ArticlePDF Available

The Prevalence of Lying in America: Three Studies of Self‐Reported Lies


Abstract and Figures

This study addresses the frequency and the distribution of reported lying in the adult population. A national survey asked 1,000 U.S. adults to report the number of lies told in a 24-hour period. Sixty percent of subjects report telling no lies at all, and almost half of all lies are told by only 5% of subjects; thus, prevalence varies widely and most reported lies are told by a few prolific liars. The pattern is replicated in a reanalysis of previously published research and with a student sample. Substantial individual differences in lying behavior have implications for the generality of truth-lie base rates in deception detection experiments. Explanations concerning the nature of lying and methods for detecting lies need to account for this variation.
Content may be subject to copyright.
Human Communication Research ISSN 0360-3989
The Prevalence of Lying in America: Three
Studies of Self-Reported Lies
Kim B. Serota, Timothy R. Levine, & Franklin J. Boster
Department of Communication, Michigan State University, East Lansing, MI 48823, USA
This study addresses the frequency and the distribution of reported lying in the adult
population. A national survey asked 1,000 U.S. adults to report the number of lies told in
a 24-hour period. Sixty percent of subjects report telling no lies at all, and almost half of
all lies are told by only 5% of subjects; thus, prevalence varies widely and most reported
lies are told by a few prolific liars. The pattern is replicated in a reanalysis of previously
published research and with a student sample. Substantial individual differences in lying
behavior have implications for the generality of truthlie base rates in deception detection
experiments. Explanations concerning the nature of lying and methods for detecting lies
need to account for this variation.
Humans are ambivalent about deception. On one hand, virtually all human cultures
have some prohibition against lying. On the other hand, the ability to deceive well
may be essential for polite interaction and, at times, self-preservation. Considerable
research exists on the topic of deception, yet surprisingly little is known about
the base prevalence of deception. Instead, much of this research has relied on
untested assumptions and anecdotal evidence or on a few studies with small and
nonrepresentative samples.
The dearth of deception prevalence research is a symptom of a broader systemic
concern regarding research in the social sciences. Asch (1952, reprinted 1987)
observed that ‘‘before we inquire into origins and functional relations, it is necessary
to know the thing we are trying to explain.’’ Influenced by Asch, Rozin (2001)
argued that social scientific research often emphasizes experimental studies and
formal hypothesis testing to the exclusion of more basic descriptive work. In line
with Rozin’s critique, more than 30 years of experimental detection research has
proceeded without much attention to the basic nature of the phenomena itself. We
believe that inquiry into deception and related behaviors associated with deception
detection requires basic descriptive research examining the extent and distribution of
deceptive communication in the population. In the extensive literature on deception,
Corresponding author: Timothy R. Levine; e-mail:
2Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
the question of prevalence remains without a clear, well-documented answer. Thus,
our research investigates reports of how often people lie.
In order to study the prevalence of lying, it is necessary to consider what
constitutes a lie. Simply and broadly put, lying occurs when a communicator seeks
knowingly and intentionally to mislead others. Ford, King, and Hollender (1988)
suggest the ‘‘consciousness of falsity’’ to distinguish ‘‘normal’’ lies from pathological
ones. Thus, it is not sufficient that something is false for it to be a lie; it is the intent
that distinguishes the lie. As Bok (1999) observes: ‘‘The moral question of whether
you are lying or not is not settled by establishing the truth or falsity of what you say.
In order to settle this question, we must know whether you intend your statement to
mislead’’ (p. 6).
Bok (1999) argues for the principle of veracity that involves a moral asymmetry
between honesty and lies. Lying requires justification, whereas truth telling does not.
Given prohibitions against deceit, people may try to avoid situations in which there
is pressure to lie. Research finds that this principle of veracity guides everyday com-
munication and people consider using deception only when the truth is problematic
(Levine, Kim, & Hamel, 2009). But this tells us about the situations in which people
lie and not how often people lie.
Despite this moral asymmetry, most deception research has presumed the ubiquity
of lying and moved past the question of frequency to focus on the behavioral correlates
of lying or lie detection. The frequency question remains mostly unanswered. A
notable exception, and the best and most cited prevalence research, however, is
the 1996 diary study of lying in everyday life by DePaulo, Kashy, Kirkendol, Wyer,
and Epstein (1996). Using two small samples, students and recruited members of
the local community, DePaulo et al. reported the mean number of lies per day as
1.96 (SD =1.63, N=77) for the students and 0.97 (SD =0.98, N=70) for the
subsequent nonstudent sample. Importantly, the aim of the second sample was to
replicate findings regarding the nature and reasons for lying with a different but not
necessarily representative sample of the population. Nonetheless, a brief synopsis of
this study in Psychology Today (‘‘The Real Truth About Lying,’’ 1996) reported that
DePaulo conducted research to answer the question ‘‘how often do people lie ... ?’’
and in many subsequent research reports the finding that people tell one to two lies
per day has been reified.
More recently, two smaller and lesser-known studies have sought to replicate and
extend elements of the DePaulo et al. (1996) diary study. Hancock, Thom-Santelli,
and Ritchie (2004) examined differences between reports of face-to-face lies and
reports of lying through computer-mediated communication. Results from a student
sample yielded an average of 1.58 lies per day (SD =1.02, N=28) and a significant
difference for the rates of lying (lies as a proportion of interactions) between face-to-
face, telephone, instant message, and e-mail interactions; the highest rates occurred
during phone conversations and the lowest rates with e-mail. George and Robb
(2008) replaced the pencil-and-paper diary with a personal digital assistant (PDA).
They report fewer lies per day; M=0.59 (SD =0.37, N=25) with the 10-minute
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 3
Prevalence of Lying in America K. B. Serota et al.
definition of interaction used by DePaulo et al. and M=0.90 (SD =0.54, N=25)
in a second study shortening the interaction definition to 5 minutes (increasing the
number of reporting opportunities). Thus, the current literature provides estimates
ranging from 0.59 to 1.96 lies per day. Variation in estimates from study to study is
expected due to small sample sizes and large standard deviations.
Other broad estimates of prevarication cited by deception researchers have come
from nonacademic sources. In a poll conducted for the book The Day America Told
the Truth (Patterson & Kim, 1991), 90% of the subjects admitted being deceitful about
a list of subjects, the most common being true feelings, income, accomplishments,
sex life, and age. In a Reader’s Digest poll (Kalish, 2004) of 2,861 of the magazine’s
readers, 93% reported one or more kinds of dishonesty at work or school, 93%
reported one or more dishonest acts in the market place, and 96% reported lying to
or committing other dishonest acts toward family and friends.
Some studies report on various facets of prevalence but provide limited insight
into the overall extent of lying because they deal with specific situations such as
lying by job applicants, students lying to parents, or lying about spousal infidelity.
Levashina and Campion (2007) found that 90% of undergraduate job candidates
used some form of deception during job interviews; however, the distinction between
impression management and outright lies is often blurred and their report notes that
behaviors that are semantically close to lies are difficult to confirm. They estimate
actual lies occur in the wide range of 2875% of job interviews. Jensen, Arnett,
Feldman, and Cauffman (2004) examined lying to parents by adolescents and young
adults, and quantified the extent of lying by topic over the course of a year. This
study found that 82% of all students reported lying to their parents on at least one
of six topical issues (money, alcohol/drugs, friends, dating, parties, and sex) with
the mean incidence of lying ranging from 0.6 to 2.4 lies depending on the issue.
Much of the research that seeks to quantify lying behavior is concentrated in the area
of relational communication. Cochran and Mays (1990) studied dating dishonesty
among college students and found that 60% of women claimed to have been lied to
in order to obtain sex, whereas 34% of the men in the study admitted lying to obtain
sex. Knox, Schacht, Holt, and Turner (1993) found that 92% of students (when given
the opportunity to report anonymously) admitted to lying to a current or potential
sexual partner.
It is not difficult to understand why many scholars believe lying is a frequent event.
Life experiences and anecdotal evidence encourage acceptance of the proposition. A
typical research report discussion statement illustrates this view: ‘‘Lying is ubiquitous
and comes in many forms, from cherished beliefs about Santa Claus to the self-
deception commonly encountered in the treatment situation’’ (Tosone, 2006).
General acceptance of the ubiquity assumption has implications for studies on
lying and deception detection. If everyone lies and lying is an everyday occurrence
for most people, this would suggest that individual differences should not have much
influence on the identification of lying behaviors. If this is the case, it should be
possible to understand the nature of lying and deceptive behavior and find ways to
4Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
detect it by studying anyone telling lies. For example, this presumption is recurrent in
studies that look for regularities in nonverbal cues, microexpressions, and the leakage
of emotions. Individual differences are typically considered of less relevance than
situational considerations such as whether the lie is a minor everyday lie or if the lie is a
consequential, high-stakes lie. If, on the other hand, base rates for lying (the frequency
with which one lies or the ratio of truths to lies) vary across groups of individuals
or if the ubiquitous average masks variation that is not normally distributed in the
population, researchers looking into the nature of the phenomenon need to take into
account this variation. Research designs and sample selection procedures ought to
control for this variation, or they should examine the nature of the variation itself.
Consistent with this second possibility, recent meta-analysis suggests substantial
individual differences in people’s ability to lie convincingly (Bond & DePaulo, 2008).
