱ
2. I often feel as if people aren’t being completely truthful with meⴱ 2. When in my daily work I request information that may somehow
be related to a crime I feel that that people aren’t being completely
truthful with meⴱ
3. Most people only tell you what they think you want to hearⴱ —
4. When I am in a conversation with someone, I frequently wonder 3. When I am interrogating someone, I frequently wonder whether
whether they are really telling me the truthⴱ they are really telling me the truthⴱ
5. People rarely tell you what they’re really thinkingⴱ 4. People rarely tell you what they’re really thinking when they have
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.