ArticlePDF Available

Abstract and Figures

Evaluating other people's sincerity is a ubiquitous and important part of social interactions. Fourteen experiments (total N = 7,565; 10 preregistered; 11 in the main article, three in the online supplemental materials; with U.S. American and British members of the public, and French students) show that response speed is an important cue on which people base their sincerity inferences. Specifically, people systematically judged slower (vs. faster) responses as less sincere for a range of scenarios from trivial daily conversations to high stakes situations such as police interrogations. Our findings suggest that this is because slower responses are perceived to be the result of the responder suppressing automatic, truthful thoughts, and fabricating a novel answer. People also seem to have a rich lay theory of response speed, which takes into account a variety of situational factors. For instance, the effect of response delay on perceived sincerity is smaller if the response is socially undesirable, or if it can be attributed to mental effort. Finally, we showed that explicit instructions to ignore response speed can reduce the effect of response speed on judgments on sincerity. Our findings not only help ascertain the role of response speed in interpersonal inference making processes, but also carry important practical implication. In particular, the present study highlights the potential effects that may be observed in judicial settings, because the response speed of innocent suspects may mislead people to judge them as insincere and hence guilty. (PsycInfo Database Record (c) 2021 APA, all rights reserved).
Content may be subject to copyright.
Slow Lies: Response Delays Promote Perceptions of Insincerity
Ignazio Ziano
1
and Deming Wang
2
1
Department of Marketing, Grenoble Ecole de Management
2
School of Psychology, James Cook University
Evaluating other people’s sincerity is a ubiquitous and important part of social interactions. Fourteen
experiments (total N7,565; 10 preregistered; 11 in the main article, three in the online supplemental
materials; with U.S. American and British members of the public, and French students) show that
response speed is an important cue on which people base their sincerity inferences. Specifically, people
systematically judged slower (vs. faster) responses as less sincere for a range of scenarios from trivial
daily conversations to high stakes situations such as police interrogations. Our findings suggest that this
is because slower responses are perceived to be the result of the responder suppressing automatic, truthful
thoughts, and fabricating a novel answer. People also seem to have a rich lay theory of response speed,
which takes into account a variety of situational factors. For instance, the effect of response delay on
perceived sincerity is smaller if the response is socially undesirable, or if it can be attributed to mental
effort. Finally, we showed that explicit instructions to ignore response speed can reduce the effect of
response speed on judgments on sincerity. Our findings not only help ascertain the role of response speed
in interpersonal inference making processes, but also carry important practical implication. In particular,
the present study highlights the potential effects that may be observed in judicial settings, because the
response speed of innocent suspects may mislead people to judge them as insincere and hence guilty.
Keywords: response speed, perceived sincerity, thought suppression, answer fabrication, impression
formation
Supplemental materials: https://doi.org/10.1037/pspa0000250.supp
Imagine asking a friend for their thoughts on a dish you cooked
and being met with immediate praise of your culinary skills. Now
imagine the same scenario with one difference, the response was
delivered after a slight delay. While in everyday life, communica-
tors’ sincerity and truthfulness can be inferred from what they say
(Rosenblum et al., 2020) and how they say it (Hornsey et al., 2020)
the aforementioned scenario seems to suggest that it somehow also
depends on when they say it. In the present study, we asked the
question whether and why slower, slower responses can be per-
ceived as less sincere.
1
Much like the loudness and pitch of people’s response (Fish et
al., 2017) and the genuineness of their smile (Bernstein et al.,
2010), if response speed is indeed one of the many ubiquitous
social cues that people base their inferences on, then it would be
imperative to establish the nature of this phenomenon and its
underlying mechanism. The reason for this is that this effect, if
present, may manifest without people’s awareness, and across
many high stakes situations, such as those involving business deals
and legal processes. While it may be accurate that slower re-
sponses are less sincere, on average, when the situation leaves
room for doubt or when the situation is highly important, “accurate
on average” is not enough. If outcomes of these high stakes
scenarios are contingent on individuals’ automatic perceptions of
the responder’s response speed, then it makes sense for people to
be at least aware of such tendencies to minimize such tendencies
in decision making.
Why would slower responses be perceived as less sincere? To
answer this question, it is important to first highlight whether
responses could be less sincere in actuality, according to existing
research.
Response Speed and Actual Sincerity
Combined with physiological measures such as galvanic skin
response, response speed (often indicated as “reaction time”) has
long been a part of “lie detector” tests used in psychological
1
In this article, we use “response speed” to indicate the length of delay
between a question and an answer, as it has been done in some recent
research (Efendic´ et al., 2020; Van de Calseyde et al., 2014), rather than
speech rate (how many words per minute are spoken by the actor).We
strove to keep to this terminology throughout this article. Further, we use
the terms “slower responses” to indicated delayed responses, and “faster
responses” to indicate immediate responses.
Ignazio Ziano Xhttps://orcid.org/0000-0002-4957-3614
Deming Wang Xhttps://orcid.org/0000-0002-7325-9924
The findings of the present study were partially presented at the Society
for Consumer Psychology Conference, March 2020. We thank Tony Evans
for his feedback on an earlier version of this article. Data, analyses, and
materials are available at https://osf.io/pqmz2/.
Correspondence concerning this article should be addressed to Ignazio
Ziano, Department of Marketing, Grenoble Ecole de Management, 12 Rue
Pierre Semard, F-38000 Grenoble, France. Email: Ignazio.ziano@grenoble-
em.com
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Personality and Social Psychology:
Attitudes and Social Cognition
© 2021 American Psychological Association 2021, Vol. 2, No. 999, 000
ISSN: 0022-3514 https://doi.org/10.1037/pspa0000250
1
research such as the Concealed Information Test (Lykken, 1959;
Verschuere et al., 2010). People seem to be slower responders
when they lie compared with when they are honest, both in
controlled laboratory studies (Seymour & Kerlin, 2008; Seymour
et al., 2000; Suchotzki et al., 2017) and in the field (Mann et al.,
2002; Vrij & Mann, 2001).
Researchers have attributed response speed during deception to
the higher levels of cognitive load that typically accompany at-
tempts to deceive (vs. truth telling), such as working memory load
and thought inhibition (Miyake et al., 2000). Meta-analytic neu-
ropsychological evidence (Christ et al., 2009) suggests that lying
taxes working memory, because people have to hold more infor-
mation in their mind when lying (both the truth and the fabricated
version) compared with when telling the truth, activating prefron-
tal brain regions and causing slower responses. This mechanism
may be further exacerbated by thought inhibition processes be-
cause people need to take the additional step of inhibiting the
spontaneous, raw, and truthful response before generating their
actual response (Spence et al., 2001). The notion that lying takes
more time than truth-telling is also supported by findings from
social cognition research, which suggests that actual experiences
and genuine opinions are more accessible in memory and more
readily activated (Meyer & Schvaneveldt, 1976) compared with
those that are fabricated (Miyake et al., 2000). This ease of
retrieval is then reflected by faster response speed and reduced
delay. When lying, on the other hand, responders not only have to
suppress the spontaneous and truthful thought activated by a
question, but also generate an alternative answer that matches their
motive, thereby delaying the response (Duran et al., 2010; Sey-
mour & Schumacher, 2009; Suchotzki et al., 2015; Walczyk et al.,
2003). Despite the abundance of research examining whether
people are in fact slower to respond when the response is a
dishonest one, research to date has not given a clear-cut answer as
to whether response speed affects observers’ perceptions of re-
sponder sincerity.
Response Speed and Perceived Sincerity
Given the ubiquity and importance of sincerity inferences and
lie detection in everyday communication, there has been consid-
erable scholarly interest on the relationship between perceived
sincerity and a host of other variables, including response speed
(sometimes also referred to as “response latency” or “response
delay” in the literature; e.g., Baskett & Freedle, 1974; Hartwig &
Bond, 2011; Kraut, 1978). For instance, one meta-analytical study
provided evidence of a weak relationship between response speed
and perceived sincerity (r.18; Cohen, 1992), hinting that slower
responses may be perceived as less sincere (Hartwig & Bond,
2011). However, a closer inspection of their meta-analyzed studies
reveals substantial heterogeneity in both the direction and magni-
tude of the relationship. While some studies find that slower
responses are perceived as insincere (Vrij et al., 2001), others find
no statistically significant effect in either direction (Vrij et al.,
2004; Zuckerman et al., 1981), some find a curvilinear relationship
between response speed and sincerity (Baskett & Freedle, 1974;
Boltz, 2005), and some even find that slower responses are per-
ceived as more sincere than faster ones (Kraut & Lewis, 1982).
We believe the aforementioned inconsistent findings are the
result of various limitations. Methodologically, many of these
studies are exploratory and correlational in nature (e.g., Kraut,
1978; Kraut & Lewis, 1982) where participants are exposed to the
same material (such as a brief excerpt of a conversation from a
cassette tape or a short video), and then asked to rate it on several
dimensions that are then correlated. Such correlational designs
make it difficult to conclude whether there is a directional and
causal effect of response speed on sincerity judgments, and are
also prone to effects of confounding variables. Even experimental
studies suffered from the confounding problem because they
mostly measured response speed and perceived sincerity as depen-
dent variables following the manipulation of a different indepen-
dent variable. Most notably, studies on deception intention would
have one group of actors tell the truth and another group lie.
Participants would then be asked to rate the actors on a variety of
measures including response speed and perceived sincerity (Buller
et al., 1991; Cornetto, 2002; Harrison et al., 1978; Stiff & Miller,
1986; Swinkels, 1991). This type of study design again precludes
causal inferences to be drawn between response speed and per-
ceived sincerity.
The uncertainty of the relationship between response speed and
perceived sincerity is exacerbated further by the low statistical
power of many studies in this domain. For instance, Kraut and
Lewis (1982) recruited 96 participants, which implies only 50%
power of detecting an effect close to the average in social psy-
chology (Cohen’s d0.40 or Pearson’s r.20; Richard et al.,
2003) with a typical two-tailed alpha level of 5%; Vrij et al. (2001)
recruited 39 participants, which implies 24% power of detecting
r.20, both far short of the recommended 80% power level.
Many studies measuring response speed and perceived sincerity
also measure and correlate with each other multiple other aspects
connected to perceptions of verbal and nonverbal behavior, some-
times more than 15 (e.g., Kraut & Lewis, 1982; Zuckerman et al.,
1981), inflating the likelihood of false positives.
Taken together, previous research serves as a valuable starting
point for the present investigation as it provides indirect evidence
for, and hence a basis on which we can predict, the causal role of
response speed in sincerity judgments. It also highlights the ne-
cessity of the present study because low statistical power, con-
founding factors, and diversity of study designs has precluded
researchers from arriving at a conclusive answer as to the exis-
tence, nature, and causal direction of this effect. More important,
this lack of empirical consistency hinders theoretical developments
and hence meaningful investigations of the underlying mecha-
nisms and boundary conditions of this phenomenon. Most notably,
experimental studies thus far have yet to test the mechanism
underlying the relationship between response speed and perceived
sincerity in a direct manner (Boltz, 2005). A systematic investi-
gation into the nature and the mechanism underlying the relation-
ship between response speed and sincerity judgment is highly
necessary, for several reasons. First, to determine whether there
exist a causal effect of response speed on perceived sincerity, as
previous research provides inconclusive results. Second, to exam-
ine the direction of this effect, as, again, previous research points
toward several directions. Finally, this work delves deep into the
mechanisms and boundary conditions of the effect of response
speed on sincerity judgments, which previous research could not
investigate.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2ZIANO AND WANG
Are Slower Responses Perceived as Less Sincere?
The literature reviewed above provides an empirical basis for
the association between response speed and perceived sincerity,
but is there reason to believe that response speed has a specific
causal effect on perceived sincerity? We believe so. Response
speed is likely not only a reflection of actual sincerity, but also able
to influence observers’ perceptions of responder sincerity such that
slower responses are perceived as less sincere. The reason for this
is twofold. First, because lying and being lied to are both frequent
occurrences in people’s social lives, observers should have an
experience-informed lay intuition that mirrors the research find-
ings on deception and response speed: slower responses should be
perceived as less sincere than faster ones because people believe
that response speed could be a reflection of underlying cognitive
labor, such as the inhibition of an automatic and truthful response,
or the fabrication of an alternative one. That is, individuals may
feel as if slower responders are using the delay to produce a novel
and untruthful answer.
Second, it has been well-documented that people rely on re-
sponse speed as a social cue in a range of interpersonal judgment
and impression formation processes. For instance, response speed
is considered indicative of internal states and decisional readiness,
such that when responders are quicker to agree to morally dubious
requests, they are dealt a harsher moral judgment, and when the
decision is morally virtuous, faster responders are likely to be dealt
a more positive moral judgment (Critcher et al., 2013). A similar
mechanism seems to be at play in expectations of collaboration:
observers expect other participants who respond more quickly in
collaborative economic games to provide more extreme responses;
either very collaborative or very hostile (Evans & van de Calseyde,
2017). Faster responses are preferred in negotiations, when receiv-
ing job offers, and within client-customer relationships (Van de
Calseyde et al., 2014). Across these examples, people sometimes
demonstrated a preference for swift responses because faster re-
sponse speeds likely show greater willingness to engage and lower
levels of hesitation, indecision, and doubt on the part of the
responder (Van de Calseyde et al., 2014). Taken together, people
seem to rely on response speed in a plethora of interpersonal
judgment processes, and seem to have a rather sophisticated set of
intuition relating to what response speed reflects in various social
settings. This allows us to reason that sincerity could be yet
another facet that people infer from others based on the social cue
of response speed.
Boundary Conditions
We do not expect the effect of response speed on sincerity
judgments to be universal. In fact, we expect that the effect of
response speed on sincerity judgments may be attenuated if the
delay can be attributed to factors other than thought suppression or
answer fabrication. Further, we expect that the effect of response
speed on sincerity judgments may be stronger for socially desir-
able responses.
Delay Attribution
Our belief that slower responses are perceived to be less sincere
than faster responses is predicated on the premise that the response
delay is attributed to a cognitive process that is related to lying,
such as thought suppression or answer fabrication. However, dur-
ing social interactions, response delays can also be attributed to
factors other than thought suppression or answer fabrication, such
as mental effort. This is supported by research associating response
speeds with mental effort. For instance, it has been shown that
observers believe recommendations provided more slowly to have
involved a higher degree of mental effort (Efendic´ et al., 2020). As
a result, not only can response speed sometimes be attributed to
mental effort, when they are, they can also be evaluated more
positively. For example, it has been found that difficult decisions
are evaluated more favorably when they are made after a longer
(vs. shorter) delay (Kupor et al., 2014), although the opposite
happens for easy decisions. These findings underscore the poten-
tial role played by contextual factors such as response type and
perceived mental effort when people are making inferences from
response speed. Given this, and given that thought suppression and
answer fabrication attribution are the proposed mechanisms un-
derlying the prediction that slow responses are perceived as insin-
cere, we expect that the main effect may be weakened if response
delays are attributed to sincerity-unrelated reasons. Specifically, if
people have reason to believe that the source of the response delay
is mental effort rather than thought suppression or answer fabri-
cation, they may be less likely to evaluate slower responses as less
sincere. For instance, if observers believe that the question relates
to an event from the distant past, they will be more likely to
attribute a slower response speed to memory effort compared with
if the question pertains to a recent event.
Response Social Desirability
In addition to response delay attribution, we believe that the
social desirability of a response may also affect the degree to
which slower responses are seen as less sincere. This is because it
is in people’s interest to project themselves in a socially desirable
(rather than undesirable) light. For instance, people are more likely
to downplay negative attitudes and values, such as how much they
resent being asked a favor (Crowne & Marlowe, 1960), and their
materialistic values (Mick, 1996; Van Boven et al., 2010), rather
than positive ones. Consequently, the response speed of a socially
desirable response may be attributed to the inhibition of an au-
thentic and “raw,” but socially undesirable response, as the re-
sponder needs time to decorate or “edit” the response in a socially
desirable way (Holtgraves, 2004), a process similar in nature to
answer fabrication, that is, the addition of untrue details to an
answer. However, when the response itself is socially undesirable
to begin with, people may be less likely to believe that any form
of response editing and inhibition was implemented. As such,
socially undesirable responses should be more likely to be per-
ceived as sincere, regardless of response speed. This is also con-
sistent with empirical evidence suggesting that people who express
themselves in a socially undesirable manner are ironically judged
to be more sincere. For instance, while swearing is considered rude
and even immoral (Jay, 2009), witnesses who use more swear
words are considered more credible (Rassin & Van Der Heijden,
2005). Moreover, while politically incorrect language may be
considered disrespectful to certain subgroups of the population,
people using it are perceived as more authentic (Rosenblum et al.,
2020). Taken together, there is reason to believe that the effect of
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
SLOW LIES
response speed on perceived sincerity may be reduced when the
response is socially undesirable.