Study 1
Examination of the literature reveals few attempts to document the extent to which
everyday lies occur. The few studies that offer behavioral rates have performed so as
an adjunct to the main objectives of the research and the rates obtained are restricted
to the specific conditions of the studies. This situation is exemplified by the DePaulo
et al. (1996) studies. Although lie frequency is among the interests that prompted
their research, most of the design and analysis is devoted to the topics of what people
lie about, to whom they lie, and what motivates them to lie. DePaulo et al. noted that
their observations of lying frequency are based on students and an adult sample that
was chosen not for representativeness but, instead, to provide a dissimilar sample
in order to determine whether or not the results from the student sample could
be replicated. Still, DePaulo et al. reported: ‘‘Participants in the community study,
on the average, told a lie everyday; participants in the college student study told
two.’’ So despite being based on data that were (correctly) noted by its authors as
lacking generalizability, this research report has become the standard reference for
the prevalence of everyday lies in the deception literature. The goal of Study 1 is to
test this claim by obtaining data from a large cross-section of the adult population.
Participants and design
In order to examine the proposition that most people tell one to two lies per day with
projectable data, an Internet survey of 1,000 American adults (18 years of age or older)
was conducted using the Synovate eNation omnibus panel.1The omnibus panel is a
commercial survey research tool used for daily, multiclient studies and approximates
a nationally representative sample with some limitations. Panelists are recruited
into a pool of more than 1 million subject households using banner advertisements,
mailing lists, and related procedures to promote participation; subjects must formally
opt-in to confirm awareness that they are participating in research and consent to
participation. Each survey day, a new sample (N=5,500) is drawn from the pool.
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 5
Prevalence of Lying in America K. B. Serota et al.
Subjects are randomly selected within strata defined by population characteristics. The
response rate is typically 19 20%, accounting for both nonresponse and incomplete
surveys. Responses exceeding the 1,000 daily quota are deleted using systematic
(random start, nth selection) sampling. The panel is matched on age, gender, income,
and region to the U.S. Census Bureau’s monthly Current Population Survey (CPS;
U.S. Census Bureau, 2008). Results are poststratification weighted (Kish, 1965) to
these CPS criteria. In addition, Synovate uses weighting in order to adjust partially
for underrepresentation of Hispanics and ethnic minorities in the sample. Subjects
were included in a prize drawing as the incentive to participate. Table 1 provides
the unweighted and weighted sample demographics and compares them to recent
U.S. Census data and to eNation weighting targets, which are based on U.S. Census
This study used a nonexperimental survey design in order to obtain descriptive
measures for the incidence of lying in the population. Results from this survey are
compared with the popular standard established by the DePaulo et al. (1996) studies.
Procedure and measures
Subjects received an invitation e-mail asking them to participate in an omnibus
survey on the date of the study. The invitation was directed to a specific member of
the household identified by age and gender. The invitation instructed that individual
to click on a link to the survey website. When subjects accessed the site, they were
provided with instructions, asked questions confirming participant identification,
asked the omnibus survey questions for several unrelated topics, and asked a series of
demographic questions. On the day of the lying study, subjects were asked about four
topics (in order of presentation): packaged meals, cat litter products, lying behavior,
and water softeners.
The DePaulo et al. (1996) diary study (and subsequent diary studies) used subject
training to make the topic less sensitive and provide the subjects with a common
definition of lying. Training of survey respondents was not possible, so to encourage
accurate reporting, the self-report lying question was preceded by a definition of
lying that incorporated the elements of foreknowledge and intent described at the
beginning of this article. A brief description of types of lies was also included. Both
were presented in a nonpejorative manner:
We are interested in truth and lies in people’s everyday communication. Most
people think a lie occurs any time you intentionally try to mislead someone.
Some lies are big while others are small; some are completely false statements
and others are truths with a few essential details made up or left out. Some lies
are obvious, and some are very subtle. Some lies are told for a good reason. Some
lies are selfish; other lies protect others. We are interested in all these different
types of lies. To help us understand lying, we are asking many people to tell us
how often they lie.
6Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
Table 1 Comparison of the Unweighted and Weighted Demographics of the Study Sample
to U.S. Census Data and eNation Demographic Targets
CPSaeNation Unweighted Weighted Weighted
(January 2007) Target Results Results ±CPS
Gender (%)
Male 49.1 48.448.448.30.8
Female 50.9 51.651.651.70.8
Region (%)
Northeast 18.2 18.718.418.80.5
MidWest 22.1 22.221.622.30.2
South 36.4 36.235.936.00.4
West 23.3 22.924.123.00.3
Age (%)
Under 25 12.8 12.75.312.50.3
25–34 17.9 17.917.518.00.1
35–44 19.2 19.219.719.30.1
45–54 19.5 19.522.619.60.1
55–64 14.5 14.523.314.50.0
65+16.2 16.211.616.20.0
Income (%)
Under $25,000 18.8 18.714.318.50.3
$25,000–$49,999 24.4 24.524.224.60.2
$50,000–$74,999 19.7 19.622.319.70.0
Race (%)b
White 80.1 81.591.083.02.8
Refused/NA 1.5
Hispanic (%)b
No 84.9 92.595.492.47.5
Note: CPS =Current Population Survey.
aSynovate weights are based on most current monthly CPS data at the time of study (March
bRace and Hispanic eNation targets are adjustments toward the U.S. Census data, not
Census-based targets.
Subjects were then asked, using an open-ended format, how many times they had
lied in the past 24 hours. They responded separately for lies to family members,
friends, business contacts, acquaintances (‘‘people you do not know but might see
occasionally’’), and total strangers; for each type of receiver, they were asked about
lies in both face-to-face and mediated situations. Response categories were used as
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 7
Prevalence of Lying in America K. B. Serota et al.
Figure 1 Screen shot of the Internet survey page giving a definition and description of lying.
a mnemonic device and to proved additional detail for the analysis. The results of
the 10 receiver-mode combinations were aggregated. Specifically, the question was
Think about where you were and what you were doing during the past 24 hours,
from this time yesterday until right now. Listed below are the kinds of people
you might have lied to and how you might have talked to them, either
face-to-face or some other way such as in writing or by phone or over the
Internet. In each of the boxes below, please write in the number of times you
have lied in this type of situation. If you have not told any lies of a particular
type, write in ‘‘0.’’ In the past 24 hours, how many times have you lied?
The subjects were presented with a response grid showing the five types of people and
two modes of communication. Subjects were instructed to enter a number in each
box. The Internet questionnaire required a response in each of the 10 boxes before
allowing the subject to continue to the next Web page. Figure 1 presents a screen shot
of the lying description used in the Web survey; Figure 2 is a screen shot of the survey
question. Based on the 5 ×2 individual categories of responses, the row, column,
and grand total frequencies (of lies per day) were aggregated for each subject.
The results of this national study are consistent with the oft-cited observation that
on average Americans tell one to two lies per day (M=1.65 lies per day, SD =4.45,
Mdn =0, Mode =0, N=998, Max =53 lies, 95% CI =1.371.93).2But the most
intriguing finding is the distribution of responses, not the mean. As Figure 3 illustrates
the majority of people report telling no lies during the past 24 hours and most of the
reported lies are told by few people. The 40.1% who reported lying told a total of
8Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
Figure 2 Screen shot of the Internet survey page with the question and response grid.
1.6 1.2 1.0 1.5
0.7 1.1
0.3 0.3 0.1 0.4 0.1 0.1 0.0 0.0 0.1
(B) Number of lies in the past 24 hours
% (of total) Reporting
(A) Occurrence
Lies vs. No Lies
No Lies
Figure 3 (A) The majority of Americans reported they did not lie in the past 24 hours
(59.9%). (B) Percentage distribution by number of lies told; 32.2% told one to five lies and
7.9% reported telling six or more lies.
1,646 lies (M=4.11, SD =6.26, n=400). Of these, 22.7% of all reported lies were
told by 1% of the national sample. Results indicate that one-half of all reported lies
are told by just 5.3% of American adults (M=15.61, SD =11.22, n=53).
Figure 4 indicates that, among those who reported lying, the proportion of
people who report a particular number of lies per day decreases as a function of
the number of lies. Moreover, observation of this curve suggests that the decrease
is a standard power function. Fitting a power function curve to these data produces
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 9
Prevalence of Lying in America K. B. Serota et al.
Prevalence of Lies - Summary Curves
0 5 10 15 20 25 30 35 40 45
Number of Lies
y = 152.225 *+ x^-1.209 Total
y = 135.584 * x^-1.301 F2F
y = 135.207 * x^-1.782 Mediated
Figure 4 Frequencies of total, face-to-face, and mediated lies are represented by similar
standard power functions.
the equation y=152.225 ×x1.209 where xis the number of lies reported per day
and ythe frequency of people reporting a given number of lies. The function’s
sizeable intercept indicates that the majority of respondents who report lying do
so in moderation, and the steep negative slope indicates the substantial decrease
in frequency as the number of daily lies increases. Figure 4 also indicates that this
pattern holds regardless of mode of communication. Although the slopes for both
face-to-face lies and mediated lies are even steeper than that of the aggregated data,
both retain the power law character of the total data set. Regardless of mode, most
people report telling no lies and as the curve-fitting for number of lies reinforces,
among those who report lying, most tell very few lies; but in each case, there are a
few subjects who account for a large proportion of the lies being told.
Similarly, Figure 5 (among those telling face-to-face lies) and Figure 6 (among
those telling mediated lies) indicate that lying behavior replicates the fractal character
of power functions observed in other disciplines such as biological systems (Brown
et al., 2002) and market segmentation (Anderson, 2008). Although each group of lies
told to family members or friends or other types of message receivers represents only
a small portion of the total lies, within each of the 10 mode-receiver combinations,
the power function pattern is repeated. Most of the variation is among the intercepts
and reflects that more lies are typically told to family members or friends than to
acquaintances or total strangers.
These data do not include the number of interactions by type; therefore, it is
possible that this variation is as much due to the number of opportunities for lying as it
10 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
Prevalence of Face-to-Face Lies - by Receiver Type
0 5 10 15 20 25 30 35 40 45
Number of Lies
y = 75.358 * x^-1.586 Family
y = 93.526 * x^-1.909 Friends
y = 74.270 * x^-1.748 Business
y = 52.481 * x^-2.090 Acquainted
y = 39.753 * x^-1.855 Strangers
Figure 5 Standard power function curves fit the frequencies of face-to-face lies plotted by
number of lies for all receiver types.
is that people are more likely to lie to others with whom they are familiar. Establishing
the proportion of interactions involving lies may be more difficult than establishing
the rate of lying for a prescribed time period. All of the diary studies to which these
results are compared are flawed with regard to the interaction ratio. In order to
capture as many lies as possible, DePaulo et al. (1996) specified that subjects were to
record all interactions of 10 minutes or more. Subjects were told to record all lies that
occurred during these interactions and, importantly, record lies occurring during
shorter interactions as well. As a result, the ratio of total lies, regardless of interaction
duration, to 10-minute interactions distorts the true relationship. If the number of
10-minute interactions varies across respondents or modes of communication, these
comparisons may be more misleading than comparisons of the number of lies in each
category during the fixed 24-hour time frame. Subsequent diary studies incorporate
the lie per interaction distortion created by the DePaulo et al. methodology.
In order to consider more fully the possibility of individual differences in the
propensity to lie, a multiple regression analysis was performed. Initially, a natural
logarithm transformation was applied to the continuous lying measure as a means of
reducing its nonnormality. Although not eliminating the nonnormality completely,
this transformation had the effect of decreasing skewness by a factor of approximately
4 and kurtosis by a factor of approximately 18. To assess the impact of the
demographic measures on lying, the natural logarithm transformed lying measure
was regressed onto all demographic measures. Trivial predictors were dropped, the
analysis was iterated, and two important, albeit modest in magnitude, predictors
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 11
Prevalence of Lying in America K. B. Serota et al.
Prevalence of Mediated Lies - by Receiver Type
0 5 10 15 20 25 30 35 40 45
Number of Lies
y = 80.271 * x^-2.139 Family
y = 62.774 * x^-1.651 Friends
y = 32.337 * x^-1.448 Business
y = 12.534 * x^-1.284 Acquainted
y = 35.555 * x^-2.187 Strangers
Figure 6 Standard power function curves fit the frequencies of mediated lies plotted by
number of lies for all receiver types.
emerged (R=.21, F(2, 979) =22.82, p<.001). First, age was an important predictor
(β=−.18, t(979) =−5.71, p<.001), and the effect was such that a decrease in
lying was associated with increasing age. Second, race (Caucasian vs. other) was
an important predictor (β=−.09, t(979) =−2.99, p=.003), and the effect was
such that Caucasians reported lying less than other racial groups. This analysis was
replicated with the dichotomous not lie/lie measure, with the same two important
predictors emerging in the subsequent logistic regression analysis.
No sex differences were observed when controlling for other demographic
predictors. Bivariate analysis showed that the overall rate of lying by men (M=1.93,
SD =4.81, n=482) and women (M=1.39, SD =4.08, n=516) is directionally but
not significantly different when using conventional criteria for statistical significance
(t(996) =1.89, p=.059/ns, two-tailed, d=0.12).3This apparent gender difference
is in the opposite direction observed by DePaulo et al. (1996) but is consistent with
the finding that women find lying less acceptable than men (Levine, McCornack, &
Avery, 1992).
The results of this national study are consistent with the DePaulo et al. (1996) diary
study and suggest that on average Americans lie once or twice a day. However, the
important findings are that many people do not lie on a given day, the majority of
lies are told by a few prolific liars, and because the distribution is highly skewed, the
mean number of lies per day is misleading. This pattern is consistent across modes of
12 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
communication and varies little on the basis of who is being lied to. Examination of
individual differences suggests some variation but in most cases the differences are
The representativeness of online panel data is debatable. Use of poststratification
(such as employed by Synovate) and propensity (likelihood of response) weighting
schemes are the usual solutions to matching panel samples to the population. In
general, the use of a large number of small and internally homogeneous strata
will enhance the proportional fit of the sample to the population. However, non-
coverage and self-selection can create sampling problems in Web-based research
that weighting does not solve (Loosveldt & Sonck, 2008). These representative-
ness issues are of particular concern when the measured values are correlated with
the underlying reasons for the selection bias. However, when measuring socially
undesirable behavior, assessment of representativeness is confounded. Loosveldt and
Sonck obtained measures for validation of questions influenced by social desirability
using face-to-face interviews; this introduces potential mode effects. Even with the
selection problem, the social distance advantage of the Internet may actually produce
better data. Birnbaum (2004) provides an argument for representativeness when the
results are homogeneous across strata. Because selection bias tends to vary across
the levels of a stratified sample, small individual differences for a measure across the
stratification variables is evidence that the aggregate finding transcends the sampling
issues and is indicative that the sample result is representative of the phenomenon in
the population as a whole. In Study 1, confidence in representativeness is enhanced by
the homogeneity of the results across strata, and convergent validity is subsequently
established by consistency with the results in Studies 2 and 3, and by the advantage
of the Internet for creating social distance in the measurement of a sensitive topic.
Although the findings were generally homogenous across the sample of adult
Americans, some small demographic differences were apparent. Notably, age and
race/ethnicity account for small but statistically significant variation. Further, the
difference between reports of lying by men and women approached statistical
significance. These findings may have theoretical, social, and cultural implications.
Perhaps the most interesting individual difference is the negative association
between lying and age. Lying is acquired by children in early childhood and the
ability to lie is correlated with the acquisition of perspective-taking, theory of mind,
and communication skills (Vasek, 1986; see Knapp, 2008, pp. 91– 116 for a summary
discussion of lying and development). As the child reaches adolescence, lying skill is
perfected, but lying declines in acceptance in early adulthood (Jensen et al., 2004).
The difference between rates of lies reported by the DePaulo et al. (1996) student
and (adult) community samples suggests that maturity tempers the usage of lying as
a strategy for goal attainment, and the current findings of the national data in Study
1 are consistent with the claim that the lying declines with age.
With regard to the finding that Caucasians report fewer lies than those of other
ethnic or racial groups, it would be irresponsible to simply conclude that White
people are more honest in general than those of other races. Research on race and
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 13
Prevalence of Lying in America K. B. Serota et al.
deception is limited, and more research is required to make sense of this finding.
The current study also found a marginally significant trend toward men reporting
more lies than women. Some studies suggest men lie more than woman while others
suggest the opposite (e.g., DePaulo et al., 1996; Levine et al., 1992). Other research
finds that sex differences vary by the topic of the lie (Haselton, Buss, Oubaid, &
Angleitner, 2005).
Study 2
The results of Study 1 replicate, with a large and nationally representative sample,
the often repeated conclusion that people lie, on average, once or twice per day. The
results also document that the distribution of lies per day is substantially skewed.
Most reported lies were told by a few prolific liars.
These findings have important implications. Most importantly, the nature of the
distribution makes conclusions drawn from sample means misleading. Although the
mean lies per day reported in the literature appear reflective of aggregate reality,
the mean as a central tendency does not reflect the lying behavior of the typical
person. Instead, most people reported telling no lies at all on a given day, with the
median and mode both being zero. We are not the first to note this shape for the
lie frequency distribution. A similar pattern, with many telling a few lies and a few
telling most of the lies, was reported by Feldman, Forrest, and Happ (2002) in a
laboratory study on self-presentation. But in their analysis of the data, the skew was
treated as a methodological limitation rather than as a finding. The clarity of the
results observed in the national study raises a question of whether or not this pattern
existed in previous studies reporting lying frequency.
The raw data were obtained from the student phase of the DePaulo et al. (1996)
study and from both phases of the George and Robb (2008) study. A distribution
of lie frequency was partially reconstructed from the results reported by Feldman
et al. (2002). The shape of each of the four distributions (excluding those reporting
no lies) is examined, and the resulting distributions are compared with the overall
results from the national survey. Data from the DePaulo et al. community sample
and Hancock et al. (2004) studies were not available.