Automaticity
To facilitate completeness of the proposed theoretical model, in
addition to establishing the basic phenomenon, its mechanism, and
boundary conditions, the present research sought to answer a final
question: can people voluntarily ignore response speed when making
sincerity inferences if they are told to? Research has shown that
people can be persuaded to discount irrelevant or invalid information
in mock jury tasks, such as information gathered without a warrant
(Dietvorst & Simonsohn, 2019; Kassin & Sommers, 1997). This
suggests that people can indeed control what sources of evidence they
allow themselves to be influenced by in the judgment process. How-
ever, response speed is not material information such as a testimony
or a piece of evidence that jurors can simply disregard. Rather, it is an
ever-present and integral feature of social interactions in general, and
hence, likely inextricably intertwined with various forms of evidence
such as interrogation videos. As such, compared with invalid evi-
dence, response speed may be a factor that is more difficult to
disentangle and ignore in the judgment process.
Study Overview
To test the notion that slower responses are perceived as less
sincere, its proposed psychological mechanisms, and its boundary
conditions, we conducted 14 experiments, summarized in Table 1 (11
presented in the main article, and three in the online supplemental
materials). In Study 1a, using audio stimuli, we tested whether slower
responses are considered less sincere, across several response speeds
(from zero to 10 s), actors, and scenarios. In Study 1b, using video
stimuli, we tested whether people believe that slower responses are
less sincere than faster ones, and are more likely to judge slower
responders as guilty of the crime of which they are accused. Study 1c
sought to replicate these results using vignette scenarios. In Study 2a,
we tested whether the effect is mediated by thought suppression
inferences in a between-subjects design. In Study 2b, we tested
whether slower responses are perceived as less sincere across a greater
variety of statements, and whether the effect is mediated by thought
suppression inferences in a within-subjects design. In Study 2c, we
independently manipulated thought suppression inferences without
manipulating response speed, to test their unique causal effect on
sincerity perceptions. This way, Studies 2a–2c elucidate one pro-
cess underlying response speed and sincerity judgments using a
“causal chain” approach. Studies 3a and 3b investigated a
second mediator, answer fabrication inferences, again by using
a causal chain approach. Study 3a used statistical mediation
tests and Study 3b manipulated answer fabrication inferences
alone and tested their unique causal effect on sincerity judg-
ments. In Study 4, we investigated whether the effect of re-
sponse speed on sincerity judgments is reduced when the re-
sponse is socially undesirable. Study 5 explored the role of
cognitive process attribution: when the question pertains to an
event from the distant past, the effect may be attenuated,
Table 1
The Effect of Response Speed on Sincerity Judgments, Study Overview
Study Sample size
(sample source) Result Moderator
Study 1a 1,197 (MTurk) Slower responses are perceived as less sincere, audio stimuli
Study 1b 605 (MTurk) Slower responders are perceived as less sincere and more likely to be
judged as guilty, video stimuli
Study 1c 584 (MTurk) Slower responders are perceived as less sincere, vignette scenario
stimuli
Study 2a 199 (Prolific) Thought suppressions inferences mediate the effect of response speed on
sincerity judgment, between-subjects
Study 2b 600 (MTurk) Thought suppressions inferences mediate the effect of response speed on
sincerity judgment, within-subjects Opinion share of response
Study 2c 303 (Prolific) Causal chain study: Thought suppression inferences promote perceived
insincerity
Study 3a 199 (MTurk) Answer fabrication inferences mediate the effect of response speed on
sincerity judgments
Study 3b 202 (MTurk) Causal chain study: Answer fabrication inferences promote perceived
insincerity
Study 3 588 (MTurk) The effect of response speed on sincerity judgment is weaker when the
response is socially undesirable Response social desirability
Study 4 800 (MTurk) The effect of response speed on sincerity judgment is weaker when the
response speed can be attributed to mental effort Perceived mental effort associated
with response; triviality of
event
Study 5 605 (MTurk) The effect of response speed on sincerity judgment is weaker when
people are instructed to ignore response speed in the judgment
process
Instruction to ignore the speed vs.
no instructions
Study S1 275 (French students) Conceptual replication of Study 1, with different scenarios, response
speeds, and with French students as participants
Study S2 604 (MTurk) Conceptual replication of Study 3, with stranger rather than friend as the
actor Response social desirability
Study S3 804 (MTurk) Conceptual replication of Study 3: Admission of guilt moderates effect Response social desirability
Note. MTurk Amazon Mechanical Turk. S1, S2, and S3 are included in the online supplemental materials.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4ZIANO AND WANG
because observers may be more likely to attribute slower re-
sponses to memory effort rather than thought suppression. In
Study 6, we examined the controllability of the predicted phe-
nomenon by testing whether the effect of response speed on
perceived sincerity can be mitigated when people are explicit
instructed to ignore response speed in the judgment process.
The online supplemental materials presents three additional
studies. Study S1 replicated the main effect and our mediation
results with French students. Studies S2 and S3 replicated Study
4 with different scenarios and an additional guilt judgment
measure to explore the downstream effects of the predicted
phenomenon.
To test the generalizability of the predicted phenomena, we at-
tempted to increase the diversity and inclusiveness of our samples in
terms of participant demographics. Specifically, we sampled members
of the public from the United States, the United Kingdom, as well as
French university students. Independent samples were used for each
study reported in the present article. Sample sizes were predetermined
for all studies. We did not have a strong a priori notion of the effect
size we would encounter. We sought to maximize power by always
using more than 100 participants per cell when we used a two-cells
design. This sample size implies 80% power of detecting an effect
size Cohen’s d0.40 (with a two-tailed ␣⫽.05), close the average
effect size in social psychology (Richard et al., 2003). Throughout all
the studies, we used an alpha level of 0.05 (5%), two-tailed. We used
larger sample sizes per cell when we conducted a study with more
than two cells or tested an interaction effect, as they both require a
higher sample size to keep the same statistical power to detect the
same effect size (Simonsohn, 2015). We also used attention checks in
some of the studies (especially those where participants needed to
have functioning earphones or headphones), to ensure that partici-
pants were attentive and followed instructions, and excluded from
analyses participants who failed them. Across studies, we used a
variety of audio, video, and vignette stimuli depicting a range of
scenarios to assuage concerns of stimuli selection bias (Judd et al.,
2012) and statistical power (Meyvis & Van Osselaer, 2018).
To assuage concerns for demand and conformity effects, we used
experimental procedures where participants had to spontaneously pay
attention to response speed, as it was not indicated by the experi-
menter where their focus should be. As such, participants could rely
on the actors’ voice tone, pitch, and pace in some studies (e.g., Study
1a), and even clothing, attractiveness, and appearance in others (e.g.,
Study 6) as contextual factors. Further, in all studies using vignette
scenarios, we ensured the provision of contextual elements such as
gender, occupation, age, names, and background of the scenario (e.g.,
Studies 2a and 2b), for the same realism reasons.
Raw data, analyses, complete stimuli and scenario, and prereg-
istrations are available at https://osf.io/pqmz2, Ziano (2020). Stud-
ies S1, S2, S3, additional statistical analyses, deviations from the
preregistrations, and participant compensation are available in the
online supplemental materials.
Study 1a—Inferring Sincerity From Response Speed
In Study 1a, we tested the prediction that slower responses are
considered less sincere than fast responses using realistic audio
stimuli. This study was preregistered at https://aspredicted.org/
s5yx5.pdf.
Method
Participants
We recruited 1,197 participants from Amazon Mechanical Turk
(MTurk; 727 males, 462 females, 8 preferred not to disclose;
M
age
37.99, SD
age
12.73). The HIT description explicitly
required earphones or headphones. To make sure that participants
were listening to the audio stimuli, a snippet in which the word
“Three” was spoken was presented to participants before they
commenced the task. Participants were asked whether they had
listened to the snippet (Yes/No), and in the next screen, they were
asked which word was spoken in the initial snippet (the choices
were “Four,” “Seven,” and “Three”). Sixty-four participants did
not answer the attention check correctly and were excluded from
subsequent analyses, leaving a final sample of 1,133 participants
(684 males, 441 females, 8 preferred not to disclose; M
age
37.97,
SD
age
12.35).
Procedure
Participants were randomly assigned to one of six conditions
(response speed; 0, 1, 2, 3, 5, 10 s). In each condition, they were
presented with four situations in randomized order—two with
male actors and two with female actors, and within each gender
category, two pertaining to a theft scenario and two pertaining
to a taste preference scenario. This was meant to maximize
generalizability across actor gender and scenario. Participants
in each condition were presented four short audio snippets to
listen to, each corresponding to one situation, and each includ-
ing a recorded question and answer. Participants were told that
each of the actors presented was either asked whether they had
liked a cake a friend made or whether they had stolen money
from the company in which they used to work. In each situation,
the question and the answer were exactly the same (either
replying “Yes I did” in the cake scenario or “No I didn’t” in the
money scenario). We first recorded sample stimuli (one per
scenario). Then, we edited each snippet by inserting silent
pauses of the desired duration depending on the condition to
keep features of both questions and answers (such as tone and
volume) constant.
After each audio snippet, participants were asked how sincere
the person in question was (our focal dependent variable), on a
scale from 1 (not at all)to7(very much). In the next screen,
participants rated the responder’s response speed (manipulation
check), on a scale from 1 (very slow)to7(very fast).
Results
Manipulation Check
A64 mixed measures analysis of variance (ANOVA) with
response speed as between-subjects factor and scenario as within-
subjects factors found a significant effect of response speed, F(5,
1127) 140.42, p.001,
2
0.315, indicating that slower
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
SLOW LIES
responses were perceived as such.
2
We also found a significant
effect of scenario, F(3, 3381) 13.22, p.001,
2
0.002,
indicating that the actors in some scenarios were considered faster
overall, and no significant scenario by speed interaction, F(15,
3381) 1.63, p.058,
2
0.001, giving no support to the
notion that the impact of response speed on perceived speed
significantly changed across scenarios. Overall, these analyses
show that slower responses were in fact perceived as slower than
faster responses.
Sincerity Judgments
A64 mixed measures ANOVA with response speed as
between-subjects factor and scenario as within-subjects factor
found a significant effect of response speed, F(5, 1127) 14.50,
p.001,
2
0.038, indicating that slower responses were
perceived as less sincere. We also found a significant effect of
scenario, F(3, 3381) 133.86, p.001,
2
0.040, indicating
that actors in some scenarios were considered significantly more
sincere than in others, and no significant scenario by response
speed interaction, F(15, 3381) 1.46, p.11,
2
0.002, giving
no support to the notion that the effect of response speed on
sincerity judgments systematically varied across scenarios. Over-
all, these analyses show that slower responses were perceived as
less sincere than faster responses. Tukey-corrected post hoc tests
(we used this correction for post hoc tests in all studies) comparing
each of the response speed condition to each other (see Table 2)
show that, compared with an immediate response, as little as a 2-s
delay is sufficient to produce a sizable and statistically significant
decrease in sincerity judgments. The relationship between re-
sponse speed and sincerity judgments (graphically presented in
Figure 1) seems relatively linear, plateauing aftera5sdelay.
Discussion
Study 1a provides initial evidence for our main proposition;
slower responses are perceived as less sincere. We used realistic
stimuli (audio recordings) to assuage concerns of demand effects,
as participants had to spontaneously notice response speed when
listening to the snippet, and it was not indicated by the experi-
menter. These results are in contrast with some prior literature,
which found that, compared with extremely slow and extremely
fast responses, intermediate responses were considered more sin-
cere (Baskett & Freedle, 1974; Boltz, 2005; but see Boltz et al.,
2010, for a failed replication of the curvilinear effect). However,
compared with those studies, we used a larger variety of scenarios,
and had higher statistical power.
Study 1b—Inferring Sincerity From Response Speed
in Video
In Study 1b, we aimed to test whether the effect of response
speed on sincerity judgments resulted in downstream social ef-
fects, specifically, whether slower responders who are judged as
insincere are consequently more likely to be judged as guilty of an
accused crime. A secondary aim of Study 1b was to test whether
results of Study 1a can be replicated using video stimuli. This
study was preregistered at https://aspredicted.org/7pk6d.pdf.
Method
Participants
We recruited 605 participants on MTurk (291 males, 314 fe-
males; M
age
37.70, SD
age
12.62). Earphones or headphones
were required for this task. We used two attention check to ensure
that participants were attentive to the presented stimuli. At the
beginning of the task, participants were shown a video in which a
hand with two fingers was shown and the word “cabbage” was
spoken. In the next screen, participants were asked which word
was spoken in the video (among “Tomato,” “Potato,” and “Cab-
bage”), and how many fingers were shown in the video (“Four,”
“Seven,” and “Two”). Five-hundred and 62 participants correctly
answered both attention checks (270 males, 292 females; M
age
37.88, SD
age
12.54) and their data was retained for analysis.
Procedure
Participants were randomized across two conditions, fast and
slow. In each condition, they were shown two videos (in random-
ized order, of a man and a woman), in which they received the
description that this was a police interrogation and that the person
in the video was accused to having stolen a few thousand dollars
from the place they worked. In the fast condition, after the question
of the actor playing the police officer (“Did you steal the
money?”), the actor playing the suspect immediately replied “No,
I didn’t!”, whereas in the slow condition, the suspect replied after
a delay of about 5 s.
After each video, participants completed four questions: their
sincerity judgment about the statement (“What do you think about
the statement?”) “Sincere”; “Truthful,” both anchored at 1 (not at
all)and7(very much), further averaged in two Sincerity measures,
Cronbach’s woman .97; Cronbach’s man .97), and a guilt
judgment (“Do you think this person is guilty?”, with a “Yes/No”
answer). Finally, they completed a manipulation check measuring
perceived response speed (“How was the response?”), anchored at
1(very slow)to7(very fast).
Results
Manipulation Check
A mixed-model ANOVA with perceived response speed as the
dependent variable, actor gender as the within-subjects factor and
response speed as the between-subjects factor showed a significant
main effect of speed, as hypothesized, F(1, 560) 756.629, p
.001,
2
.404. On average, participants considered slower re-
sponders (M1.73, SD 0.93) as slower than faster responders
(M5.33, SD 1.13), d3.49, 95% confidence interval (CI)
[3.23, 3.75]. Further, we found a significant effect of actor gender,
2
The analyses pertaining to the manipulation check and sincerity judg-
ments are slightly different from the preregistered analyses. In the prereg-
istered analyses we included response speed, actor gender and scenario as
factors in a 6 22 ANOVA. In hindsight, we realized we could not test
for gender, because we did not match the male and female actor for features
such as status and attractiveness that may influence sincerity judgments,
but included several scenarios for generalizability purposes. The main
effect of response speed on the manipulation check and sincerity judgments
is exactly the same across these and the preregistered analyses (reported in
detail in the online supplemental materials).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6ZIANO AND WANG
F(1, 560) 33.72, p.001,
2
.013, indicating that the female
actor was perceived to be faster to respond; and a significant
interaction between response speed and actor gender, F(1, 560)
191.634, p.001,
2
.072, indicating that the effect of response
speed on perceived response speed was relatively stronger for the
male actor, yet strong and significant for both actors.
3
Descriptive
statistics and effect sizes per gender are presented in Table 3.
Sincerity Judgments
A mixed-model ANOVA with perceived sincerity as the depen-
dent variable, actor gender as the within-subjects factor and re-
sponse speed as the between-subjects factor showed a significant
main effect of response speed, as hypothesized, F(1, 560)
172.37, p.001,
2
.177. On average, participants considered
the slower responder (M2.44, SD 1.23) as less sincere than
the faster responder (M3.84, SD 1.30), d1.08, 95% CI
[0.90, 1.26]. Further, we found a significant effect of actor gender,
F(1, 560) 17.40, p.001,
2
.007, indicating that the female
actor was perceived as more sincere in general; and no significant
interaction between response speed and actor gender, F(1, 560)
.02, p.89,
2
.001. Descriptive statistics and effect sizes per
actor gender are presented in Table 3.