DePaulo et al. (1996)
Of the 77 students sampled by DePaulo et al., 76 reported telling at least one lie
over the period of 1 week (Mweek =13.74, SDweek =11.40; this is equivalent to
Mday =1.96, SDday =1.63); the total number of lies told was 1,058; and the most lies
told by one person was 46 (equivalent to 6.6 lies per day). Curve-fitting yields a power
function for these data of y=5.366 ×x0.290(n=76, r2=.289). This function has
a poor fit; however, no equation provided a fit better than r2=.350. Even so, the
14 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
data exhibit the overall distributional properties of a few lies told by most of the
subjects and most of the lies told by a few liars. Of the total students in this study,
66.2% told the equivalent of two lies per day or less. Conversely, the seven most
frequent liars (9.2% of the sample) told more than the equivalent of five lies per day,
or 26.2% of all lies reported. Although raw data for the community sample are not
available, DePaulo et al. reported 64 of 70 subjects told 477 lies with Mweek =6.79
(SDweek =6.86) and Mdnweek =4.5; these measures suggest a positive skew similar
to that observed with the student sample.
George and Robb (2008)
Two studies were conducted following the diary methodology used by DePaulo et al.
(1996) and Hancock et al. (2004). The objective of the two studies was to examine
variations in deceptive behavior by media use. A key difference between these studies
and the prior diary studies was the use of PDAs instead of paper and pencil to
record interactions and lies. In the first study, George and Robb used the 10-minute
minimum time established in the prior diary studies to define an interaction; in the
second study, the minimum interaction time was reduced to 5 minutes. In both
studies, subjects were instructed to record lies even when the interaction was shorter
than the prescribed time.
Results of this research are notably different from the prior diary studies. In each
study, the mean number of lies was small. In the first study, 24 of 25 subjects reported
lying over the period of a week (Mweek =4.16, SDweek =2.59; this is equivalent to
Mday =0.59, SDday =0.37; Mdn =4); the total number of lies told was 104; and
the most lies told by one individual was 11 (the equivalent of 1.6 lies per day).
Despite the low average, only one subject who reported lying told fewer than two lies;
consequently, a power function could not be fit to the distribution (if this individual
is eliminated a power function with a modest r2=.758 can be fit to the remaining
data). Even with this poor fit, the distributional properties of the study demonstrated
a positive skew similar to the national survey and the DePaulo et al. (1996) diary
study. Half of the subjects (48% of the sample) told just 25% of the lies while the
three subjects (12%) reporting the most lies told 26.9% of the lies.
In George and Robb’s (2008) second study, the length of interaction was shortened
in order to encourage more reporting. As a result, 23 of 24 subjects reported lying
over the period of a week and the lying frequency increased from the first study
(Mweek =6.33, SDweek =3.78; this is equivalent to Mday =0.90, SDday =0.54;
Mdn =5). The total number of lies reported was 152 and the most lies told by one
individual was 14 (the equivalent of two lies per day). Similar to the first study, and
despite the overall low number of lies told, those reporting lies told a minimum of
two lies; consequently, a power function could not be fit to the data. Nonetheless, the
positive skew is again apparent in the pattern of responses. Just three subjects (12.5%
of the sample) told 40 of 152 lies (26.3% of the total).
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 15
Prevalence of Lying in America K. B. Serota et al.
Feldman et al. (2002)
Although addressing a much shorter time frame in an experimental setting, Feldman
et al. (2002) observe the same distinct skew pattern in their lying data as was observed
in the daily national study and the weekly student diary studies. This study examined
lying as a component of self-presentation; a sample of 121 students was divided into
two induction groups and an experimental control. Across all subjects, a total of 211
lies were analyzed with a mean of 1.75 lies per subject and 2.92 lies per subject who
lied.4Of the 121 subjects, 49 told no lies (40.5%), 23 told one lie (19.0%), and 18
told two lies (14.9%). The maximum number of lies was 12. Thus, 74.4% of the total
sample accounted for only 28% of the lies told. The remaining 31 subjects (25.6%)
told 312 lies, accounting for 72% of the lies (M=4.9 lies per subject).
When prior research reporting the frequency of lies is reexamined, results show
the small diary samples, the experimental self-presentation study, and our large,
national self-report survey have similarly skewed distributions. In all cases, the
infrequent liars are a large part of the sample and account for a disproportionately
small share of the lies reported. Conversely, each study includes a small number of
individuals who account for a disproportionately large share of the lies.
Study 3
Several issues are of concern despite the consistency of the findings across studies
and the face validity obtained by reanalysis of prior studies reporting lie frequency.
Primarily, these concerns have to do with the accuracy of the study findings using
the survey approach to obtaining self-report data. Further, the apparent discrepancy
between the numbers of nonliars when the time frame is 1 day versus the number
when the time frame is 1 week needs to be resolved. For these reasons, the national
survey was replicated using student samples and additional measures.
Accuracy of reporting lies
An obvious concern with self-report, mass survey research is accuracy of reporting.
Bias in the national study data, if it exists, would likely manifest itself as under-
reporting. Given the pervasive cultural prohibitions against lying, self-presentation
motives favor under- rather than overreporting. Methodological research on other
sensitive topics such as drinking (Lemmens, Tan, & Knibbe, 1992) and sexual behavior
(Ramjee, Weber, & Morar, 1999) indicates that diary studies produce higher mean
scores than self-report questionnaires. The diary method used by DePaulo et al.
(1996) is expected to be less susceptible to underreporting bias even though the study
had the added limitations of a smaller and nonrepresentative sample. Because the
mean number of lies per day in the national survey data was greater than the mean
observed in the diary research, concern that the national study may be underreporting
due to the use of the self-report questionnaire method is not consistent with results
of most comparisons of survey research to an alternative diary method. Thus, the
16 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
observed frequency of lying reported is not likely attributable solely to the survey
methodology used to collect the national study data.
Nonetheless, a survey of lying behavior invites the question: ‘‘How do you know
the subjects are not lying?’’ One answer resides in the methods used to address
sensitive questions. Tourangeau and Yan (2007) identify four techniques for asking
sensitive questions that might contribute to improved prevalence measurement:
Use of self-administered rather than interviewer-administered questions, forgiving
wording, randomized response techniques (RRT), and use of the bogus pipeline
(BPL) approach. The national survey of lying was conducted as a self-administered
study and used a forgiving wording preamble. The meta-analysis by Tourangeau and
Yan (2007) provides clear evidence that self-administration yields more behavioral
reporting for sensitive topics than does interviewer-administration. There is limited
evidence that forgiving wording may also be helpful. Catania et al. (1996) found
increased reporting of sexual activity but a series of experiments by Holtgraves, Eck,
and Lasky (1997) suggest that forgiving wording is more likely to improve attitudinal
responses than behavioral reporting.
A second answer can be obtained through the use of additional measures in
order to assess the social desirability bias (SDB) in the self-report measures (Fisher &
Katz, 2000). This is sometimes done directly by obtaining measures of SDB using a
scale of SDB traits (Crowne & Marlowe, 1960; Reynolds, 1982) and adjusting for the
subjects’ level of bias. Alternatively, Fisher and Katz suggest the validity of reports of
a socially undesirable behavior such as lying can be assessed indirectly, for example,
by comparing the self-report of that behavior to subjects’ estimates of the extent of
that behavior in others. Subjects in Study 3 were asked to report the total number of
times others had lied to them in the past 24 hours.
Minimum observable differences
A second key issue raised by the results of the national study is the minimum
observable difference of one lie per day. Those who lie but do so less than once a
day may be recorded as having told no lies. Because the frequency of lies is typically
reported as the rate of lying in a fixed interval of time (i.e., one to two lies per day),
expanding the duration in which the behavior can be observed is likely to increase
the overall reported incidence of the behavior; measures taken over a longer period
and converted to a daily rate will also increase precision. In order to observe those
who lie once a week or once a month or even less frequently, it is necessary to use a
wider time aperture. In the student sample of the DePaulo et al. (1996) study, 30%
of diaries recorded six or fewer lies per week (less than once per day). If the number
of lies told in a week is evenly distributed across the days of the week (and there is
no reason to believe an even distribution is a good assumption), we might expect
that as many as 17% (those reporting four, five, or six lies in a week) have an above
average likelihood of being included as liars in a 1-day study. But the other 13%
(reporting one, two, or three lies in a week) would be more likely to report not lying
on the survey day. Identifying infrequent liars by repeating the daily survey on 2 or
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 17
Prevalence of Lying in America K. B. Serota et al.
more days with the same sample, or by asking those who report no lies to identify
other times when they had told a lie, we would expect to narrow substantially the gap
between the national study observation that 40% told a lie on one given day and the
DePaulo et al. result that 95% of subjects (both studies combined) reported at least
one lie over the period of a week.
Participants and design
Students were recruited from communication, advertising, and marketing classes at
two large universities in the Midwestern United States. Of the 229 subjects, two failed
to provide complete information regarding the number of lies told and two provided
answers that were determined to be statistical outliers using a discordancy test. These
subjects were excluded from the analysis. Of the remaining 225 respondents, 140
were female (62.2%) and 85 were male (37.8%). Participation was voluntary.
The primary purpose of this study was to cross-validate the results of the national
study using a separate sample. A nonexperimental survey design similar to that of the
national study was used in order to obtain comparable descriptive measures for the
incidence of lying in the student sample. Results from this survey are compared with
those of Study 1 and the reanalysis of prior studies in Study 2.
Procedures and measures
The study was administered in five separate classroom settings using a paper
questionnaire. The introduction and frequency of lies question were identical to that
of the online survey used with the national sample except that students could leave
parts of the 5 ×2 grid blank (the online survey forced subjects to place zeroes in all
cells for which no lies were told before they could continue with the questionnaire).
Blank cells were coded as zero lies. In addition, those subjects who reported telling
no lies in the past 24 hours were asked a follow-up closed-ended question regarding
when they last told a lie. Response categories included, ‘‘more than 24 hours ago but
within the last 2 days,’’ ‘‘more than 2 days ago but within the last week,’’ ‘‘more than
a week ago but within the last month,’’ ‘‘more than a month ago,’’ and ‘‘never.’’
Subjects were asked how many lies they had been told by others in the past 24 hours
and what percentage of the U.S. adult population lies on a given day. One group of
subjects was asked about the difficulty of the lie reporting task.
The pattern of lie frequency for the total student sample is consistent with the
distribution of lies for the national survey using the same frequency measure. The
data fit a power function of y=53.891 ×x1.012 with a goodness-of-fit of r2=.894.
The pattern of many telling few lies versus a few telling many lies is reproduced in the
student sample frequencies. Of the total subjects in the student sample, 68.4% told one,
two, or no lies (accounting for 24.5% of 526 total lies); conversely, the 13 most prolific
liars (5.8% of the total sample) told 22.4% of the lies. On the basis of the relationship
18 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
of the means in the DePaulo et al. (1996) student and community diary studies, we
expected the mean of the student survey to be higher than the mean of the national
adult sample. Results from the student sample (M=2.34 lies per day, SD =2.94,
Mdn =1, Mode =0, N=225, Max =21 lies, 95% CI =1.952.72) indicate higher
frequency than in the national survey (nonoverlapping 95% confidence intervals;
t(1221) =2.62, p<.01, d=0.18).
Those subjects who reported telling no lies in the past 24 hours (28.9%) were
asked the follow-up question regarding when they last told a lie. Combining those
who indicated their most recent lie was within the past week (though not in the past
24 hours) with those who reported at least one lie in the past 24 hours, a total of
92.4% reported telling a lie in the past week (204 of 221 subjects; 4 did not answer
the follow-up question). This result is reasonably consistent with the DePaulo et al.
(1996) result of 95%.
Since there is a legitimate concern about SDB in the self-report of lying, a check
on the subjects’ reported rate of lying (lies per day) was obtained by asking subjects in
the student sample how many times they believe they were lied to by others in the past
24 hours. Subjects reported that others told an average of 2.79 lies in the past 24 hours
(SD =2.82, N=198). Although 28.9% reported telling no lies, 17.2% reported not
being lied to by others. Own lies (M=2.34, SD =2.94) are significantly lower
than others’ lies; the paired sample t(197) =2.18, p<.05. Despite this significant
difference, the distribution of others’ lies is strikingly similar to that of own lies
following the long-tail pattern observed in the self-report of lies (the power function
for others’ lies is y=44.552 ×x0.790 with a modest goodness-of-fit of r2=.700).
It is important to note that individual subject’s reports of others’ lies did not mimic
their self-report of lying. The correlation of the two measures is a meager r=.073
(p=.152, ns).
Participants were also asked what proportion of the U.S. adult population
(18 years or older) told at least one lie on any given day. The mean estimate is
75.8% (SD =21.7). The proportion of students (who are a subsegment of the adult
population) telling at least one lie in the past 24 hours is 71.1% (CI 5.9% points).
Although not directly comparable due to the different ways in which the proportions
were derived, in both instances one percentage falls within the confidence interval of
the other, suggesting that the two proportions would not be significantly different.
The results of Study 3 fit a power function similar to that of the national survey
and further indicated that, based on self-reports of lying behavior, most people tell
few or no lies in a given day but a few prolific liars tell a disproportionately large
share of the daily lies. Survey results, diary studies, and the distribution of lies in an
experimental setting share this pattern, and the consistency of results provides strong
evidence that the frequency of lying has a strong positively skew. The similarity of
subjects’ estimates of others’ lies and the distribution of number of lies told provides
convergent validity for the self-report measure of lying.
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 19
Prevalence of Lying in America K. B. Serota et al.
With regard to the accuracy of mean prevalence figures, Study 3 provides mixed
results. The mean reported by this student sample is 2.34 lies per day and expands
the range of responses across all studies to 0.592.34 lies per day. The mean for
the student survey is significantly higher than the mean of the national survey of
American adults. This is similar to the relationship observed by DePaulo et al. (1996)
for their student and community samples. Consistent with the Study 1 finding that
younger people tend to lie more than older people, the predominantly young student
sample has a higher frequency of lying. Two measures were obtained to assess the
validity of the reported lying frequency. One measure estimated the proportion of the
population that lies on a given day; the result of 75.8% is similar to the Study 3 result
showing that 71.1% of the subjects told at least one lie in the past 24 hours. However,
an estimate of others’ lies to the subjects (M=2.79) is significantly higher than the
average of the subjects’ own lies. Either the subjects truly believe that other people
lie more than they do or the subjects are slightly but systematically underestimating
their own lying behavior.
Regardless, the difference is one of degree and not magnitude, and the relevant
finding is the consistent skew of the distribution and not the average number of lies
told. In fact, the mean of a long-tail (power) distribution is in part a function of
sample size. As the sample size increases, the probability of rare but legitimate extreme
values that are not statistical outliers being reported also increases. In Study 1, the
maximum number of lies by one person was 46 (N=998). Study 3 consisted of five
subsamples (separate classroom administrations of the survey) and the maximum
number of lies told was closely associated with sample size. In the largest subgroup
(n=118), the maximum number of lies was 21. In the next largest group (n=31),
the most lies told was 12. In the three smallest subsamples (n=26, 25, and 25,
respectively), the most lies told by one person were 10, 8, and 6.
A question related to accuracy may be raised with regard to the difficulty of the
task (since it might be expected that a more difficult task would produce increased
variance). We note that researchers familiar with the topic of deception tend to
struggle more with the question asked in the self-administered survey than do naive
subjects. Those familiar with the deception literature wrestle with the explication of
lying and, perhaps because of this cognitive effort, their own ability to recall precisely
over a 24-hour period. Naive subjects seem to have no such problem. As a check,
25 students in one Study 3 subsample completed the recall of lies task and were
subsequently asked to rate the difficulty of answering the question. On a 1 10 scale
(1 =not at all difficult; 10 =extremely difficult), the subjects reported M=3.48
(SD =3.03) with the Mdn =2andtheMode=1. Most subjects seemed to feel that
they were able to complete the task with little difficulty. During a verbal debriefing
of this subject group, subjects reiterated that most had made an earnest effort to
estimate the number of times they had lied and most felt the task was not difficult.
Finally, Study 3 appears to have resolved the discrepancy in the proportion of
nonliars reported in the daily survey and weekly diary results. All of the diary studies
report that more than 90% of the subjects told at least one lie in the period of 1
20 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
week. However, many of those subjects lie at a rate that translates to less than one lie
per day. When Study 3 subjects who did not report lying in the 24 hours preceding
the survey were asked when they last told a lie, results indicate that most did lie in
the previous week. Therefore, the apparent discrepancy is due to the precision of
the question being asked. The daily questions do not allow for fractional responses,
whereas the conversion of weekly data to daily data inherently creates fractional
reporting. When the precision issue is accounted for by additional measurement, the
discrepancy disappears.
General discussion
The results of Study 1 reproduce with a large, national sample the conclusion that
people lie, on average (arithmetic mean), once or twice per day. More importantly,
the results document that the distribution of self-reported lies per day is substantially
skewed. Most people report telling few or no lies on a given day and most reported
lies are told by a few prolific liars. The reanalyses in Study 2 and the replication of the
survey methodology in Study 3 provide strong evidence in support of this finding.
All the data reported here are consistent with the claim that most lies are told by a
few prolific liars.
The highly skewed, long-tail distribution that results from reports of lying may
be emblematic of a larger class of behaviors. Although much of what we measure
and observe relies on the principle of randomness and results in the well-known
bell-shaped curve of observations distributed around a central tendency for a
phenomenon, scientists continue to find new evidence for a self-organizing principle
in nature that produces distributions that are scale-free, or lacking in a characteristic
tendency (Barab´
asi, 2003; Barab´
asi & Albert, 1999). This principle is found in the
examination of atomic structure, biological systems, economic theory (the Pereto
Law or ‘‘80/20 Rule’’), and the emergence of nodes on the Internet. The latter
describes a pattern consisting of a few very large nodes that are connected with
extremely high frequency (e.g., Google, eBay, and that are surrounded
by over 100 million smaller and, in many cases isolated, Websites. But it is the human
tendency to search, use, and link to these few extreme large sites while ignoring so
many others that creates this apparent scale-free pattern in the first place. It may be
less the case that lying is some unique form of communicative behavior that divides
people into those who do it (with vigor) and those who do not than an indicator
that we need to reexamine a broad range of social and symbolic behaviors, look
for scale-free distributions, and consider broadly the implications for all forms of
communication research.
Our finding with regard to the distribution of reported lies has specific impli-
cations for the conclusions drawn from deception detection accuracy experiments.