Guilt Judgment
A repeated-measures logistic regression with gender as a within-
subjects factor and response speed as between-subjects factor
showed a significant effect of response speed,
2
(1) 122.52, p
.001, as hypothesized. In total, participants considered the faster
actor guilty 40% of the time (223/558 of choices), and the slower
actor guilty 73% of the time (416/566 of total choices). Further, we
found a significant effect of actor gender,
2
(1) 9.70, p.002,
indicating that overall, the woman was less likely to be considered
guilty compared the man; and no significant actor gender by
response speed mixed interaction,
2
(1) .001, p.99. Descrip-
tive statistics and effect sizes per actor gender are presented in
Table 3.
Discussion
Study 1b provides additional evidence for our central claim.
Using video stimuli, we showed that response speed not only
influences sincerity inferences, but also has downstream effects on
perceptions of guilt in interrogation contexts. Actors responding
3
In this study, we found an effect of gender and a stronger effect of
response speed for the male actor. However, we caution against over
interpreting these results as we presented different actors in different
scenarios for generalizability purposes, not to test the effect of gender per
se, which would require more accurate matching of the male and the female
actors and scenarios.
Table 2
Tukey-Corrected Post Hoc Comparisons Testing the Effect of Response Speed on Sincerity
Judgments, Study 1a
Comparison
Delay (seconds) Mean difference SE df t value p
tukey
0 — 1 0.101 0.132 1127.000 0.767 0.973
2 0.414 0.129 1127.000 3.203 0.018
3 0.489 0.130 1127.000 3.765 0.002
5 0.829 0.131 1127.000 6.302 .001
— 10 0.848 0.131 1127.000 6.469 .001
1 — 2 0.312 0.130 1127.000 2.396 0.158
3 0.387 0.131 1127.000 2.956 0.037
5 0.727 0.133 1127.000 5.481 .001
— 10 0.747 0.132 1127.000 5.644 .001
2 — 3 0.075 0.128 1127.000 0.584 0.992
5 0.415 0.130 1127.000 3.192 0.018
— 10 0.435 0.130 1127.000 3.353 0.011
3 — 5 0.340 0.131 1127.000 2.603 0.097
— 10 0.360 0.130 1127.000 2.761 0.065
5 — 10 0.020 0.132 1127.000 0.149 1.000
Figure 1
The Impact of Response Speed on Sincerity Judgments, Study 1a
Note. The black line indicates the “money” scenario and the gray line
indicates the “cake” scenario. Error bars indicate 1SEM.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
SLOW LIES
aftera5sdelay were perceived as significantly less sincere, and
consequently more likely to be guilty, compared with actors re-
sponding immediately.
Study 1c—Varying Response Speed in a Vignette
Scenario
Thus far, we have demonstrated the basic prediction that slower
responses are perceived to be less sincere using audio and video
stimuli. Study 1c aimed to replicate the basic effect using vignette
scenarios. This study was preregistered at https://aspredicted.org/
nd2nh.pdf.
Method
Participants
We recruited 584 participants from MTurk (372 males, 210
females, 2 preferred not to disclose; M
age
38.19, SD
age
12.11). Fifty participants replied “Yes” to the question “Have you
ever had a fatal heart attack?” (presented at the end of the survey)
and were excluded from analyses, leaving a valid sample of 534
participants (328 males, 204 females, 2 preferred not to disclose;
M
age
38.05, SD
age
11.44).
Procedure
We randomly assigned participants to one of three conditions
(immediate, 2 s delay, 5 s delay), where we varied the response
speed of the actor. Participants read the following scenario (in
brackets, the corresponding scripts for the 2 s delay and 5 s delay
conditions):
Maria is a woman in her 40s. She is asked by a friend if she liked the
movie they just saw together.
Maria immediately [waits 2 s/waits 5 s] replies: “Yes, I did!”
Participants then rated how sincere Maria was, on a scale
ranging from 1 (not at all)to7(very much). In the next and final
screen, participants completed a manipulation check by rating how
fast Maria’s response was, on a scale ranging from 1 (very slow)
to7(very fast).
Results
Manipulation Check
A one-way ANOVA found a significant effect of actual re-
sponse speed on perceived response speed, F(2, 531) 140.18,
p.001,
2
0.346. Tukey-corrected post hoc tests showed that
participants rated the responder as faster in the immediate condi-
tion (M6.20, SD 0.95) compared with the 2 s delay condition
(M4.37, SD 1.80), t(531) ⫽⫺11.59, p.001, d1.29,
95% CI [1.07, 1.51], and compared with the 5 s delay condition
(M3.55, SD 1.80), t(531) ⫽⫺16.06, p.001, d1.91,
95% CI [1.66, 2.16]. Further, participants rated the responder as
slower in the 5 s delay condition compared with the 2 s delay
condition, t(531) ⫽⫺4.87, p.001, d0.46, 95% CI [0.24,
0.68].
Sincerity Judgments
A one-way ANOVA with response speed as the between-
subjects factor found a significant effect of response speed on
sincerity judgments, F(2, 531) 36.55, p.001,
2
0.121.
Tukey-corrected post hoc tests showed that compared with the
immediate response condition (M5.93, SD 1.02), participants
rated the responder as less sincere in the 2-s delay condition (M
5.10, SD 1.40), t(531) ⫽⫺5.97, p.001, d0.68, 95% CI
[0.47, 0.88], and in the 5-s delay condition (M4.75, SD 1.61),
t(531) ⫽⫺8.18, p.001, d0.90, 95% CI [0.68, 1.12].
Participants further thought that the responder was less sincere in
the 5-s delay compared with the 2-s delay, t(531) ⫽⫺2.42, p
.042, d0.24, 95% CI [0.03, 0.45].
Discussion
This study shows that responses are considered less sincere if
delivered after a delay as brief as 2 s, and that responses
delivered aftera5sdelay were seen as even more insincere.
These results not only consolidate the findings of Studies 1a
and 1b, but also provide generality to our effect through the use
of vignette scenario stimuli. This is important because it sug-
gests that people are indeed sensitive to response speed as a cue
on which to base their sincerity inferences, and that the findings
of Studies 1a and 1b were not simply because of the conspic-
uous nature of the response delays. Study S1 (online supple-
mental materials) replicates the main effect of response speed
on sincerity judgments with a French student sample, confirm-
ing its generalizability to a different country and population. In
the next three studies (2a, 2b, and 2c), we sought to systemat-
ically test the mediational role of thought suppression infer-
ences in the causal relationship between response speed and
sincerity perception using a causal chain approach. We did this
by using a between-subjects mediation analysis (Study 2a),
within-subjects mediation analysis (Study 2b), and a sole ma-
nipulation of the mediator (Study 2c).
Table 3
The Effect of Response Speed on Sincerity and Guilt Judgment, Study 1b
Male actor Female actor
Variable Fast M(SD) Slow M(SD) Effect size (95% CI) Fast M(SD) Slow M(SD) Effect size (95% CI)
Speed (manipulation check) 5.52 (1.41) 1.76 (1.07) d3.00 [2.76, 3.24] 5.15 (1.44) 1.76 (1.07) d2.74 [2.51. 2.97]
Sincerity judgments 3.70 (1.61) 2.30 (1.29) d.96 [0.78, 1.13] 3.99 (1.63) 2.58 (1.49) d.91 [0.74, 1.08]
Guilt judgment 125/279 (45%) 219/283 (77%) ␸⫽.33
a
98/279 (35%) 197/283 (70%) ␸⫽.35
a
a
No accepted method to calculate 95% confidence interval (CI) around Cramer’s .
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8ZIANO AND WANG
Study 2a—Slower Responders Are Perceived as Less
Sincere Because Observers Infer Thought Suppression,
Between-Subjects
The objective of this study was to test whether thought suppres-
sion inferences mediate the effect of response speed on sincerity
judgments. This study was preregistered at https://aspredicted.org/
my9tr.pdf.
Method
Participants
We recruited 199 U.K. nationals from Prolific. One participant
failed the attention check at the end of the survey, replying, “Yes”
to the question “Have you ever been to your own funeral?” which
left a final sample of 198 participants (92 males, 104 females, 2
other; M
age
36.30, SD
age
13.47).
Procedure
Participants were randomly assigned to one of two conditions,
fast and slow. In both conditions, they read about a woman named
Michelle, who responded either immediately or after2stoa
friend’s question. Participants read the following scenario (in
brackets, the slow condition):
Michelle is a woman in her 30s. She works as a copy-writer in New
York. She is tall and dark-haired.
One night, Michelle goes to the movies with a friend, to watch a
movie they both have never seen before.
Michelle is asked by her friend if she likes the movie that they are
watching.
Michelle immediately (waits 2 s and) replies: “Yes, I do”
Participants were then asked to rate Michelle’s sincerity (“How
sincere was Michelle?”), whether they believed that Michelle was
suppressing another thought (“Do you think Michelle was sup-
pressing another thought?”), and, as a manipulation check, how
they judged the speed of Michelle’s response (“How fast was
Michelle in replying?”), all on items anchored at 1 (not at all) and
7(very much).
Results
Manipulation Check
Participants rated the delayed response (M3.79, SD 1.61)
as slower than the immediate response (M5.95, SD 1.31),
t(196) 10.41, p.001, d1.48.
Sincerity Judgments
Participants rated the delayed response (M4.35, SD 1.44),
as less sincere than the immediate response (M5.20, SD
1.38), t(196) 4.25, p.001, d0.61.
Thought Suppression Inferences
Participants were more likely to believe that Michelle was
suppressing another thought when she gave a delayed response
(M3.67, SD 1.60) compared with when she gave an imme-
diate response (M2.83, SD 1.40), t(196) 3.96, p.001,
d0.56.
Mediation Analysis
A mediation analysis conducted with PROCESS for SPSS v3.1
(5,000 bootstraps) Model 4, with response speed as the indepen-
dent variable, thought suppression inferences as the mediator, and
sincerity judgments as the dependent variable found a significant
indirect effect of response speed on sincerity judgments through
thought suppression inferences, ab (SE)0.37 (0.11), 95% CI
[0.17, 0.60].
Discussion
Results of Study 2a show that observers believe that responders
who responded after a delay are more likely to have suppressed
another thought. In turn, observers rate slower responses as less
sincere. This study provides evidence in line with the notion that
thought suppression inferences underlie the effect of response
speed on sincerity judgments.
Study 2b—Slower Responders Are Perceived as Less
Sincere Because Observers Infer Thought Suppression,
Within-Subjects
As with Study 2a, the primary goal of Study 2b was to test
whether people perceive slower responses to be less sincere be-
cause the response delay is attributed to the responder suppressing
another thought, this time using a within-subjects experimental
design. Study 2b also aimed to investigate whether the social
desirability of a response is a boundary parameter of the effect of
response speed on perceived sincerity. We predict that if the
response is socially undesirable (e.g., a minority opinion), the
effect of response speed on perceived sincerity will be attenuated,
as people often intentionally inhibit spontaneous responses and
deliberate their response to make it more, not less, socially desir-
able (Crowne & Marlowe, 1960). Consequently, socially undesir-
able responses may be considered more authentic, less deliberated,
and hence more sincere (Rosenblum et al., 2020). Finally, Study 2
tested whether the main effect of response speed on sincerity
judgments is robust to a change to a within-subjects design (Lamb-
din & Shaffer, 2009).
Given that Study 2b had a number of aims, we sought to
maximize power to facilitate the detection of predicted effects
(Pieters, 2017; Simonsohn, 2015). To achieve this, in addition to
recruiting a sufficiently large sample, we also used a longer delay
duration (10 s).
Method
Participants and Procedure
We recruited 600 participants from MTurk (328 females and
272 males; M
age
38.63, SD
age
14.05). We manipulated one
factor between-subjects (opinion share), thereby randomly assign-
ing participants to one of two conditions (minority and majority),
and two factors within-participants: response speed (fast vs.
slow) and scenario (four scenarios: Cola, Italian election, CA
election, and Sweaters). In the minority condition, participants
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
9
SLOW LIES
replied that they liked the element also preferred by the minority;
in the majority condition, that they liked the element also preferred
by the majority. In each condition, participants were presented four
scenarios. In each scenario, two interviewees were asked about
their preferences for one of two objects (one also preferred by a
majority, and a second one preferred by a minority), in a wide
range of topics ranging from their electoral preferences (in the
United States and in Italy—California election and Italian election
scenarios), to their soft drinks preferences (Cola scenario), to their
clothing preferences (Sweaters scenario). The first interviewee
responded immediately (fast), and the second one after a 10 s delay
(slow). Complete scenarios are available in the online supplemen-
tal materials.
After reading the scenarios, participants were asked to estimate
how “Sincere” responders were, and how much they “Meant what
they said” on scales from 1 (not at all)to7(very much), which
showed high reliability (all s.80) for all four scenarios and
averaged for each scenario, resulting in eight sincerity judgments.
Participants were also asked—per scenario—to rate the sentence
“Another thought popped up in his/her head,” anchored at 1 (not at
all)and7(very much), as a measure of thought suppression
inferences.
Results
Sincerity Judgments
A mixed-model ANOVA with response speed (fast vs. slow)
and scenario (High School; Cola; Party; Election) as within-
subjects factors, opinion share (majority vs. minority) as between-
subjects factor, and sincerity judgments as dependent variable
found a significant effect for response speed, F(1, 598) 488.17,
p.001,
2
.203. On average, slower responders were con-
sidered less sincere (M4.42, SD 1.44) than faster responders
(M5.89, SD 1.09), d0.90, 95% CI [0.73, 1.07]. Further,
the ANOVA found significant effects for scenario, F(3, 1794)
4.52, p.004,
2
.001, indicating that the responses were
considered more sincere in some scenarios, and for the interaction
between response speed and scenario, F(3, 1794) 6.91, p
.001,
2
.001, indicating that the effect size varied across
scenarios (while remaining statistically significant and in the same
direction). No significant effect was found for the interaction
between response speed and opinion share, F(1, 598) 0.03, p
.87, for the three-way mixed interaction between response speed,
scenario, and opinion share, F(1, 1794) 1.45, p.23, and the
between-subjects main effect of opinion share, F(1, 598) 0.01,
p.91, all with
2
.001. Descriptive statistics and effect sizes
per scenario are presented in Table 4.
Thought Suppression Inferences
A mixed-model ANOVA with response speed (fast vs. slow)
and scenario (High School; Cola; Party; Election) as within-
subjects factors, opinion share (majority vs. minority) as between-
subjects factor, and thought suppression inferences as dependent
variable found a significant effect for response speed, F(1, 598)
509.66, p.001,
2
.247. On average, participants were more
likely to think that participants were suppressing a thought if they
replied slower (M4.77, SD 1.43) compared with when they
replied faster (M2.67, SD 1.74), d0.92, 95% CI [0.75,
1.09]. The ANOVA also found significant main effects for sce-
nario, F(3, 1794) 3.67, p.012,
2
.001, indicating that on
average, participants were considered more sincere in some sce-
narios, and for the interaction between share and scenario, F(3,
1794) 2.91, p.033,
2
.001, indicating a small change in
the effect of opinion share in different scenarios. The main effect
of opinion share, F(1, 598) 2.16, p.14, the two-way inter-
action between response speed and opinion share, F(3, 1794)
1.09, p.30, and the three-way interaction between response
speed, scenario, and opinion share, F(3, 1794) 1.09, p.35,
were not statistically significant, all yielding an effect size
2
.001. Descriptive statistics and effect sizes per scenario are pre-
sented in Table 4.
Mediation Analysis
Within-subjects mediation analysis using the MEMORE
macro for SPSS (Montoya & Hayes, 2017, 5,000 bootstraps),
showed a significant indirect effect of response speed (pseu-
doindependent variable) on the average of sincerity judgments
(dependent variable) through the average measure of thought
suppression inferences (mediator), ab (SE)0.85 (0.09), 95%
CI [0.67; 1.04], providing statistical evidence for our theoretical
model, and supporting the notion that the effect of response
speed on sincerity judgments is driven by thought suppression
inferences.
Discussion
Results of Study 2b suggest that slow responses are perceived as
less sincere because observers infer that slow responders are more
likely to be suppressing an alternative thought. Further, it does not
seem that the opinion share moderates the effect. Finally, findings
Table 4
Sincerity Judgements Across Response Delays and Vignette Scenarios, Results of Study 2
Sincerity judgments Thought suppression inferences
Vignette scenario Fast M(SD) Slow M(SD) Cohen’s d(95 % CI) Fast M(SD) Slow M(SD) Cohen’s d(95 % CI)
Cola 5.79 (1.27) 4.46 (1.54) 0.72 [0.55, 0.88] 2.57 (1.92) 4.69 (1.46) 0.96 [0.79, 1.13]
Italian elections 5.90 (1.24) 4.32 (1.67) 0.80 [0.63, 0.97] 2.65 (1.98) 4.84 (1.69) 0.82 [0.65, 0.99]
High school 5.93 (1.23) 4.52 (1.64) 0.77 [0.60, 0.94] 2.64 (1.92) 4.75 (1.70) 0.82 [0.65, 0.99]
California elections 5.95 (1.22) 4.39 (1.73) 0.82 [0.65, 0.99] 2.61 (1.98) 4.70 (1.73) 0.79 [0.62, 0.96]
Average 5.89 (1.09) 4.42 (1.44) 0.90 [0.73, 1.07] 2.67 (1.74) 4.77 (1.43) 0.92 [0.75, 1.09]
Note.CIconfidence interval. All differences between the slow and the fast condition were significant, p.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10 ZIANO AND WANG
of Study 2b also suggest that the effect of response speed on
perceived sincerity is observed even when the same observer rated
both the slow and the fast responder, and even for longer response
delays. Study S1 in the online supplemental materials replicates
both the main effect and the mediation results of Studies 2a–2b
using a between-subjects design with French participants.
Study 2c—Testing the Unique Causal Effect of
Thought Suppression Inferences on Sincerity
Judgments
A limitation of Studies 2a and 2b is that while thought suppres-
sion inferences were analyzed as the mediator, it was measured
along with the dependent measure. As such, whether or not
thought suppression inferences per se have a causal effect on
sincerity judgments remains to be tested. To this end, Study 2c was
conducted. We manipulated thought suppression inferences with-
out manipulating response speed to provide stronger evidence for
the directional and causal role of thought suppression inferences
on sincerity judgments. This study was preregistered at https://
aspredicted.org/cc775.pdf.
Method
Participants and Procedure
We recruited 303 members of the British public on Prolific (146
males, 155 females, 2 other/prefer not to disclose; M
age
31.12,
SD
age
10.05). Participants were randomly assigned to one of two
conditions, control and thought suppression. They were presented
the following scenario, containing the description of the response
of a woman to a mundane question (in brackets, the text added in
the thought suppression condition and absent in the control con-
dition):
Michelle is a woman in her 30s.
Michelle is asked by a friend if she likes the movie that they are
watching.
[Michelle seems to be suppressing another thought when replying.]
Michelle replies: “Yes, I do”
Then, participants rated Michelle’s sincerity on a scale from 1
(not at all)to7(very much), and, as a manipulation check (pre-
sented in the next screen), they rated whether she seemed to be
suppressing another thought, also on a scale from 1 (not at all)to
7(very much).
Results
Thought Suppression Inferences
Michelle was rated as more likely to be suppressing a thought in
the thought suppression condition (M5.25, SD 1.14) com-
pared with the control condition (M3.52, SD 1.55), t(301)
11.04, p.001, d1.27, 95% CI [1.02, 1.52].
Sincerity Judgments
Michelle was rated as less sincere in the thought suppression
condition (M3.31, SD 1.17) compared with the control
condition (M4.97, SD 1.28), t(301) 11.79, p.001, d
1.35, 95% CI [1.10, 1.60].
Discussion
Results of Study 2c show that individuals suppressing another
thought while responding (as compared with those not suppressing
another thought) are perceived as less sincere. This result strength-
ens our theoretical model by showing that thought suppression
inferences is indeed a causal mechanism driving observers’ sin-
cerity judgments. Findings of Study 2c also augment the findings
of Studies 2a and 2b by demonstrating a causal chain. Specifically,
while Studies 2a and 2b showed that the predictor (response speed)
affects both the mediator (thought suppression inferences) and the
dependent variable (perceived sincerity), Study 2c showed that the
mediator has an independent causal effect on the dependent vari-
able. Many effects in social psychology are multiply mediated, and
we do not expect thought suppression to be the one and only
inference generated by response speed that, in turn, promotes
insincerity judgments. In the next two studies, we test whether a
second mediator—answer fabrication inferences—may also under-
lie our effect.
Study 3a—The Mediating Role of Answer Fabrication
Inferences
The objective of this study was to test whether the effect of
response speed on sincerity judgments can be mediated by answer
fabrication inferences. This study was preregistered at https://
aspredicted.org/my9tr.pdf.
Method
Participants
We recruited 199 participants from MTurk (97 males, 99 fe-
males, 2 other, 1 preferred not to disclose; M
age
40.09, SD
age
11.44). Twelve participants failed the attention check at the end of
the survey, by replying “Yes” to the question “Have you ever been
to your own funeral?” This left 187 valid participants (90 males, 94
females, 2 other, 1 preferred not to disclose; M
age
40.42,
SD
age
11.61).
Procedure
Participants read a scenario in which a woman was asked a
question and replied either immediately or after a short pause.
Participants read this scenario (in the fast condition, they also read
the additional text in round brackets; in the slow condition, they
did not read the text in round brackets, but read the additional text
in square brackets).
Michelle is a woman in her 30s, who works as a software developer
in Plano, Texas, but is originally from Butte, Montana.
She is taller than average and likes to dress in black. Over the years,
she acquired a slight Texas Drawl.
Michelle is asked by a friend if she likes the play that they are
watching together, “Cats.”
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
11
SLOW LIES
Michelle (immediately) replies [after 3 s]: “Yes, I do”
Measures
Participants were then asked four questions, each anchored at 1
(not at all)and7(very much). First, they were asked how sincere
they believed Michelle was (“How sincere was Michelle?”); then,
they were asked whether they believed that Michelle had to come
up with an improvised response and whether they believed that
Michelle had to invent an answer on the spot (“Do you think
Michelle had to come up with an improvised response?”; “Do you
think Michelle had to invent an answer on the spot?”) These two
items were intended as a measure of answer fabrication. Because
they showed very high correlation (Pearson’s r.85, p.001),
they were averaged in an answer fabrication index. Finally, as a
manipulation check, they were asked how fast Michelle’s response
was (“How fast was Michelle in responding?”).
Results
Manipulation Check
Participants rated Michelle as faster when she replied immedi-
ately (M6.39, SD 0.89) compared with when she replied after
3s(M3.82, SD 1.69), t(185) 13.16, p.001, d1.93,
95% CI [1.58, 2.28].
Sincerity Judgments
Participants rated Michelle as more sincere when she replied
immediately (M5.70, SD 1.14) compared with when she
replied after3s(M4.82, SD 1.42), t(185) 4.71, p.001,
d0.69, 95% CI [0.39, 0.98].
Answer Fabrication Inferences
Participants were more likely to rate Michelle as fabricating an
answer when she replied after3s(M3.29, SD 1.80)
compared with when she replied immediately (M2.31, SD
1.59), t(185) 3.93, p.001, d0.58, 95% CI [0.29, 0.87].
Mediation
A mediation analysis (PROCESS v3.5 for SPSS, 5,000 boot-
straps) with response speed as the independent variable, answer
fabrication impressions as the mediator, and sincerity judgments as
the dependent variable found a significant indirect effect, ab
(SE)⫽⫺0.40 (0.12), 95% CI [0.66, 0.17]. This supports the
notion that observers perceive slower responders to be less sincere
because they believe the response delay is the result of the re-
sponder fabricating their response.
Discussion
Study 3a replicates the main effect investigated in this article,
and also shows that people believe that slower responders are more
likely to be fabricating answers. Furthermore, Study 3a shows that
in addition to thought suppression inferences, the effect of re-
sponse speed on sincerity judgments is also mediated by answer
fabrication inferences.
Study 3a contributes to the present work in two ways. First, we
establish another cognitive process that, in addition to thought
suppression, people infer from response speed. Second, we show
that this cognitive process mediates sincerity judgments engen-
dered by response speed, adding considerable nuance to the me-
diating mechanism we proposed in this article. Next, to better
carve out the causal nature of answer fabrication inferences as an
additional mediator, we test whether answer fabrication inferences
alone have a causal effect on sincerity judgments.
Study 3b—Testing the Unique Causal Effect of
Answer Fabrication Inferences on Sincerity Judgments
Study 3b was preregistered at https://aspredicted.org/bg5fm
.pdf.
Method
Participants
We recruited 202 participants from MTurk (96 males, 103
females, 1 other, 2 preferred not to disclose; M
age
38.23,
SD
age
11.28). Eleven participants failed the attention check at
the end of the survey, by replying “Yes” to the question “Have you
ever been to your own funeral?” This left 191 valid participants
(92 males, 96 females, 1 other, 2 preferred not to disclose; M
age
38.56, SD
age
11.45).
Procedure
Participants were randomly assigned to one of two conditions,
control and fabrication. They read a version of the following
scenario (specifically, participants in the fabrication condition also
read the sentence in brackets, while those in the control condition
did not):
Michelle is a woman in her 30s, who works as a software developer
in Plano, Texas, but is originally from Butte, Montana.
She is taller than average and likes to dress in black. Over the years,
she acquired a slight Texas drawl.
Michelle is asked by a friend if she likes the play that they are
watching together.
[Michelle seems to be coming up with an improvised thought when
responding.]
Michelle replies: “Yes, I do”:
Measures
Participants were then asked three questions, each anchored at 1
(not at all)and7(very much), taken from Study 3a. First, they
were asked how sincere they believed Michelle was (“How sincere
was Michelle?”); then, they were asked whether they believed that
Michelle had to come up with an improvised response and whether
they believed that Michelle had to invent an answer on the spot
(“Do you think Michelle had to come up with an improvised
response?”; “Do you think Michelle had to invent an answer on the
spot?”). These two items were intended as a measure of answer
fabrication impressions, which in this case was used as a manip-
ulation check. Since they showed very high correlation (Pearson’s
r.81, p.001), they were averaged in an answer fabrication
index.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12 ZIANO AND WANG
Results
Manipulation Check
Participants believed that Michelle was more likely to be fab-
ricating an answer in the fabrication condition (M4.04, SD
1.63) compared with the control condition (M2.27, SD 1.40),
t(189) 8.06, p.001, d1.17, 95% CI [0.86, 1.48].
Sincerity Judgments
Participants believed that Michelle was less sincere in the fab-
rication condition (M4.26, SD 1.65) compared with the
control condition (M5.65, SD 1.06), t(189) ⫽⫺7.05, p
.001, d⫽⫺1.02, 95% CI [1.32, 0.72].
Discussion
Study 3b shows that people who are seen as improvising their
response are perceived as less sincere. This again completes the
causal chain by showing an effect of a candidate mediator (answer
fabrication inferences) on the dependent variable (sincerity judg-
ments) without the measurement or manipulation of the indepen-
dent variable (response speed). Next, we investigate boundary
conditions for the effect of response speed on sincerity percep-
tions.
Study 4—Sincerity Judgments of Socially Undesirable
Responses Are Less Affected by Response Speed
A limitation of Study 2b was that we had conflated opinion
share (majority vs. minority) with social desirability. To test the
role of social desirability in the predicted relationship more accu-
rately, Study 4 was conducted.
Method
Participants
We recruited 588 participants on MTurk (327 females, 261
males; M
age
39.68, SD
age
14.87). We randomly assigned
participants to one of four conditions in a 2 (response speed: fast
vs. slow) 2 (social undesirability of the response: desirable vs.
undesirable) between-subjects design. All participants were asked
to write down the initials of their best friend at the beginning of the
survey. Then, they were asked to read a vignette in which they
asked their best friend for his or her opinion, to which their best
friend would reply either immediately (fast condition) or after a 15
s delay (slow condition), giving either a socially desirable (“it’s
really good”) or socially undesirable (“it’s really bad”) response.
Participants read one of the four versions of the following scenario
(the slow and the socially undesirable variants are presented in
brackets):
Imagine you have made a cake for a party. You are unsure if people
will like it.
Your best friend tastes it and you ask him/her for his or her opinion.
He or she immediately (after 15 s) replies: “It’s really good (really
bad)!”
Participants were then asked four questions in randomized or-
der, which included the focal measure of perceived sincerity (“Do
you think your best friend was sincere in this occasion?” on a scale
from 1 (not at all)to7(very much)). Two were manipulation
checks, for response speed (“How fast was your friend in reply-
ing?”) on a scale from 1 (very slow)to7(very fast) and for social
desirability (“Was the answer positive or negative?”) on a scale
from 1 (very negative)to7(very positive). As an additional check,
we also measured perceived social desirability (“Did your friend
risk upsetting you with this answer?”) on a scale from 1 (not at all)
to7(very much), to ascertain whether the negative response was in
fact more socially undesirable than the positive one.
Results
Manipulation Check, Response Speed
Participants were perceived to be slower in the slow condition
(M3.72, SD 1.77), compared with the fast condition (M
6.31, SD 1.07), t(586) ⫽⫺21.48, p.001, d1.77, 95% CI
[1.58, 1.96].
Manipulation Check, Response Valence
Answers were perceived more negative when socially undesir-
able (M2.33, SD 1.81) compared with when they were
socially desirable (M5.89, SD 1.34), t(586) ⫽⫺27.12, p
.001, d2.23, 95% CI [2.02, 2.44].
Manipulation Check, Social Desirability
Negative responses were in fact perceived as more socially
undesirable (M4.17, SD 2.06) compared with positive
responses (M2.68, SD 1.84), t(586) 9.22, p.001, d
0.76, 95% CI [0.59, 0.93].
Sincerity Judgments
A22 ANOVA with social desirability and response speed as
between-subjects factors found the predicted two-way interaction
between response speed and politeness, F(1, 584) 10.00, p
.002,
2
.015. Tukey-corrected post hoc tests show that when
the response was socially desirable, slow responses (M4.84,
SD 1.66) were considered less sincere than fast responses (M
6.01, SD 1.14, t(584) ⫽⫺7.04, p.001, d0.81, 95% CI
[0.57, 1.05]), but when the response was socially undesirable, the
difference between slow responses (M5.78, SD 1.54) and fast
responses (M6.21, SD 1.28) was reduced, t(584) ⫽⫺2.59,
p.049, d0.30, 95% CI [0.07, 0.53]. Descriptive statistics are
depicted in Figure 2.
Discussion
The results of Study 4 add an important boundary condition to
our theoretical model. Specifically, while response speed seems to
play a significant role in sincerity inferences if the response is
socially desirable, the magnitude of this influence is much smaller
when the response is socially undesirable. Studies S2 and S3 in the
online supplemental materials replicate this study with different
vignette scenarios and actors and extend it to guilt judgments.
In a similar vein, another important boundary condition to
examine is that of delay attribution. If people have reason to
believe that the response delay is not the result of a cognitive
process related to insincerity such as thought suppression or an-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
13
SLOW LIES
swer fabrication but a different reason, such as memory effort, they
may be less likely to judge slower responders as insincere. We test
this prediction in Study 5.
Study 5—Perceived Memory Effort Moderates the
Effect of Response Speed on Perceived Sincerity
Study 5 was preregistered at https://aspredicted.org/5kr2q.pdf.
Method
Participants and Procedure
We recruited 800 participants from MTurk (428 males, 372
females; M
age
38.21, SD
age
12.34). Participants were ran-
domly assigned to one of eight conditions in a 2 (response speed:
fast vs. slow) 2 (distance in time of the violation: recent past vs.
distant past) 2 (severity of the violation: low vs. high) between-
subjects design. In each condition, participants were presented
with a vignette in which a man named John was asked whether he
had murdered a man (high severity) or stolen some candy (low
severity), either earlier the same day (recent past) or 10 years ago
(distant past), to which John responded “No, I didn’t” either
immediately (fast) or after a 10 s delay (slow). Participants read a
scenario as below (in square brackets, the high severity condition,
the distant past condition and the fast condition variants of the
scenario):
John is accused of having stolen some candy [murdered a man] earlier
today [10 years ago]. He is asked whether he did it and after about 10
s [immediately] he replies: “No, I didn’t!”