A meta-analysis shows that people do only slightly better than chance (54%) when
distinguishing between truths and lies (Bond & DePaulo, 2006). In the experiments
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 21
Prevalence of Lying in America K. B. Serota et al.
cited by Bond and DePaulo, the deception base rate is almost always 50% and sub-
jects are typically truth-biased. Truth bias leads to the veracity effect, meaning that
accuracy is higher for detecting honest messages than for detecting lies (Levine, Park,
& McCornack, 1999). As a consequence, honesty base rates impact accuracy such that
people are increasingly accurate as the proportion of truthful to deceptive messages
increases (Levine, Kim, Park, & Hughes, 2006). The finding that most people are
honest most of the time suggests that experiments employing nonrepresentative base
rates (i.e., a lower proportion of truths to lies than is typical of everyday interactions)
will underestimate accuracy. Findings from this general population study indicate
that truth bias may be functional. Since self-report data suggest most people do not
lie or tell very few lies in the course of a typical day, it is reasonable for people to
believe others (that is, to be truth-biased) most of time.
However, although laboratory deception detection experiments may underesti-
mate overall accuracy due to nonrepresentative base rates, lie detection rates may
be overestimated. The current findings suggest substantial individual differences for
truth telling (or lying) with an honest majority and a few prolific liars. Bond and
DePaulo (2008) report substantial variation in both people’s demeanor and people’s
ability to lie. People’s ability to lie successfully likely impacts how often they lie (Kashy
& DePaulo, 1996). We speculate that the prolific liars are likely those people with
especially honest demeanor and that unusually transparent liars avoid lying. If most
lies outside the laboratory are told by people who are usually believed, lie detection
rates would be lower than those observed in randomized experiments.
One caution relates to Rozin’s (2001) concern for achieving ‘‘context- and
culture-sensitive scientific social psychology.’’ The large sample of American adults
used in the current research is more representative and allows greater generalizability
than previous studies using students and other convenience samples. Nevertheless,
caution should be taken before assuming that these results will hold for other cultures,
individuals under 18 years of age, or subsets of the population with characteristics or
beliefs that are sharply discrepant from the norm. Anecdotal evidence suggests that
base rates may well vary across cultures and studies of lying involving young children
and adolescents show considerable variation in the ways lying occurs (Knapp, 2008).
Criminals, political extremists, and those at the far ends of the socioeconomic
spectrum may well behave differently than the vast majority of the American
population. The domain of deception research would benefit from cross-cultural,
subcultural, and cross-generational examination of the prevalence and distribution
of lying behavior.
Over time, most Americans probably lie at least occasionally. But the inference of
pervasive daily lying, drawn from the statistical average of one to two lies per day,
and reinforced by media coverage of corporate deception and political malfeasance,
as well as pop culture portrayals of deception detectors, belies the basic honesty
22 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
present in most people’s everyday communication. Self-report data for the U.S. adult
population show the average rate of lying is around 1.65 lies per day; but the data are
not normally distributed. On any given day, the majority of lies are told by a small
portion of the population, and nearly 6 out of 10 Americans claim to have told no
lies at all. As researchers continue to examine the nature of lying and look for ways to
detect deception effectively, both the theories of deception, and the methods used to
test those theories, necessarily must take into account that veracity is not a constant
across the population and that the propensity to lie can be an important moderator.
The authors thank Deborah Kashy and Joey George for providing the original data
from their studies for additional analysis.
1 Synovate, Inc. (Chicago) conducts marketing research and strategic studies for business,
industry, and government agencies. White papers on the omnibus methodology are
available at The omnibus panel was surveyed on April 30, 2008.
2 Two subjects, one with an extreme value of 134 lies (20.8 standard deviations above the
mean of 1.80 lies) and a corresponding subject with no lies, were eliminated from the
sample by α-trimming. Using procedures outlined by Barnett and Lewis (1978), a
discordancy test appropriate for gamma distributions was applied to the data. Even
though the potential for extreme values is inherent in the tail of the power function, the
most extreme value recorded in these data is well beyond the limits of statistical
3 The application of a t-test for establishing the significance of differences for nonnormal
distributions is problematic. Nonetheless, in those situations where the shapes of the
distributions are similar, the sample sizes are similar, and the samples are sufficiently
large the t-test will provide an acceptable test of significance even though the assumption
of normality has been violated (Johnson, 1978). Although most of the small student
sample studies fail to meet all of these conditions, the power distributions in this
large-scale national study replicate for most subsamples and do so with large enough
samples to allow for robust statistical testing. An alternative is to make the nonparametric
assumption and apply difference testing accordingly. Calculating chi-square for age
(under 45 years old vs. 45 years and older) and lying (did vs. did not lie) yields
χ2(1, N=998) =36.59, p<.001; this leads to a conclusion similar to the one reported
in the text, specifically that younger people are more likely to lie than older people.
Calculating chi-square for gender and lying yields χ2(1, N=998) =2.84, p<.092;
again the result is consistent with the reported t-test showing that men are directionally
but not significantly more likely to lie than are women.
4 Despite including the induction groups, using a student sample, and being constrained to
only those who told lies, the mean of 2.92 lies has been misinterpreted in the media
(online program description, About Lie to Me, 2009,
and in scholarly work as claiming ‘‘in the course of a 10-minute interaction ...the
average person tells two to three lies’’ (Harrington, 2009, p. 4).
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 23
Prevalence of Lying in America K. B. Serota et al.
About Lie to Me. (2009). Online program description. Retrieved April 10, 2009, from
Anderson, C. (2008). The long tail.NewYork:Hyperion.
Asch, S. E. (1987). Social psychology (Original work published 1952). New York: Oxford
University Press.
asi, A.-L. (2003). Linked.NewYork:Plume.
asi, A.-L., & Albert, R. (1999). Emergence of scaling in random networks. Science,286,
Barnett, V., & Lewis, T. (1978). Outliers in statistical data.Chichester,UK:Wiley.
Birnbaum, M. H. (2004). Human research and data collection via the Internet. Annual
Review of Psychology,55, 803832.
Bok, S. (1999). Lying: Moral choice in public and private life.NewYork:Vintage.
Bond, C. F., Jr., & DePaulo, B. M. (2006). Accuracy of deception judgments. Review of
Personality and Social Psychology,10, 214 234.
Bond, C. F., Jr., & DePaulo, B. M. (2008). Individual differences in judging deception:
Accuracy and bias. Psychological Bulletin,134, 477–492.
Brown, J. H., Gupta, V. K., Li, B.-L., Milne, B. T., Restrepo, C., & West, G. B. (2002). The
fractal nature of nature: Power laws, ecological complexity, and biodiversity.
Philosophical Transactions of the Royal Society B,357, 619626.
Catania, J. A., Binson, D., Canchola, J., Pollack, L. M., Hauck, W., & Coates, T. J. (1996).
Effects of interviewer gender, interviewer choice, and item wording on responses to
questions concerning sexual behavior. Public Opinion Quarterly,60, 345375.
Cochran, S. D., & Mays, V. M. (1990). Sex, lies and HIV. New England Journal of Medicine,
322(11), 774–775.
Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of
psychopathy. Journal of Consulting Psychology,24, 349354.
DePaulo, B. M., Kashy, D. A., Kirkendol, S. E., Wyer, M. M., & Epstein, J. A. (1996). Lying in
everyday life. Journal of Personality and Social Psychology,70, 979995.
Feldman, R. S., Forrest, J. A., & Happ, B. R. (2002). Self-presentation and verbal deception:
Do self-presenters lie more? Basic and Applied Social Psychology,24(2), 163 170.
Fisher, R. J., & Katz, J. E. (2000). Social desirability bias and the validity of self-reported
values. Psychology & Marketing,17(2), 105 120.
Ford, C. V., King, B. H., & Hollender, M. H. (1988). Lies and liars: Psychiatric aspects of
prevarication. American Journal of Psychiatry,145, 554 562.
George, J. F., & Robb, A. (2008). Deception and computer-mediated communication in daily
life. Communication Reports,21, 92–103.
Hancock, J. T., Thom-Santelli, J., & Ritchie, T. (2004). Deception and design: The impact of
communication technology on lying behavior. CHI Letters,6(1), 129134.
Harrington, B. (Ed.). (2009). Deception: From ancient empires to Internet dating.Stanford,
CA: Stanford University Press.
Haselton, M. G., Buss, D. M., Oubaid, V., & Angleitner, A. (2005). Sex, lies, and strategic
interference: The psychology of deception between the sexes. Personality and Social
Psychology Bulletin,31, 3–23.
Holtgraves, T., Eck, J., & Lasky, B. (1997). Face management, question wording, and social
desirability. Journal of Applied Social Psychology,27, 16501671.
24 Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association
K. B. Serota et al. Prevalence of Lying in America
Jensen, L., Arnett, J., Feldman, S., & Cauffman, E. (2004). The right to do wrong: Lying to
parents among adolescents and emerging adults. Journal of Youth and Adolescence,33(2),
Johnson, N. J. (1978). Modified ttests and confidence intervals for asymmetric populations.
Journal of the American Statistical Association,73, 536544.
Kalish, N. (2004, January). How honest are you? Reader’s Digest, 114– 119.
Kish, L. (1965). Survey sampling.NewYork:JohnWiley&Sons.
Kashy, D. A., & DePaulo, B. M. (1996). Who lies? Journal of Personality and Social Psychology,
70, 1037–1051.
Knapp, M. L. (2008). Lying and deception in human interaction. Boston: Pearson Education.
Knox, D., Schacht, C., Holt, J., & Turner, J. (1993). Sexual lies among university students.
College Student Journal,26, 269272.
Lemmens, P., Tan, E. S., & Knibbe, R. A. (1992). Measuring quantity and frequency of
drinking in a general population survey: A comparison of five indices. Journal of Studies
on Alcohol,53, 476486.