Participants were then asked eight questions, all on 7-point
scales ranging from 1 (not at all)to7(very much). These included
first the focal dependent variable, perceived sincerity (“How sin-
cere was John?”); followed by the measure of thought suppression
inferences (“Was John suppressing any other thoughts when re-
sponding?”); perceived memory effort (“Was John trying to re-
member whether he did what he was accused of or not?”); and
perceived ease of retrieval (“Was the event John was asked about
easy to remember?”). Finally, participants completed three manip-
ulation checks, one for response speed (“How fast was John’s
response?”), one for violation severity (“How serious is the vio-
lation of which John is accused?”), and one for distance in time
(“How far ago in the past was the action that John was asked
about?”).
Results
Manipulation Check, Response Speed
Attest using response speed as the independent variable con-
firmed the success of our manipulation, as slower responses (M
3.42, SD 2.02) were perceived as slower than faster responses
(M6.28, SD 1.12), Welch’s t(621.044) ⫽⫺24.71, p.001,
d1.75, 95% CI [1.59, 1.91].
Manipulation Check, Violation Severity
Attest using violation severity as the independent variable
confirmed the success of our manipulation, as stealing some candy
(M3.61, SD 1.80) was considered less severe than murdering
someone (M6.52, SD 1.00), Welch’s t(624.55) ⫽⫺28.20,
p.001, d1.99, 95% CI [1.82, 2.16].
Manipulation Check, Temporal Distance
Attest using distance in time as the independent variable
confirmed the success of our manipulation, as the violation com-
mitted earlier the next day (M2.69, SD 1.75) was considered
less distant in the past than the violation committed 10 years earlier
(M5.62, SD 1.34), Welch’s t(753.22) ⫽⫺26.59, p.001,
d1.88, 95% CI [1.71, 2.05].
Sincerity Judgments
A222 between-subjects ANOVA found the predicted
three-way interaction between response speed, violation severity
and temporal distance, F(1, 792) 3.94, p.048,
2
.004,
indicating that the effect of response speed was significantly
smaller and not statistically significant when the violation was of
low severity and had been committed a long time ago. When the
violation had low severity and was committed in the recent past
(stealing candy earlier the same day), slower responders (M
3.03, SD 1.63) were rated as less sincere than faster responders
(M4.65, SD 1.73, t(792) ⫽⫺7.10, p.001, d0.97, 95%
CI [0.69, 1.25]); when the violation was low in severity and
committed in the distant past, we found a smaller and nonsignif-
icant difference between slower (M3.61, SD 1.74) and faster
responders (M4.07, SD 1.68, contrast t(792) ⫽⫺1.86, p
.58, d0.27, 95% CI [0.02, 0.56])—the effect attenuation we
predicted. When the violation was high in severity and committed
in the recent past (murdering a man earlier today), slower respond-
ers (M3.51, SD 1.75) were judged as less sincere than faster
responders (M4.55, SD 1.44, t(792) ⫽⫺4.29, p.001, d
0.64, 95% CI [0.35, 0.93]); when the violation was high in severity
and committed in the distant past, again slower responders (M
3.65, SD 1.75) were judged as less sincere than faster responders
(M4.48, SD 1.50, t(792) ⫽⫺3.47, p.013, d0.52, 95%
CI [0.36, 0.91]). Descriptive statistics are depicted in Figure 3. The
Figure 2
The Effect of Response Speed and Response Social Desirability
on Sincerity Judgments, Study 4
Note. Error bars represent 1SEM.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
14 ZIANO AND WANG
ANOVA also found a significant main effect of response speed,
F(1, 792) 68.36, p.001,
2
.078, indicating that faster
responses were considered more sincere than slower ones, but no
significant main effects of temporal distance, F(1, 792) 0.20,
p.89,
2
.001, and violation severity, F(1, 792) 3.94, p
.083,
2
.003. The two-way interaction between response speed
and temporal distance was significant, F(1, 792) 8.34, p.004,
2
.010, indicating that the effect of response speed on sincerity
was smaller but still significant when events were far away in the
past compared with closer events, but the other two-way interac-
tions between temporal distance and violation severity, F(1,
792) .02, p.88,
2
.001, and between violation severity and
response speed, F(1, 792) .16, p.69,
2
.001, were not
statistically significant.
Memory Effort
A222 between-subjects ANOVA with response speed,
violation severity, and temporal distance as factors found a signif-
icant two-way interaction between temporal distance of the event
and violation severity, F(1, 792) 15.71, p.001,
2
.018.
When the event being questioned was one in the distant past,
participants perceived that the responder exerted higher effort if
the event was a low-severity violation (M3.95, SD 1.88) as
compared with a high-severity violation (M2.98, SD 1.96,
contrast t(792) 4.65, p.001, d0.50, 95% CI [0.30, 0.50]).
No such effect was found when the event under question was one
in the recent past (low severity M2.98, SD 1.79; high severity
M3.15, SD 1.89, t(792) ⫽⫺0.92, p.80, d0.09, 95%
CI [0.13, 0.27]). Overall, this analysis indicates that participants
expected the responder to be more likely to exert memory effort
for an event that happened in the distant past and that was not a
severe violation. Note that it is exactly in this case that we found
a smaller and nonsignificant effect of response speed on sincerity
judgments: when the event is considered as involving more mem-
ory effort, we observed a smaller effect of response speed. This
finding corroborates our reasoning that the attribution of response
speed to cognitive effort may reduce inferences of insincerity. The
ANOVA also found a significant main effect of response speed,
F(1, 792) 29.86, p.001,
2
.035, indicating that people
attributed higher memory effort to slower responses, significant
main effects of temporal distance, F(1, 792) 13.96, p.001,
2
.016, indicating that participants attributed higher memory
effort to events in the distant past, and violation severity, F(1,
792) 7.23, p.007,
2
.008, indicating that participants
believed that recalling the memory of a trivial violation required
greater memory effort compared with a recalling the memory of a
severe violation. We found no significant effect of the two-way
interactions between temporal distance and response speed, F(1,
792) .53, p.53,
2
.001, and violation severity and
response speed, F(1, 792) 1.00, p.32,
2
.001. Finally, we
found no three-way interaction between response speed, violation
severity, and temporal distance, F(1, 792) 1.46, p.23,
2
.002.
Ease of Retrieval
A222 between-subjects ANOVA with response speed,
violation severity, and temporal distance as factors found no sig-
nificant main effect of response speed, F(1, 792 .001, p.99,
2
.001, a significant main effects of temporal distance, F(1,
792) 101.70, p.001,
2
.097, indicating that participants
thoughts events distant in the past were harder to retrieve, and
violation severity, F(1, 792) 83.30, p.001,
2
.079,
indicating that participants thought the trivial violation was easier
to retrieve compared with the severe violation. We found a sig-
nificant two-way interaction between violation severity and tem-
poral distance, F(1, 792) 69.76, p.001,
2
.066, indicating
that the effect of violation severity was larger for actions more
distant in the past (low severity M3.75, SD 1.97; high
severity M5.85, SD 1.59), contrast t(792) ⫽⫺12.22, p
.001, d1.18, 95% CI [0.97, 1.39] compared with actions closer
in the past (low severity M5.90, SD 1.46; high severity M
6.03, SD 1.44), contrast t(792) ⫽⫺0.55, p.95, d0.09,
95% CI [0.11, 0.29]. If the memory is trivial and in the distant
past, observers consider it harder to retrieve and are more likely to
attribute a delay to a memory effort. The two remaining two-way
interactions were not significant (response speed by temporal
distance, F(1, 792) 1.00, p.32,
2
.001, and response speed
by violation severity, F(1, 792) 0.16, p.69,
2
.001).
Finally, we found no three-way interaction between response
speed, violation severity, and temporal distance, F(1, 792) 2.95,
p.086,
2
.003.
Thought Suppression Inferences
A222 between-subjects ANOVA with response speed,
violation severity, and temporal distance as factors found a signif-
icant main effect of response speed, F(1, 792) 86.84, p.001,
2
.098, indicating that people attributed slower responses to
thought suppression to a greater extent than faster responses, but
no other significant main effects of temporal distance, F(1, 792)
.01, p.93,
2
.001, and violation severity, F(1, 792) .01,
p.94,
2
.001. Critically, we found a statistically significant
two-way interaction between response speed and temporal dis-
tance, F(1, 792) 8.34, p.018,
2
.006, indicating that the
effect of response speed on thought suppression inferences was
Figure 3
The Effect of Violation Severity, Temporal Distance, and Re-
sponse Speed on Sincerity Judgments, Study 5
Note. Errors bars represent 1SEM.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
15
SLOW LIES
smaller when events were more distant in the past compared with
recent events. The other two-way interactions between temporal
distance and violation severity, F(1, 792) .24, p.63,
2
.001, and between violation severity and response speed, F(1,
792) .03, p.86,
2
.001, were not statistically significant.
We did not find a statistically significant three-way interaction
between response speed, violation severity and temporal distance,
F(1, 792) 2.67, p.10,
2
.003.
Additional moderated mediation analyses (discussed in detail in the
online supplemental materials) found significant indirect effects of
both thought suppression inferences and memory effort linking the
relationship between response speed and sincerity perceptions, both
moderated by violation severity and temporal distance. These results
replicate our previous results relating the mediating role of thought
suppression inferences, and provide additional support for the con-
textual role of memory effort inferences if the context requires the
actor to remember information that is hard to retrieve.
Discussion
The moderation effect established by the present study adds
important nuance to our results. It shows that perceived mental
effort, specifically memory effort, can attenuate the effects of
response speed of sincerity judgments. When the event that people
are asked about should be easily retrievable in memory (such as a
mundane experience in the recent past, or a salient memory from
the distant past), people readily infer sincerity from response
speed. However, people are less likely to do so if they have reason
to believe that the event is something that is difficult to remember,
presumably because they are more likely to attribute the response
delay to memory effort, rather than thought suppression or answer
fabrication. This study shows that people attribute response speed
to different cognitive processes in different contexts, and that such
differential attribution affects sincerity judgments. This study is
important because it further expands our understanding of the
circumstances in which response speed matters for sincerity per-
ceptions. If observers believe that the answer requires a cognitive
effort process that is not connected to insincerity, then the effect of
response speed is attenuated.
Study 6—Can People Voluntarily Ignore Response
Speed When Making Sincerity Judgments?
Can people ignore response speed in the interpersonal judgment
process if they are told to? Answering this question would reveal
the extent to which the effect is controllable: if the effect of
response speed is used in the judgment process in an automatic and
uncontrollable way, instructions to ignore response speed should
have no impact on the judgment outcome (Hütter & Sweldens,
2018). If, however, response speed can be taken into account in a
controllable way, one should observe a reduction in the effect of
response speed on sincerity judgments following instructions to
ignore response speed. Whether or not response speed can be
disregarded in the sincerity judgment process is imperative to
examine because the answer likely has implications in high stakes
contexts such as judicial settings. For instance, if it appears that
response speed can be discounted in the sincerity judgment pro-
cess, then jurors should be asked to do so when there are legitimate
reasons for slower responses, such as a language barrier between
the interrogator and the suspect. If our results show that response
speed cannot be discounted in the sincerity judgment process, then
it may be necessary for jury systems to consider concealing the
element of response speed, where appropriate, when presenting
various forms of evidence. For instance, video or audio recordings
of interrogations may be replaced with written dialogues where
response speed is not reflected. This may prevent people from
misattributing response delays to thought suppression or answer
fabrication, which may be connected to insincerity. This study was
preregistered at https://aspredicted.org/jm3q3.pdf.
Method
Participants
We initially recruited 605 participants on MTurk (310 males,
295 females; M
age
38.26, SD
age
11.75). As in Study 1a and
1b, earphones or headphones were required for this task. We used
two attention checks to verify that participants were paying atten-
tion to the video and to the presented audio. At the beginning of the
task, participants were shown a video in which a hand with two
fingers was shown and the word cabbage was spoken. In the next
screen, participants were asked which word was spoken in the
video (among Tomato, Potato, and Cabbage), and how many
fingers were shown in the video (Four, Seven, and Two). Five-
hundred and 67 participants correctly answered both attention
checks (282 males, 285 females; M
age
38.53, SD
age
11.86)
and were retained for analyses.
Procedure
Participants were randomly assigned to one of four conditions in
a 2 (response speed: fast vs. slow) 2 (instruction: ignore vs. no
instruction) between-participants design. Participants were first
shown two videos (in random order, of a man and a woman; also
used in online Supplemental Materials Study S1), in which they
received the description that this was a police interrogation and
that the person in the video was accused of having stolen a few
thousand dollars from their workplace. In the fast condition, after
the actor playing the police officer had asked the question: “Did
you steal the money?”, the actor playing the suspect immediately
replied “No, I didn’t!”, whereas in the slow condition, the suspect
replied aftera5sdelay. Following this, participants were either
not given any additional instructions (no instructions condition), or
told: “IMPORTANT: In answering the questions, please do not
take into account the speed with which the person replied.”
Participants subsequently completed five questions: a manipu-
lation check of response speed (“How was the response?”; mea-
sured on a scale ranging from 1 [very slow]to7[very fast]), a
measure of perceived sincerity (“What do you think about the
statement?” “Sincere”; “Truthful”, on 7-point scales from 1 (not at
all)to7(very much), averaged in two sincerity measures, both
Cronbach’s s.90), a guilt judgment item (“Do you think this
person is guilty?”, which was answered by choosing “Yes” or
“No”), and a manipulation check of instruction compliance (“Did
you take response speed into account in replying to the above
questions?”) that was answered on a 7 point scale from 1 (not at
all)to7(very much).
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
16 ZIANO AND WANG
Results
Manipulation Check, Response Speed
A22 (actor gender: man vs. woman, within-subjects) 2
(response speed: fast vs. slow, between-subjects) mixed-model
ANOVA revealed a significant, between-subjects effect of re-
sponse speed, F(1, 565) 1638.72, p.001,
2
.638. On
average, slower responders (M1.91, SD 1.00) were consid-
ered slower than faster ones (M5.17, SD 0.91),
t(565) ⫽⫺40.48, p.001, d3.40, 95% CI [3.14, 3.66]. The
ANOVA found a significant effect of actor gender, such that the
man was on average perceived faster, F(1, 565) 18.60, p.001,
2
.004, and a significant mixed interaction between response
speed and actor gender, F(1, 565) 7.66, p.006,
2
.002,
showing that the effect of response speed was stronger for the male
actor.
Manipulation Check, Instructions to Ignore
A 2 (actor gender: man vs. woman, within-subjects) 2 (in-
structions to ignore: presented vs. no instructions, between-
subjects) mixed-model ANOVA revealed a significant between-
subjects main effect of instructions to ignore, F(1, 565) 204.16,
p.001,
2
.237. On average, participants who were not
instructed to ignore response speed (M5.18, SD 1.51)
reported that they took response speed into account to a larger
extent than participants who were instructed to ignore it (M
3.08, SD 1.96), t(565) 14.29, p.001, d1.20, 95% CI
[1.02, 1.38]. The ANOVA revealed no significant within-subjects
effect of scenario, F(1, 565) .08, p.78,
2
.001, and no
significant mixed interaction between actor gender and instruc-
tions to ignore, F(1, 565) 2.54, p.11,
2
.001,
Sincerity Judgments
A 2 (instructions to ignore: presented vs. no instruction; be-
tween-subjects) 2 (actor gender: man vs. woman, within-sub-
jects) 2 (response speed: fast vs. slow, between-subjects) mixed-
model ANOVA revealed a significant between-subjects main
effect of response speed, F(1, 563) 178.79, p.001,
2
.082,
a significant between-subjects main effect of instructions to ignore,
F(1, 563) 8.31, p.004,
2
.004, which were both qualified
by the predicted, significant interaction between speed and instruc-
tions to ignore, F(1, 563) 10.74, p.001,
2
.005, indicating
that the effect of response speed on sincerity was reduced by
instructions to ignore it. When participants were not given instruc-
tions to ignore response speed, they considered slower responders
(M2.58, SD 1.18) to be less sincere than faster responders
(M4.20, SD 1.12, t(563) 11.74, p.001, d1.40, 95%
CI [1.14, 1.66]). When participants were given instructions to
ignore response speed, the difference between slower responders
(M3.18, SD 1.21) and faster responders (M4.16, SD
1.12) was significantly smaller, t(563) 7.16, p.001, d0.84,
95% CI [0.60, 1.08]. Descriptive statistics are presented in Table 5.