Levashina, J., & Campion, M. (2007). Measuring faking in the employment interview:
Development and validation of an interview faking behavior scale. Journal of Applied
Psychology,92(6), 16381656.
Levine, T. R., Kim, R. K., & Hamel, L. M. (2009). People lie for a reason: An experimental test
of the principle of veracity. Unpublished manuscript, Michigan State University.
Levine, T. R., Kim, R. K., Park, H. S., & Hughes, M. (2006). Deception detection accuracy is a
predictable linear function of message veracity base-rate: A formal test of Park and
Levine’s probability model. Communication Monographs,73, 243260.
Levine, T. R., McCornack, S. A., & Avery, P. B. (1992). Sex differences in emotional reactions
to discovered deception. Communication Quarterly,40, 289 296.
Levine, T. R., Park, H. S., & McCornack, S. A. (1999). Accuracy in detecting truths and lies:
Documenting the ‘‘veracity effect.’’ Communication Monographs,66, 125144.
Loosveldt, G., & Sonck, N. (2008). An evaluation of the weighting procedures for an online
access panel survey. Survey Research Methods,2(2), 93105.
Patterson, J., & Kim, P. (1991). The day America told the truth.NewYork:Prentice-Hall.
Ramjee, G., Weber, A. E., & Morar, N. S. (1999). Recording sexual behavior: Comparison of
recall questionnaires with a coital diary. Sexually Transmitted Diseases,26, 374380.
Reynolds, W. M. (1982). Development of reliable and valid short forms of the
Marlowe-Crowne social desirability scale. Journal of Clinical Psychology,38, 119 125.
Rozin, P. (2001). Social psychology and science: Some lessons from Solomon Asch.
Personality and Social Psychology Review,5(1), 214.
The Real Truth About Lying (1996, September/October). Psychology Today,29, 16.
Tosone, C. (2006). Living everyday lies: The experience of self. Clinical Social Work Journal,
34(3), 335348.
Tourangeau, R., & Yan, T. (2007). Sensitive questions in surveys. Psychological Bulletin,
133(5), 859883.
U.S. Census Bureau (2008). Current population survey. Retrieved June 5, 2008, from
Vasek, M. E. (1986). Lying as a skill: The development of deception in children. In
R. W. Mitchell & N. S. Thompson (Eds.), Deception perspectives on human and
nonhuman deceit (pp. 271 –292). Albany: State University of New York Press.
Human Communication Research 36 (2010) 2–25 ©2010 International Communication Association 25
... Based on an influential couple of diary studies (DePaulo et al., 1996), the accepted lore in the deception literature was that people lie once or twice per day. Although this finding has held up well, it applies only on average (Serota et al., 2010). The catch is that lie frequency data have long-tail distributions. ...
... Prior findings consistently align with TDT's predictions regarding lie prevalence. Initial evidence for the few prolific liar predictions came from a representative survey of 1,000 American adults, a survey of 225 American students, and the re-analyses of several previously published diary and experimental studies (Serota et al., 2010). Serota et al.'s results were then replicated twice in the Netherlands with both student (Halevy et al., 2014) and non-student (Debey et al., 2015) samples. ...
... In all seven countries, similar self-reported lie prevalence measures were obtained. Respondents were given the preamble developed by Serota et al. (2010): ...
Truth-default theory (TDT), a theory of human deception and deception detection, has two propositions that focus on the overall rate of lying and individual variation in the frequency of lying behavior. The distribution of lie prevalence is specified to exhibit a non-normal, positively skewed distribution in which the majority of people are normatively honest, and most lies are told by a few prolific liars. Together, these predictions form the few prolific liars modules in TDT. Although the findings of prior research align with TDT predictions, the pan-cultural scope of TDT warrants testing such predictions with new and diverse samples. The current studies (total N = 3,463) sampled participants from China, Germany, Mexico, Israel, Kenya, Russia, and Brazil. Similar long-tail distributions were observed in each of the seven locations, and in language and cultural subsamples. These findings add to a growing empirical literature providing pan-cultural evidence consistent with TDT.
... Moreover, people also care about the loses of others along with their own gains. Serota et al. [42] conducted an experiment with 1000 Americans, and it was found that on average 1.65 lies were told by each American per day. Similarly, other experiments [43][44][45][46] proved that lies were told between frequency 0.6-2.0 per day, where face-to-face intersections had a lower chance of deception in comparison to telephonic conversations. ...
Full-text available
Dunbar's number is the cognitive limit of an individual to maintain stable relationships with others in his network. It is based on the size of the neocortex of the human brain. On the other hand, trust is one of the major issues for one while selecting members for his social network and the evolution of his social network with time. Trust and Dunbar's number are interconnected in the case of one's stable social network. Trust needs time to be built after several social interactions, intimacy, etc. In this paper, we try to provide answers to the following important questions related to social networks: (i) Do trust levels remain the same for individuals from one's perspective in his social network when the network size increases? (ii) What is the relation between the power-law exponent α and the trust cutoff? (iii) Do trust levels help to diffuse information quickly or vice versa to reach Dunbar's number 150 along with hierarchy layers of 5, 15, and 50 individuals in networks of different sizes? We find that there is a requirement for trust levels to increase among the same individuals in one's social network if the size of the network increases. As a relation between the power-law exponent α and the trust cutoff, it is found that α∝ 1/(trust cutoff). Moreover, we also find that trust levels never help to diffuse information quickly or vice versa to reach Dunbar's number 150, along with hierarchy layers of 5, 15, and 50 individuals in networks of different sizes.
... However, conversations also provide a stage for trust to be violated. For example, people admit to telling an average of one or two lies per day, with a minority of individuals engaging in many more 6,7 . Although people generally tell the truth and admit to lying only rarely, discovering these falsehoods can damage trust between parties 8 . ...
Full-text available
Lies can have major consequences if undetected. Research to date has focused primarily on the consequences of deception for receivers once lies are discovered. We advance deception research and relationship science by studying the social consequences of deception for the sender—even if their lies remain undetected. In a correlational study of video conversations (Study 1; N = 776), an experimental study of text conversations (Study 2; N = 416), and a survey of dispositional tendencies (Study 3; N = 399), we find consistent evidence that people who lie tend to assume that others are lying too, and this impedes their ability to form social connections. The findings provide insight into how (dis)honesty and loneliness may go together, and suggest that lies—even when undetected—harm our relationships.
... That argument is limited in my view. Human communication is generally truthful, but it is incontrovertible that, occasionally, everybody lies (DePaulo & Bell, 1996;Serota et al., 2010;Serota & Levine, 2015). Hence, we cannot definitively associate lying or truth-telling with any personality trait (cf. ...
Full-text available
This article examines the idea that cognitive load interventions can expose lies—because lying is more demanding than truth-telling. I discuss the limitations of that hypothesis by reviewing seven of its justifications. For example, liars must suppress the truth while lying, and this handicap makes lying challenging such that one can exploit the challenge to expose lies. The theoretical fitness of each justification is variable and unknown. Those ambiguities prevent analysts from ascertaining the verisimilitude of the hypothesis. I propose research questions whose answers could assist in specifying the justifications and making cognitive load lie detection amenable to severe testing.
... Journalists and political fact-checkers should be aware of uniquely high-frequency deceivers (prolific liars, or those who lie prolifically; [43][44][45][46][47]) and when they might be more likely to lie (e.g., as suggested by TDT, when topics are consequential or when the truth is a problem; [32]). For example, during his final year as president and as the 2020 U.S. presidential election approached, President Trump's lying rate increased dramatically compared to other years [23]. ...
Full-text available
Professional fact-checkers and fact-checking organizations provide a critical public service. Skeptics of modern media, however, often question the accuracy and objectivity of fact-checkers. The current study assessed agreement among two independent fact-checkers, The Washington Post and PolitiFact, regarding the false and misleading statements of then President Donald J. Trump. Differences in statement selection and deceptiveness scaling were investigated. The Washington Post checked PolitiFact fact-checks 77.4% of the time (22.6% selection disagreement). Moderate agreement was observed for deceptiveness scaling. Nearly complete agreement was observed for bottom-line attributed veracity. Additional cross-checking with other sources (Snopes,, original sources, and with fact-checking for the first 100 days of President Joe Biden's administration were inconsistent with potential ideology effects. Our evidence suggests fact-checking is a difficult enterprise, there is considerable variability between fact-checkers in the raw number of statements that are checked, and finally, selection and scaling account for apparent discrepancies among fact-checkers.
... In the later, the tangle comes off as a neck-breaking exercise; much of the time attempting alchemy with self-imposed private ideas intended to impress and sedate society to tow authority's path and line of reasoning, in most climes. The means of doing that is often sheath in lots of indecipherable ornaments of spins, freeflying, in the shape of rumor; either as convenient construct of 'truths', or emotive lies and shiny actions and even inactions that model a discretize reality (Gallup, 2016;Serota, Levine, & Boster, 2010;Daza, Vilca, Salinas, Pomereda and Quico, 2021) that people barely remember to question the inner drive of the anticipated goal of the dispatched gossip, which looks a bit like grey propaganda (Baran and Davis, 2006) or the rat bite-and-blow tactics (Onobe, 2012). ...