The ANOVA also found a significant within-subjects effect of
actor gender, F(1, 563) .21, p.021,
2
.001, indicating that
the woman was considered more sincere than the man in our
scenarios; no significant mixed interaction between actor gender
and speed, F(1, 563) 1.16, p.28,
2
.001, no significant
interaction between actor gender and instructions to ignore, F(1,
563) .35, p.56,
2
.001, and a significant mixed three-way
interaction between actor gender, response speed, and instructions
to ignore, F(1, 563) 5.72, p.037,
2
.001, indicating a
larger effect of the interaction of response speed and instructions to
ignore in the woman scenario compared with the man scenario.
Guilt Judgment
A repeated-measures logistic regression with actor gender as the
within-subjects factor and response speed and instructions to ig-
nore as between-subjects factors found a statistically significant
interaction between response speed and instructions to ignore,
2
(1) 9.37, p.002, as predicted. When participants were not
instructed to ignore response speed, they considered the slower
responders (215/288 total choices, or 75% of choices) as more
likely to be guilty than the faster one (85/278 total choices, or
31%). When they were instructed to ignore response speed, the
difference in guilt judgments between slower responders (170/284,
or 60% of responses) and faster ones (100/246, or 41% of re-
sponses) was smaller. These results are represented in Figure 4 and
Table 5. The logistic regression found no significant main effect of
actor gender,
2
(1) .49, p.48; a significant effect of response
speed,
2
(1) 125.36, p.001, indicating that, in general, faster
responders were less likely to be considered guilty; a significant
main effect of instructions to ignore,
2
(1) 4.13, p.042,
indicating that, on average, actors were less likely to be considered
guilty when participants were instructed to ignore response speed;
no significant interaction between scenario and response speed,
2
(1) .66, p.42; no significant interaction between actor
Table 5
The Effect of Response Speed and Instructions to Ignore Response Speed on Sincerity Judgments
and Guilt Judgment, Study 6
Actor Response
speed Instructions to ignore
response speed Sincerity judgments
M(SD)Guilt judgment
frequency (%)
Female actor Slow No instructions 2.57 (1.33) 107/144 (74%)
Ignore response speed 3.27 (1.46) 83/142 (59%)
Fast No instructions 4.41 (1.57) 39/138 (28%)
Ignore response speed 4.19 (1.42) 49/143 (34%)
Male actor Slow No instructions 2.59 (1.30) 108/144 (75%)
Ignore response speed 3.09 (1.49) 87/142 (61%)
Fast No instructions 3.99 (1.28) 46/138 (33%)
Ignore response speed 4.14 (1.45) 51/143 (36%)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
17
SLOW LIES
gender and instructions to ignore,
2
(1) .19, p.66; and no
significant three-way interaction between actor gender, speed, and
instructions to ignore,
2
(1) 1.50, p.22.
Discussion
Study 6 shows that the effect of response speed on perceived
sincerity and guilt judgment can be reduced by explicit instructions
to ignore response speed, although we did not manage to eliminate
it completely. This partiality of the effect reduction may be be-
cause of a proportion of participants who still used response speed
as a cue to sincerity even when instructed not to, as the relevant
manipulation check showed a lower value in the “ignore” (M
3.08) than in the “no instructions” condition (M5.18), but still
higher than the scale floor (1 in this case). However, instead of
viewing the results of Study 6 as the outcome of a manipulation
failure, they might be better thought of as a reflection of reality—
perhaps people believe that response delays are so integral to
judgments of sincerity that they continue to voluntarily rely on it
despite explicit instructions not to. Regardless, these results sug-
gest that people are aware that they are taking response speed into
account when making sincerity judgments, and that the effect of
response speed on sincerity judgments and guilty votes is control-
lable to a certain extent.
General Discussion
Using diverse stimuli (audio, video, and vignette scenarios),
participants (American, French, and British), study designs (both
within- and between-subjects), and situations (from mundane opin-
ions to confessing a crime), the present study shows that slower
responders are perceived as less sincere. In Studies 1a, 1b, and 1c,
using audio, video, and vignette scenario stimuli, respectively, we
showed that observers judge slower responders as less sincere. In
Studies 2a, 2b, and 2c, we demonstrated that the effect of response
speed on perceived sincerity is driven by thought suppression
inferences such that observers rate slower responders as less sin-
cere because they attribute the response delay to the responder
suppressing another thought. Studies 3a and 3b show that an
alternative mediator, perceived answer fabrication, is also part of
the explanation of this phenomenon. In Study 4 we found that the
perceived sincerity of socially undesirable responses are less likely
to be impacted by response speed. Study 5 showed that the type of
cognitive process inferred by response speed can influence the
magnitude of the effect: if a response delay can be attributed to
memory effort, the effect of response speed on sincerity judgments
is reduced. Finally, in Study 6, we showed that people are able to
disregard response speed information to some extent in the sincer-
ity judgment process when explicitly instructed to do so. In the
online supplemental materials, Study S1 replicates the basic effect
in a different population (French students). Studies S2 and S3
replicated and extended Study 4 to different scenarios.
Theoretical Implications
Findings of the present study compliment and extend the exist-
ing literature in several ways. Numerous studies have documented
people’s reliance on observing other people’s response speed in
various impression formation and judgment processes, ranging
from morality attribution (Critcher et al., 2013) to expectations of
cooperation (Evans & van de Calseyde, 2017). However, given
that perceived sincerity is an element that is crucial for trust and
cooperation (Caza et al., 2015; Lewicki & Bunker, 1996), it is
discomforting to see that extant research could not give a clear
answer to the question of whether response speed has a causal
effect on perceptions of sincerity. The present study addresses this
gap in the literature by showing that sincerity is also inferred from
response speed. Demonstrating that response speed, alone, is caus-
ally responsible for changes in sincerity perceptions also enhances
our understanding of the relationship between these two variables
because previous research have been unable to provide direct
evidence of causality. Specifically, previous studies involving re-
sponse speed and perceived sincerity have only tangentially ex-
amined the relationship between the two, such as by correlating
response speed and perceived sincerity as dependent variables
after having manipulated deception intention (Buller et al., 1991;
Cornetto, 2002; Harrison et al., 1978; Stiff & Miller, 1986).
In addition to demonstrating causality, we consistently show
that slower responses are perceived as less sincere. This helps set
the record straight on the directionality of the relationship between
response speed and perceived sincerity, which previous research
has provided contradictory findings on. For instance, some studies
showed a curvilinear relationship (Baskett & Freedle, 1974; Boltz,
2005), some showed no statistically significant relationship (Vrij et
al., 2001), some suggesting that slower responses are perceived as
more sincere (Kraut & Lewis, 1982), and some suggesting that
slower responses are perceived as less sincere (Harrison et al.,
1978). Reliably establishing effect directionality is crucial for the
advancement of research on this phenomenon because confidence
in the superordinate phenomenon is the cornerstone of further
inquisition, such as on the mechanism underlying the phenomenon
and its boundary conditions. The robustness and high level of
replicability observed in the present research also allows us to
speculate that previous inconsistencies were indeed the result of
methodological inconsistencies and limitations such as varying
study designs and underpowered samples, respectively.
Figure 4
The Effect of Response Speed and Instructions to Ignore Re-
sponse Speed on Guilt Judgments, Study 6
Note. Error bars indicate 1 standard error around the proportion.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
18 ZIANO AND WANG
Our research also carries implications for the study of conver-
sations. Research in linguistics suggests that people tend to delay
“dispreferred” responses, that is, responses that risk offending the
counterpart, for instance by changing the order of the words if they
are Japanese speakers, or by hesitating, delaying the response, and
inserting paraverbal cues if they speak English (Pomerantz, 1984;
Tanaka, 2008). It has also been evidenced that recipients anticipate
that delayed responses will be dispreferred (Lerner, 1991, 1996).
The present work complements such previous research by showing
that there are additional consequences to a response delay, namely,
perceived insincerity. Additionally, the present findings also com-
plement the literature connecting response speed and actual de-
ception, because it suggests that slower responses are not only
more likely to be insincere (Seymour & Kerlin, 2008; Seymour et
al., 2000; Suchotzki et al., 2017), but are also more likely to be
perceived as insincere.
Interestingly, findings of the present study suggest that people
have a rather sophisticated network of intuitions about sincerity
and response speed. For instance, perceived sincerity seems to be
commensurately inferred from the magnitude of response delay, up
to a certain point: a response given after2sisgenerally considered
less sincere than a response given immediately, and more sincere
than a response given aftera5sdelay. This is not the only cue
supporting the notion that people hold a vast web of lay intuitions
regarding response speed and sincerity judgments. We also found
that sincerity judgments may be more, or less, response speed-
based depending on the cognitive process to which response delays
are attributed. Specifically, delay-induced perceptions of insincer-
ity are the result of the delay being attributed to thought suppres-
sion or answer fabrication. Consequently, when the delay can be
attributed to other cognitive processes not directly connected with
insincerity, such as mental effort, response delays are not judged as
insincere to the same extent. Taken together, the present research
extends the existing literature by elucidating two novel processes
underlying the effect of response speed and perceived sincerity, as
well as establishing the nuances associated with people’s intuitions
regarding response speed and sincerity.
Further detail was added to our theoretical model when the
effect of response speed on perceived sincerity was shown to be
smaller in magnitude when responses are socially undesirable.
Consistent with the notion that socially undesirable acts are con-
sidered a signal of sincerity (Bonnefon & Villejoubert, 2006;
Feldman et al., 2017), the present study connects two streams of
research that have—until now—proceeded in parallel: one inves-
tigating the interpersonal consequences of response speed (Evans
& van de Calseyde, 2017; Van de Calseyde et al., 2014) and one
investigating people’s impressions of socially dispreferred actions
and words (Brown & Levinson, 1987; Lee & Pinker, 2010). This
also reinforces the notion that the attribution of cognitive process
to response speed is situation-dependent, and that several contex-
tual factors are at play in determining what inferences are made
based on response speed.
In addition, previous research has largely investigated the con-
nections between response speed and observers’ inferences of
response speed one at a time. For instance, observed response
speed in judgments of decision and recommendation quality has
been attributed to inferences of thoughtfulness in one study (Kupor
et al., 2014), and to inferences of cognitive effort in another study
(Efendic´ et al., 2020). In both studies, the connection between
response speed and the inferred cognitive process (thoughtfulness
and cognitive effort) has been identified, but conditions under
which observers are more likely to attribute response speed to a
different cognitive process have not been acknowledged. This is
potentially problematic because it may give off the impression that
observers typically attribute response speed to only a single cog-
nitive process. Our research sets the record straight by showing
that people are capable of attributing response speed to different
cognitive processes (thought suppression, answer fabrication, and
memory effort) when making a judgment. We also provide some
conditions under which alternative attributions are made (e.g.,
response delay is more likely to be attributed to mental effort when
the question is related to an event in the distant past).
Finally, we show that when instructed explicitly, people are able
to voluntarily discount response speed in the sincerity inference
equation. While this discounting was only partial in magnitude, it
nevertheless suggests that the effect of response speed on sincerity
judgments does not operate completely automatically and surrep-
titiously (Hütter & Sweldens, 2018), and that presumably people
use response speed as a cue to infer sincerity willingly (Dietvorst
& Simonsohn, 2019). Therefore, similar to people’s voluntary use
of to-be-ignored information (Dietvorst & Simonsohn, 2019), the
effect of response speed on sincerity judgments can be considered
controllable, at least to an extent. This finding is important because
it connects our findings with the larger literature on inference
controllability (for reviews, see Gawronski & Bodenhausen, 2006;
De Houwer et al., 2001), crucial to the understanding of the
automaticity of the effect and of the effectiveness of potential
strategies to reduce it.
Practical Implications
Being able to detect sincerity in others is paramount for human
collaboration and trust (Boyd et al., 2003; Milinski et al., 2002;
Trivers, 1971). As such, people are constantly inferring sincerity in
others, be it for high stakes events such as business deals and
police interrogations, or trivial ones such as sensing whether their
friend really does like their cake. Given that response speed is an
ever-present feature of social interactions, our findings seem to
have worrying implications for the judicial setting. This is because
much like how people often cannot be persuaded to ignore invalid
evidence in mock jury tasks (Dietvorst & Simonsohn, 2019; Kas-
sin & Sommers, 1997), response speed appears to be a social cue
that people are not entirely willing to ignore, even when explicitly
instructed to. It is also possible that response speed is a social cue
that people involuntarily rely on to a certain extent, akin to a
mental heuristic. Consequently, in high stakes situations, such as
judging whether someone is guilty of a crime, it is imperative that
observers exercise more caution when using the cue of response
speed in the judgment process. It would be unfair for the re-
sponder, such as a crime suspect, if the response delay was
misattributed to thought suppression or answer fabrication when it
was in fact caused by a different factor, such as simply being
distracted. The present findings provide a partial solution: instruct-
ing people to ignore response speed in the sincerity judgment
process. Again, our participants, despite being explicitly instructed
to ignore response speed, still relied on it to a certain extent to
judge sincerity, testifying to the strength of response speed as a cue
that people rely on in sincerity judgments. Given this, it may be
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
19
SLOW LIES
appropriate in certain high stakes situations to replace response
speed information (e.g., video or audio evidence) completely. For
instance, judges may opt to be presented with alternative forms of
evidence that do not reflect response speed, such as transcribed
written versions of audio recordings. This method may allow
people to make sincerity judgments without taking response speed
into consideration at all, something we were not able to accomplish
in Study 6.
Limitations and Future Research
While the utilization of well powered sample sizes and a variety
of hypothetical scenarios within the previous study augments the
robustness of the predicted phenomenon, some limitations must be
acknowledged. First, our investigation and findings are confined to
relatively simple questions that required simple responses. This is
because previous research on expectancies of response times in-
dicates that people expect more complicated responses to be ac-
companied by a commensurate response delay (Evans & van de
Calseyde, 2017; Kupor et al., 2014), understandably because such
responses require more thought on the part of the responder. In
fact, Study 5 supported this notion by testing the boundary param-
eter of perceived mental effort required by the response. We found
that when a response is considered to require mental effort, people
are indeed more likely to attribute a response delay to mental effort
rather than thought suppression or answer fabrication, and the
effect of response speed on sincerity judgments consequently
diminishes. It is important to note that we did not include exper-
iments that utilize relatively complex questions such as “What is
something of which you are particularly proud?” The reason for
this is that such complex questions may evoke more than just
mental effort, for instance, it may induce more intense emotions or
memories that are so satisfying that the responder becomes im-
mersed (Wang et al., 2018); thereby, taking more time to respond.
As such, the inherent subjectivity and mechanistic complexities of
responding to such questions may introduce multiple possibilities
in terms of delay attribution, which may inevitably make any
results difficult to interpret. This is precisely why in Study 5, we
utilized stimulus materials that allow us to investigate the moder-
ation effect of mental effort attribution without the unwanted
complexities that may accompany complex questions—so that the
focus of the present work is not blurred, and scope not breached.
In summary, the notion that response delays promote perceptions
of insincerity necessarily entails a level of nuance and should not
be taken as a sweeping statement.
Nevertheless, an interesting question that future studies are
encouraged to examine is whether the effect of response speed
on sincerity judgments can still be observed for less straight-
forward questions when a response delay is introduced in ad-
dition to the expected magnitude of delay. For instance, if
people already expect the responder to respond aftera4sdelay
to account for the complexity of the question and the thought-
fulness required in the answer, would the response still be
perceived as less sincere if provided aftera9sdelay (an
additional 5 s delay)? This would help establish whether the
predicted phenomenon has a more overarching effect across
different question types, or whether our findings are confined to
straightforward questions and responses.