Ideas and government policies, whatever they are, are ventilated on step-motherly ration and fanned as ‘gossip', sometimes, to test their popularity or otherwise before they are implemented. Employing deduction from observation, and using longitudinal approach and literature, findings of this investigation reveal that governments in the Global South adopt this communication strategy as an inverted feedback mechanism for policies they are not certain will enjoy public acceptability. The exploratory study stretches the parameters of the usage of the tactics by interrogating earlier theories like grapevine, propaganda. The study lays a foundation of the operational principles and applicability of a new political communication theory, structured-deliberative gossip theory (or grapevine info-filter theory), especially on matters of plebiscites in the Global South.
... 4.2.5 Diaries and surveys were used to recall a situation or any past experience in a short period of time. In this experiment, the participants were asked to write diaries concerning what lies they told in a week (e.g., DePaulo et al., 1996) Page | 63 questionnaire about how often they had lied in the previous 24 hours (e.g., Serota, Levine, & Boster, 2010). Telling lies in unsanctioned deception has already taken place, and does not share the same limitations as sanctioned deception. ...
Prior studies on online deception detection have investigated language differences between deceivers and truth-tellers to fight against online crimes, such as scams and other online threats. As most studies derived from western contexts have centered on English language, the deceptive features identified may not be applicable to other languages, like Thai. This research, therefore, aims to examine whether these features previously identified as associated with deception in English could be applied to Thai, and whether there are any Thai specific features that might be associated with online deception. The experimental study was conducted in a controlled computer laboratory in a Thai state university. The participants were ninety-six native speakers of Thai in tertiary level. All participants were led to isolated computer terminals in the computer room where they would perform the task. These participants were randomly divided into two groups: 48 chatters and 48 naïve partners. For chatters, they were informed of the true objective of the study, namely, to examine the differences in language use between deceivers and truth-tellers in Thai online chat, so they needed to switch their roles either as truth-tellers (providing only truthful information) or deceivers (convincing their chat partners that they were someone else) in Thai online chat. The other group members as naïve partners were informed that they would participate in two chat sessions with unknown partners to investigate the use of Thai chat language on MSN Messenger. In the first 10-minute chat sessions, naïve partners were paired up with either truth-tellers or deceivers and switched their chat partners in the second 10-minute chat session. The investigated features used by truth-tellers and deceivers were then analyzed by using statistical analyses: Percentage Differences and Chi-square test. Features with no sub-categories (e.g., participation and demonstratives) were calculated by percentage differences with an arbitrary cut point of fifteen percentage points to identify the noticeable differences between these two groups. Features with sub-categories (e.g., nouns and verbs) were calculated by using chi-square test to investigate whether there were statistically significant differences in the use of features by truth-tellers and deceivers in Thai online chat. The results show noticeable differences in features with no sub-categories in initiation, opening new topics, spontaneous correction, demonstratives, and frequency of self-references used by deceivers and truth-tellers in Thai online chat. There were also statistically significant differences in features with subcategories in the use of modal verbs, self-references types, and particles between truth-tellers and deceivers at .01 level (p < .01). To sum up, the findings of the study suggest that most deceptive features, previously identified in English, cannot be transferrable to Thai, however, two Thai-specific indicators (the use of self-references and particles) were also identified as associated with deception in Thai online chat. In other words, the more deceivers manipulated their gender identity, the more they used first personal pronoun and gender particles as a deception strategy when pretending to be someone else in Thai online chat. The findings may be usefully applied in automated online deception detection for Thai and also highlight the need to supplement linguistic investigations of English with investigations of other disparate languages.
Contrary to popular belief, research indicates that individuals in the general population are poor at detecting truthfulness or deception using facial cues (Stel & van Dijk. Social Influence, 13(3):137–149, 2018). We also tend to have a truth-bias, where we judge more truths as truths than lies as lies (Baker et al. Legal and Criminological Psychology, 18(2):300–313, 2013; Bond & DePaulo. Personality and Social Psychology Review, 10(3):214–234, 2006). Successful deception detection is often based on multiple sources of information, and it may take days, weeks, or months to draw conclusions about others’ truthfulness (Park et al. Communication Monographs, 69(2):144–157, 2002). Educators encounter numerous types of deception, and the accuracy of lie detection could be quite useful when investigating allegations of academic misconduct. Findings from deception research suggest that discovery interviews used in academic misconduct cases may not be worthwhile unless educators and administrators are specifically trained to detect deception (Driskell. Psychology, Crime & Law, 18(8):713–731, 2012) and examine multiple sources of evidence before coming to conclusions (Ellis et al. Technology, policy and research: Establishing evidentiary standards for contract cheating cases. In T. Bretag (Ed.), A research agenda for academic integrity (pp. 138–151). Edward Elgar, 2020). This chapter summarizes the peer-reviewed research literature on detecting deception and outlines the implications for investigations of academic cheating.
As concerns grow regarding the deceptive nature of media content marketed to children, the demand for exploration into portrayals of deception in children’s programming content simultaneously increases. The following article uses multiple psychology and media studies theories to analyze both animated and live-action series targeted to children through Netflix Kids for their depictions of deception in frequency and evaluation. Content analyses reveal high levels of deception frequency and positive evaluations. Supplemental textual analysis demonstrates the appearance of deception in children’s television programming as frequent, insignificant and justified – illustrating further need to explore the implications of deception portrayals on children.
Full-text available
Unlike prior research that treats social‐desirability bias (SDB) as measure contamination, the present research asserts that significant associations between measures of SDB and value self‐reports are evidence of measure validity. The degree to which value self‐reports are influenced by SDB also reflects the relative importance of values within a culture. Values that are most important have the greatest self‐presentational implications and therefore should be most affected by SDB. Moreover, differences between raw and SDB‐corrected value self‐reports indicate the extent to which values are personal (i.e., private) or public in nature. The research is based on two national samples of American adults 18 years of age and older. Implications for research on values are discussed. © 2000 John Wiley & Sons, Inc.
Full-text available
Developed, on the basis of responses from 608 undergraduate students to the 33-item Marlowe-Crowne Social Desirability Scale, three short forms of 11, 12, and 13 items. The psychometric characteristics of these three forms and three other short forms developed by Strahan and Gerbasi (1972) were investigated and comparisons made. Results, in the form of internal consistency reliability, item factor loadings, short form with Marlowe-Crowne total scale correlations, and correlations between Marlowe-Crowne short forms and the Edwards Social Desirability Scale, indicate that psychometrically sound short forms can be constructed. Comparisons made between the short forms examined in this investigation suggest the 13-item form as a viable substitute for the regular 33-item Marlowe-Crowne scale.
Full-text available
This study explores sex differences in perceptions of discovered deception, and the subsequent emotional reactions that are experienced by relational partners. Drawing upon research examining deception, relational communication, and gender, several hypotheses were developed and tested in a sample of 190 respondents who had recently discovered the lie of a friend or romantic partner. The data were consistent with the hypotheses. Women were more likely than men to rate lying as an unacceptable form of behavior within both friendship and romantic relationships. In addition, women rated the act of lying (regardless of what was lied about) as more significant, and reported more negative emotional reactions upon discovering deception than did men. Generalized communicative suspicion functioned to enhance the intensity of emotional reactions for women, but not for men. Implications of the current results for the study of deception are discussed.
Opinion research is frequently carried out through the Internet and a further increase can be expected. The article focuses on the online access panel, in which respondents are previously recruited through non-probability methods. Despite substantial time- and cost-reduction, on- line access panel research mainly has to cope with limited Internet coverage and self-selection in the recruitment phase of new panel members. The article investigates whether frequently ap- plied weighting procedures, based on poststratification variables and propensity scores, make online access panel data more representative of the general population. To address this issue, the answers to identical questions are compared between an online self-administered survey of previously recruited online access panel respondents and a face-to-face survey of randomly sampled respondents of the general population. Both respondent groups were surveyed at a similar moment in time (2006-2007) in the same geographical region (Flanders, Belgium). The findings reveal many significant differences, regarding sociodemographic characteristics as well as attitudes towards work, politics and immigrants. The results can be explained by both the specific characteristics of the respondent groups and mode effects. Weighting adjustment had only a minor impact on the results and did not eliminate the differences.
Systems as diverse as genetic networks or the World Wide Web are best described as networks with complex topology. A common property of many large networks is that the vertex connectivities follow a scale-free power-law distribution. This feature was found to be a consequence of two generic mech-anisms: (i) networks expand continuously by the addition of new vertices, and (ii) new vertices attach preferentially to sites that are already well connected. A model based on these two ingredients reproduces the observed stationary scale-free distributions, which indicates that the development of large networks is governed by robust self-organizing phenomena that go beyond the particulars of the individual systems.
This study examined the effects of self-presentation goals on the amount and type of verbal de- ception used by participants in same-gender and mixed-gender dyads. Participants were asked to engage in a conversation that was secretly videotaped. Self-presentational goal was manipu- lated, where one member of the dyad (the self-presenter) was told to either appear (a) likable, (b) competent, or (c) was told to simply get to know his or her partner (control condition). After the conversation, self-presenters were asked to review a video recording of the interaction and iden- tify the instances in which they had deceived the other person. Overall, participants told more lies when they had a goal to appear likable or competent compared to participants in the control condition, and the content of the lies varied according to self-presentation goal. In addition, lies told by men and women differed in content, although not in quantity.