An additional question that awaits future investigation is
whether our findings can be generalized across cultures. De-
spite our best efforts to investigate different demographics, we
managed to collect data in three countries—United States,
United Kingdom, and France. While this provides a measure of
solace regarding the generalizability of the effect of response
speed on sincerity judgments, all samples included in the pres-
ent study came from WEIRD populations (Western, Educated,
Industrialized, Rich, Democratic; Henrich et al., 2010). It is
possible that some cultures or individuals may prize the quality
of being planned and deliberated more than others. As such,
response delays may not always be taken as a reflection of
thought suppression or answer fabrication, but qualities with a
positive connotation instead, such as cautiousness. Future stud-
ies are encouraged to examine the cross-culture generalizability
of the effect of response speed on sincerity judgments.
In a similar vein, future studies are encouraged to explore
potential individual differences in the manifestation of the
predicted phenomenon. Research suggest that individuals differ
in their tendency to use episodic cues to judge others (Scopelliti
et al., 2017). As such, some people may be more predisposed to
rely on response speed as a heuristic in making sincerity judg-
ments without further thought, while others may naturally be
more careful in casting sincerity judgments. Future research
may investigate whether some people are more likely than
others to consider slower responses as insincere, and whether
such individual difference are associated with certain person-
ality traits.
Future research may also wish to take a more solution fo-
cused approach. For instance, we have shown that people are
less likely to infer insincerity from response delays when they
are instructed to ignore response speed, and also when they
attribute response speed to a factor other than thought suppres-
sion or answer fabrication. Given this, it is possible that com-
bining instructions to ignore response speed with a believable
excuse (e.g., “IMPORTANT: please ignore the responder’s
response speed, because she has a hearing impediment”) may be
a more potent way to prevent people from inferring insincerity
from response delays.
Finally, throughout our experiments, we observed a small,
but recurrent trend for female actors to be considered more
sincere, on average. While the lack of a systematic comparison
between the genders precludes us from drawing strong conclu-
sions about this effect, research has shown, for example, that
women are given shorter sentences than men for similar felo-
nies (Doerner & Demuth, 2014). May it be because people tend
to believe women more in the courtroom? Future research may
investigate in a more systematic way whether women are really
considered more sincere than men, and if so, whether this
discrepancy engenders downstream social effects such as
women being more leniently judged, or being relied on more in
workplace settings.
Conclusion
Response speed is an inherent, ubiquitous, and noticeable cue
in social interactions. A burgeoning line of research (Evans &
Rand, 2019; Kupor et al., 2014; Van de Calseyde et al., 2014)
shows that observers use response speed to draw various infer-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
20 ZIANO AND WANG
ences about the response and the responder. The present study
shows that one such inference is sincerity—people have a
tendency to believe that response delays are the result of the
responder suppressing another thought, and are a reflection of
insincerity. We also show that people do not simply rely on
response speed as a proxy to sincerity in a dogmatic and
one-dimensional way, rather, people seem to have a sophisti-
cated understanding of what response speed represents and
what it does not represent in different contexts. Nevertheless,
our research shows that, on the whole, a fast response seems to
be perceived as more sincere, while a response that is delayed
for even a couple of seconds may be considered a slow lie.
References
Baskett, G. D., & Freedle, R. O. (1974). Aspects of language pragmatics
and the social perception of lying. Journal of Psycholinguistic Research,
3(2), 117–131. https://doi.org/10.1007/BF01067571
Bernstein, M. J., Sacco, D. F., Brown, C. M., Young, S. G., & Claypool,
H. M. (2010). A preference for genuine smiles following social exclu-
sion. Journal of Experimental Social Psychology,46(1), 196–199.
https://doi.org/10.1016/j.jesp.2009.08.010
Boltz, M. G., Dyer, R. L., & Miller, A. R. (2010). Are you lying to me?
Temporal cues for deception. Journal of Language and Social Psychol-
ogy,29(4), 458–466. https://doi.org/10.1177/0261927X10385976
Boltz, M. G. (2005). Temporal dimensions of conversational interaction
the role of response latencies and pauses in social impression formation.
Journal of Language and Social Psychology,24(2), 103–138. https://doi
.org/10.1177/0261927X05275734
Bonnefon, J-F., & Villejoubert, G. (2006). Tactful or doubtful? Psycho-
logical Science,17(9), 747–751. https://doi.org/10.1111/j.1467-9280
.2006.01776.x
Boyd, R., Gintis, H., Bowles, S., & Richerson, P. J. (2003). The evolution
of altruistic punishment. Proceedings of the National Academy of Sci-
ences,100(6), 3531–3535. https://doi.org/10.1073/pnas.0630443100
Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in
language usage. Cambridge University Press. https://doi.org/10.1017/
CBO9780511813085
Buller, D. B., Strzyzewski, K. D., & Hunsaker, F. G. (1991). Interpersonal
deception: II. The inferiority of conversational participants as deception
detectors. Communication Monographs,58, 25–40. https://doi.org/10
.1080/03637759109376212
Caza, A., Zhang, G., Wang, L., & Bai, Y. (2015). How do you really feel?
Effect of leaders’ perceived emotional sincerity on followers’ trust. The
Leadership Quarterly,26(4), 518–531. https://doi.org/10.1016/j.leaqua
.2015.05.008
Christ, S. E., Van Essen, D. C., Watson, J. M., Brubaker, L. E., &
McDermott, K. B. (2009). The contributions of prefrontal cortex and
executive control to deception: Evidence from activation likelihood
estimate meta-analyses. Cerebral Cortex,19(7), 1557–1566. https://doi
.org/10.1093/cercor/bhn189
Cohen, J. (1992). A power primer. Psychological Bulletin,112(1), 155–
159. https://doi.org/10.1037/0033-2909.112.1.155
Cornetto, K. M. (2002). Identity and illusion on the Internet: Interpersonal
deception and detection in interactive Internet environments (PhD The-
sis). University of Texas at Austin.
Critcher, C. R., Inbar, Y., & Pizarro, D. A. (2013). How quick decisions
illuminate moral character. Social Psychological and Personality Sci-
ence,4(3), 308–315. https://doi.org/10.1177/1948550612457688
Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability
independent of psychopathology. Journal of Consulting Psychology,
24(4), 349–354. https://doi.org/10.1037/h0047358
De Houwer, J., Thomas, S., & Baeyens, F. (2001). Associative learning of
likes and dislikes: A review of 25 years of research on human evaluative
conditioning. Psychological Bulletin,127(6), 853–869. https://doi.org/
10.1037/0033-2909.127.6.853
Dietvorst, B. J., & Simonsohn, U. (2019). Intentionally “biased”: People
purposely use to-be-ignored information, but can be persuaded not to.
Journal of Experimental Psychology: General,148(7), 1228–1238.
https://doi.org/10.1037/xge0000541
Doerner, J. K., & Demuth, S. (2014). Gender and sentencing in the federal
courts: Are women treated more leniently? Criminal Justice Policy
Review,25(2), 242–269. https://doi.org/10.1177/0887403412466877
Duran, N. D., Dale, R., & McNamara, D. S. (2010). The action dynamics
of overcoming the truth. Psychonomic Bulletin & Review,17(4), 486
491. https://doi.org/10.3758/PBR.17.4.486
Efendic´, E., van de Calseyde, P. P. F. M., & Evans, A. M. (2020). Slow
response times undermine trust in algorithmic (but not human) predic-
tions. Organizational Behavior and Human Decision Processes,
157(January), 103–114. https://doi.org/10.1016/j.obhdp.2020.01.008
Evans, A. M., & Rand, D. G. (2019). Cooperation and decision time.
Current Opinion in Psychology,26, 67–71. https://doi.org/10.1016/j
.copsyc.2018.05.007
Evans, A. M., & van de Calseyde, P. P. F. M. (2017). The effects of
observed decision time on expectations of extremity and cooperation.
Journal of Experimental Social Psychology,68, 50–59. https://doi.org/
10.1016/j.jesp.2016.05.009
Feldman, G., Lian, H., Kosinski, M., & Stillwell, D. (2017). Frankly, we do
give a damn. Social Psychological & Personality Science,8(7), 816
826. https://doi.org/10.1177/1948550616681055
Fish, K., Rothermich, K., & Pell, M. D. (2017). The sound of (in)sincerity.
Journal of Pragmatics,121, 147–161. https://doi.org/10.1016/j.pragma
.2017.10.008
Gawronski, B., & Bodenhausen, G. V. (2006). Associative and proposi-
tional processes in evaluation: An integrative review of implicit and
explicit attitude change. Psychological Bulletin,132(5), 692–731.
https://doi.org/10.1037/0033-2909.132.5.692
Harrison, A. A., Hwalek, M., Raney, D. F., & Fritz, J. G. (1978). Cues to
Deception in an Interview Situation. Social Psychology,41(2), 156–161.
https://doi.org/10.2307/3033575
Hartwig, M., & Bond, C. F. (2011). Why do lie-catchers fail? A lens model
meta-analysis of human lie judgments. Psychological Bulletin,137(4),
643–659. https://doi.org/10.1037/a0023589
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in
the world? Behavioral and Brain Sciences,33(2–3), 61–83. https://doi
.org/10.1017/S0140525X0999152X
Holtgraves, T. (2004). Social desirability and self-reports: Testing models
of socially desirable responding. Personality and Social Psychology
Bulletin,30(2), 161–172. https://doi.org/10.1177/0146167203259930
Hornsey, M. J., Wohl, M. J. A., Harris, E. A., Okimoto, T. G., Thai, M., &
Wenzel, M. (2020). Embodied remorse: Physical displays of remorse
increase positive responses to public apologies, but have negligible
effects on forgiveness. Journal of Personality and Social Psychology,
119(2), 367–389. https://doi.org/10.1037/pspi0000208
Hütter, M., & Sweldens, S. (2018). Dissociating controllable and uncon-
trollable effects of affective stimuli on attitudes and consumption. Jour-
nal of Consumer Research,45(2), 320–349. https://doi.org/10.1093/jcr/
ucx124
Jay, T. (2009). The utility and ubiquity of taboo words. Perspectives on
Psychological Science,4(2), 153–161. https://doi.org/10.1111/j.1745-
6924.2009.01115.x
Judd, C. M., Westfall, J., & Kenny, D. A. (2012). Treating stimuli as a
random factor in social psychology: A new and comprehensive solution
to a pervasive but largely ignored problem. Journal of Personality and
Social Psychology,103(1), 54–69. https://doi.org/10.1037/a0028347
Kassin, S. M., & Sommers, S. R. (1997). Inadmissible testimony, instruc-
tions to disregard, and the jury: Substantive versus procedural consid-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
21
SLOW LIES
erations. Personality and Social Psychology Bulletin,23(10), 1046
1054. https://doi.org/10.1177/01461672972310005
Kraut, R. E. (1978). Verbal and nonverbal cues in the perception of lying.
Journal of Personality and Social Psychology,36(4), 380–391. https://
doi.org/10.1037/0022-3514.36.4.380
Kraut, R. E., & Lewis, S. H. (1982). Person perception and self-awareness:
Knowledge of influences on one’s own judgments. Journal of Person-
ality and Social Psychology,42(3), 448–460. https://doi.org/10.1037/
0022-3514.42.3.448
Kupor, D. M., Tormala, Z. L., Norton, M. I., & Rucker, D. D. (2014).
Thought calibration: How thinking just the right amount increases one’s
influence and appeal. Social Psychological and Personality Science,
5(3), 263–270. https://doi.org/10.1177/1948550613499940
Lambdin, C., & Shaffer, V. a. (2009). Are within-subjects designs trans-
parent? Judgment and Decision Making,4(7), 554–566. https://doi.org/
10.1037/e722352011-194
Lee, J. J., & Pinker, S. (2010). Rationales for indirect speech: The theory
of the strategic speaker. Psychological Review,117(3), 785–807. https://
doi.org/10.1037/a0019688
Lerner, G. H. (1991). On the syntax of sentences-in-progress. Language in
Society,20(3), 441–458. https://doi.org/10.1017/S0047404500016572
Lerner, G. H. (1996). Finding “face” in the preference structures of
talk-in-interaction. Social Psychology Quarterly,59(4), 303–321.
https://doi.org/10.2307/2787073
Lewicki, R. J., & Bunker, B. B. (1996). Developing and maintaining trust
in work relationships. In R. Kramer & T. Tyler (Eds.), Trust in organi-
zations: Frontiers of theory and research (pp. 114–139). SAGE Publi-
cations. https://doi.org/10.4135/9781452243610.n7
Lykken, D. T. (1959). The GSR in the detection of guilt. Journal of Applied
Psychology,43(6), 385–388. https://doi.org/10.1037/h0046060
Mann, S., Vrij, A., & Bull, R. (2002). Suspects, lies, and videotape: An
analysis of authentic high-stake liars. Law and Human Behavior,26(3),
365–376. https://doi.org/10.1023/A:1015332606792
Meyer, D. E., & Schvaneveldt, R. W. (1976). Meaning, memory structure,
and mental processes. Science,192(4234), 27–33. https://doi.org/10
.1126/science.1257753
Meyvis, T., & Van Osselaer, S. M. J. (2018). Increasing the power of your
study by increasing the effect size. Journal of Consumer Research,
44(5), 1157–1173. https://doi.org/10.1093/jcr/ucx110
Mick, D. G. (1996). Are studies of dark variables confounded by socially
desirable responding? The case of materialism. Journal of Consumer
Research,23(2), 106–119. https://doi.org/10.1086/209470
Milinski, M., Semmann, D., & Krambeck, H-J. (2002). Reputation helps
solve the ‘tragedy of the commons’. Nature,415, 424–426. https://doi
.org/10.1038/415424a
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A.,
& Wager, T. D. (2000). The unity and diversity of executive functions
and their contributions to complex “frontal lobe” tasks: A latent variable
analysis. Cognitive Psychology,41(1), 49–100. https://doi.org/10.1006/
cogp.1999.0734
Montoya, A., & Hayes, A. F. (2017). Two-condition within-participant
statistical mediation analysis: A path-analytic framework. Psychological
Methods,22(1), 6–27. https://doi.org/10.1037/met0000086
Pieters, R. (2017). Meaningful mediation analysis: Plausible causal infer-
ence and informative communication. Journal of Consumer Research,
44(3), 692–716. https://doi.org/10.1093/jcr/ucx081
Pomerantz, A. (1984). Agreeing and disagreeing with assessments: Some
features of preferred/dispreferred turn shapes. In J. M. Atkinson & J.
Heritage (Eds.), Structures of social action (pp. 57–101). Cambridge
University Press. https://doi.org/10.1017/cbo9780511665868.008
Rassin, E., & Van Der Heijden, S. (2005). Appearing credible? Swearing
helps! Psychology, Crime & Law,June(11), 117–182. https://doi.org/10
.1080/106831605160512331329952
Richard, F. D., Bond, C. F., Jr., & Stokes-Zoota, J. J. (2003). One hundred
years of social psychology quantitatively described. Review of General
Psychology,7(4), 331–363. https://doi.org/10.1037/1089-2680.7.4.331
Rosenblum, M., Schroeder, J., & Gino, F. (2020). Tell it like it is: When
politically incorrect language promotes authenticity. Journal of Person-
ality and Social Psychology,119(1), 75–103. https://doi.org/10.1037/
pspi0000206
Scopelliti, I., Min, H. L., McCormick, E., Kassam, K. S., & Morewedge,
C. K. (2017). Individual differences in correspondence bias: Measure-
ment, consequences, and correction of biased interpersonal attributions.
Management Science,64(4), 1477–1973. https://doi.org/10.1287/mnsc
.2016.2668
Seymour, T. L., & Kerlin, J. R. (2008). Successful detection of verbal and
visual concealed knowledge using an RT-based paradigm. Applied Cog-
nitive Psychology,22(4), 475–490. https://doi.org/10.1002/acp.1375
Seymour, T. L., & Schumacher, E. H. (2009). Electromyographic evidence
for response conflict in the exclude recognition task. Cognitive, Affective
& Behavioral Neuroscience,9(1), 71–82. https://doi.org/10.3758/
CABN.9.1.71
Seymour, T. L., Seifert, C. M., Shafto, M. G., & Mosmann, A. L. (2000).
Using response time measures to assess “guilty knowledge.” Journal of
Applied Psychology,85(1), 30–37. https://doi.org/10.1037/0021-9010
.85.1.30
Simonsohn, U. (2015). No-way interactions. The Winnower. https://doi
.org/10.15200/winn.142559.90552
Spence, S. A., Farrow, T. F. D., Herford, A. E., Wilkinson, I. D., Zheng,
Y., & Woodruff, P. W. R. (2001). Behavioural and functional anatomical
correlates of deception in humans. NeuroReport,12(13), 2849–2853.
https://doi.org/10.1097/00001756-200109170-00019
Stiff, J. B., & Miller, G. R. (1986). “Come to think of it . . . ” interrogative
probes, deceptive communication, and deception detection. Human
Communication Research,12(3), 339–357. https://doi.org/10.1111/j
.1468-2958.1986.tb00081.x
Suchotzki, K., Crombez, G., Debey, E., van Oorsouw, K., & Verschuere,
B. (2015). In vino veritas? Alcohol, response inhibition and lying.
Alcohol and Alcoholism,50(1), 74–81. https://doi.org/10.1093/alcalc/
agu079
Suchotzki, K., Verschuere, B., Van Bockstaele, B., Ben-Shakhar, G., &
Crombez, G. (2017). Lying takes time: A meta-analysis on reaction time
measures of deception. Psychological Bulletin,143(4), 428–453. https://
doi.org/10.1037/bul0000087
Swinkels, A. H. (1991). The effects of cognitive busyness on human lie
detection ability [Unpublished doctoral dissertation, University of Texas at
Austin]. Retrieved from https://www.elibrary.ru/item.asp?id5878227
Tanaka, H. (2008). Delaying dispreferred responses in English: From a
Japanese perspective. Language in Society,37(4), 487–513. https://doi
.org/10.1017/S0047404508080743
Trivers, B. Y. R. L. (1971). The evolution of reciprocal altruism. The
Quarterly Review of Biology,46(1), 35–57. https://doi.org/10.1086/
406755
Van Boven, L., Campbell, M. C., & Gilovich, T. (2010). Stigmatizing
materialism: On stereotypes and impressions of materialistic and expe-
riential pursuits. Personality and Social Psychology Bulletin,36(4),
551–563. https://doi.org/10.1177/0146167210362790
Van de Calseyde, P. P. F. M., Keren, G., & Zeelenberg, M. (2014).
Decision time as information in judgment and choice. Organizational
Behavior and Human Decision Processes,125(2), 113–122. https://doi
.org/10.1016/j.obhdp.2014.07.001
Verschuere, B., Crombez, G., Degrootte, T., & Rosseel, Y. (2010). De-
tecting concealed information with reaction times: Validity and compar-
ison with the polygraph. Applied Cognitive Psychology,24(September
2009), 991–1002. https://doi.org/10.1002/acp
Vrij, A., Edward, K., & Bull, R. (2001). Police officers’ ability to detect
deceit: The benefit of indirect deception detection measures. Legal and
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
22 ZIANO AND WANG
Criminological Psychology,6(2), 185–196. https://doi.org/10.1348/
135532501168271
Vrij, A., Evans, H., Akehurst, L., & Mann, S. (2004). Rapid judgements in
assessing verbal and nonverbal cues: Their potential for deception re-
searchers and lie detection. Applied Cognitive Psychology,18(3), 283–
296. https://doi.org/10.1002/acp.964
Vrij, A., & Mann, S. (2001). Telling and detecting lies in a high-stake
situation: The case of a convicted murderer. Applied Cognitive Psychol-
ogy,15(2), 187–203. https://doi.org/10.1002/1099-0720(200103/04)15:
2187::AID-ACP6963.0.CO;2-A
Walczyk, J. J., Roper, K. S., Seemann, E., & Humphrey, A. M. (2003).
Cognitive mechanisms underlying to questions: Response time as a cue
to deception. Applied Cognitive Psychology,17(7), 755–774. https://doi
.org/10.1002/acp.914
Wang, D., Chatzisarantis, N. L. D., & Hagger, M. S. (2018). Why distrac-
tors with need-supportive content can mitigate ironic effects of thought
suppression. Motivation and Emotion,42, 214–224. https://doi.org/10
.1007/s11031-017-9653-3
Ziano, I. (2020, November 12). Slow lies data, analyses, and materials.
Retrieved from osf.io/pqmz2.
Zuckerman, M., Koestner, R., & Driver, R. (1981). Beliefs about cues
associated with deception. Journal of Nonverbal Behavior,6(1), 105–
114. https://doi.org/10.1007/BF00987286
Received May 23, 2020
Revision received November 17, 2020
Accepted November 30, 2020
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
23
SLOW LIES
... For instance, making people aware of the importance of depression prototypes on classification of a person as depressed might reduce their overreliance on personal features. In addition, explicitly instructing people to ignore personal features and focus on the symptoms alone may also moderate the effect of depression prototypes on classification as depression (this method was used to reduce people's reliance on extraneous cues in Dietvorst &Simonsohn, 2018 andZiano &Wang, 2021). This paper also aims to test these ideas. ...
... For instance, making people aware of the importance of depression prototypes on classification of a person as depressed might reduce their overreliance on personal features. In addition, explicitly instructing people to ignore personal features and focus on the symptoms alone may also moderate the effect of depression prototypes on classification as depression (this method was used to reduce people's reliance on extraneous cues in Dietvorst &Simonsohn, 2018 andZiano &Wang, 2021). This paper also aims to test these ideas. ...
... Then, while participants in the control condition proceeded to completing the measures in the study, only participants in the "instructions" condition were presented with the following instructions, in a procedure drawn from the literature about judgments of others (Dietvorst & Simonsohn, 2018;Ziano & Wang, 2021): ...
Preprint
Full-text available
This paper investigates the content and the consequences of the prototypes of the depressed. Twelve preregistered studies (total N = 4573, with U.S. American, British, and French participants) show that laypeople consider depressed people as having specific psychological, social, and physical features compared to people described as non-depressed or to a control condition. Target prototypicality (as opposed to the clinical symptoms) has a strong influence on the degree to which laypeople believe others are actually depressed, how much depression-like symptoms cause others to experience psychological pain, and how much professional mental healthcare is appropriate for others. This effect was not reduced by instructing people to focus on the symptoms and ignore the target features, and somewhat reduced by informing them of the effect. We discuss theoretical implications for the understanding of prototypes of depressed people and practical implications for its alleviation.
... Studies have shown that potential biases in judgement could be at least partially corrected when participants are instructed to deliberately disregard various cues or information that could lead to biases in the judgement process (Dietvorst & Simonsohn, 2018;Ziano & Wang, 2021). In this context, we therefore predict that by explicitly asking people to ignore a target individual's personal features and focus on the event instead, the effect of target prototypicality on ostracism recognition, perceptions of psychological pain, behavior intentionality, and blame on ostracizers would diminish. ...
Preprint
Full-text available
Do people imagine ostracized individuals a certain way and are they perceived differently depending on their conformity to this prototype? This was investigated in five main and five supplementary studies (all preregistered; total N = 3789) which showed that people imagine ostracized individuals as socially, psychologically, and physically different from the average person. This has significant consequences. The same act directed at prototypical (vs. nonprototypical) targets is more likely to be classified as ostracism, even when people try to ignore target features and focus on the behavior alone. Prototypical targets are seen as less deserving, more hurt, and are sympathized with more. Further, people judge alleged ostracizers more negatively and express less interest in affiliating with them when the target fits the prototype. We discuss theoretical implications for the understanding of prototypes and ostracism and practical implications for the detection and curbing of ostracism in everyday life.
... Observers may have implicitly learned that response time and connection covary. This finding extends prior work demonstrating that outside observers infer another's preferences (32), sincerity (33), and certainty (34) by how many seconds they take to make a decision between available options: Faster decisions appear to express stronger or "truer" preferences. Response time in natural conversation, on the order of milliseconds, may similarly be taken as a true signal of connection. ...
Article
Clicking is one of the most robust metaphors for social connection. But how do we know when two people "click"? We asked pairs of friends and strangers to talk with each other and rate their felt connection. For both friends and strangers, speed in response was a robust predictor of feeling connected. Conversations with faster response times felt more connected than conversations with slower response times, and within conversations, connected moments had faster response times than less-connected moments. This effect was determined primarily by partner responsivity: People felt more connected to the degree that their partner responded quickly to them rather than by how quickly they responded to their partner. The temporal scale of these effects (<250 ms) precludes conscious control, thus providing an honest signal of connection. Using a round-robin design in each of six closed networks, we show that faster responders evoked greater feelings of connection across partners. Finally, we demonstrate that this signal is used by third-party listeners as a heuristic of how well people are connected: Conversations with faster response times were perceived as more connected than the same conversations with slower response times. Together, these findings suggest that response times comprise a robust and sufficient signal of whether two minds “click.”
... People can make surprising inferences about attributes expressed in numerical forms, which may vary with the magnitude of such attributes. This includes numerical measures of response speed (Ziano & Wang, 2021), product attributes expressed in quantities (Evangelidis & van Osselaer, 2017), and the percentage of people that make the same consumer choices (Ziano & Pandelaere, 2018). People may also make inferences about the total number of people in danger in a specific situation. ...
Article
Full-text available
Is it better to save 4500 lives out of 11,000 or 4500 lives out of 250,000? Fetherstonhaugh et al. (1997) showed that people prefer the former: to save lives if they are a higher proportion of the total, a phenomenon they termed “psychophysical numbing”. We attempted to replicate Studies 1 and 2 of Fetherstonhaugh et al. (1997) (5 data collections, total N = 4799, MTurk and Prolific, USA and UK), and added several extensions (e.g., donation amounts, procedural differences, and individual-level ideology and knowledge). We found mixed support, with two successful replications of Study 2 that indeed showed psychophysical numbing (original: η²p = 0.55, 90% CI [0.45, 0.62], Study 2a: η²p = 0.62, 90% CI [0.58, 0.66], Study 2b: η²p = 0.24, 90% CI [0.21, 0.27], all in same direction), yet also three unsuccessful replications of Study 1 showing instead an opposite psychophysical sensitization, a preference for saving a smaller proportion of lives (original effect size: η²p = 0.14, 90% CI [0.02, 0.28], replications: Study 1a: η²p = 0.06, 90% CI [0.02, 0.10], Study 1b: η²p = 0.21, 90% CI [0.17, 0.26]; Study 1c: η²p = 0.13, 90% CI [0.08, 0.17], all in the opposite direction). We discuss theoretical implications and potential drivers of psychophysical numbing and sensitization, including evaluation mode, comparison procedure, ideology, knowledge, and prioritizing of one's own country, and practical implications for research on perceptions of charity, aid effectiveness, and donations. Materials, preregistrations, data, and analyses are available at https://osf.io/786jg/.
... People can make surprising inferences about attributes expressed in numerical forms, which may vary with the magnitude of such attributes. This includes numerical measures of response speed (Ziano & Wang, 2021), product attributes expressed in quantities (Evangelidis & van Osselaer, 2017), and the percentage of people that make the same consumer choices (Ziano & Pandelaere, 2018). People may also make inferences about the total number of people in danger in a specific situation. ...
Preprint
Full-text available
Is it better to save 4,500 lives out of 11,000 or 4,500 lives out of 250,000? Fetherstonhaugh ‎et al. (1997) showed that people prefer the former: to save lives if they are a higher ‎proportion of the total, a phenomenon they termed “psychophysical numbing”. We ‎attempted to replicate Studies 1 and 2 of Fetherstonhaugh et al. (1997) (5 data collections, ‎total N = 4799, MTurk and Prolific, USA and UK), and added several extensions (e.g., ‎donation amounts, procedural differences, and individual-level ideology and knowledge). ‎We found mixed support, with two successful replications of Study 2 that indeed showed ‎psychophysical numbing (original: η2p = 0.55, 90% CI [0.45, 0.62], Study 2a: η2p = 0.62, 90% ‎CI [0.58, 0.66], Study 2b: η2p = 0.24, 90% CI [0.21, 0.27], all in same direction), yet also ‎three unsuccessful replications of Study 1 showing instead an opposite psychophysical ‎sensitization, a preference for saving a smaller proportion of lives (original effect size: η2p = ‎‎0.14, 90% CI [0.02, 0.28], replications: Study 1a: η2p = 0.06, 90% CI [0.02, 0.10], Study 1b: ‎η2p = 0.21, 90% CI [0.17, 0.26]; Study 1c: η2p = 0.13, 90% CI [0.08, 0.17], all in the opposite ‎direction). We discuss theoretical implications and potential drivers of psychophysical ‎numbing and sensitization, including evaluation mode, comparison procedure, ideology, ‎knowledge, and prioritizing of one’s own country, and practical implications for research on ‎perceptions of charity, aid effectiveness, and donations. Materials, preregistrations, data, ‎and analyses are available at https://osf.io/786jg/. ‎
Article
People constantly and effortlessly acquire information about one another’s decisions, and use this information to form impressions (and judgments) of others. We review research on this process of choice perception—how people come to make sense of others’ choices. We suggest that choice perception consists of observers’ inferences about (a) what was chosen, (b) why it was chosen, (c) how (or through what process) it was chosen, and (d) broader impressions about who chose it. These inferences can affect observers in multiple ways—including erroneous beliefs about the actor due to interpersonal errors (i.e., mistakes in how observers perceive actors) and cue-perception errors (i.e., mistakes in how observers perceive chosen options), as well as changes in one’s own behavior.
Article
Past research has uncovered actions that would seem to undermine but in fact frequently enhance persuasion. For example, expressing doubt about one’s view or presenting arguments against it would seem to weaken one’s case, but can sometimes promote it. We propose a framework for understanding these findings. We posit that these actions constitute acts of receptiveness—behaviors that signal openness to new information and opposing viewpoints. We review four classes of acts of receptiveness: conveying uncertainty, acknowledging mistakes, highlighting drawbacks, and asking questions. We identify conditions under which and mechanisms through which these actions boost persuasion. Acts of receptiveness appear to be more persuasive when they come from expert or high-status sources, rather than non-expert or low-status sources, and to operate through two primary mechanisms: increased involvement and enhanced source perceptions. Following a review of this work, we delineate potentially novel acts of receptiveness and outline directions for future research.
Article
Full-text available
Algorithms consistently perform well on various prediction tasks, but people often mistrust their advice. Here, we demonstrate one component that affects people’s trust in algorithmic predictions: response time. In seven studies (total N = 1928 with 14,184 observations), we find that people judge slowly generated predictions from algorithms as less accurate and they are less willing to rely on them. This effect reverses for human predictions, where slowly generated predictions are judged to be more accurate. In explaining this asymmetry, we find that slower response times signal the exertion of effort for both humans and algorithms. However, the relationship between perceived effort and prediction quality differs for humans and algorithms. For humans, prediction tasks are seen as difficult and observing effort is therefore positively correlated with the perceived quality of predictions. For algorithms, however, prediction tasks are seen as easy and effort is therefore uncorrelated to the quality of algorithmic predictions. These results underscore the complex processes and dynamics underlying people’s trust in algorithmic (and human) predictions and the cues that people use to evaluate their quality.
Article
Full-text available
As in other social sciences, published findings in consumer research tend to overestimate the size of the effect being investigated, due to both file drawer effects and abuse of researcher degrees of freedom, including opportunistic analysis decisions. Given that most effect sizes are substantially smaller than would be apparent from published research, there has been a widespread call to increase power by increasing sample size. We propose that, aside from increasing sample size, researchers can also increase power by boosting the effect size. If done correctly, removing participants, using covariates, and optimizing experimental designs, stimuli, and measures can boost effect size without inflating researcher degrees of freedom. In fact, careful planning of studies and analyses to maximize effect size is essential to be able to study many psychologically interesting phenomena when massive sample sizes are not feasible. © The Author 2017. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved.
Article
Full-text available
This research studies a fundamental and seemingly straightforward question: can basic advertising elements such as the presence of attractive imagery have uncontrollable effects on consumers’ attitudes and consumption decisions? Answering this question is methodologically challenging, because the presence of an uncontrollable process can be masked by a simultaneously operating controllable process. We argue first that existing methods conflate the contribution of both processes and are therefore unable to measure the presence of an uncontrollable process reliably. To solve the conundrum, we present a novel application of processing tree modeling. Evaluative conditioning is employed as a paradigm to study the influence of affective visual stimuli on attitudes and behavior. Across six experiments, we demonstrate the validity of the model parameters estimating controllable and uncontrollable processes. As predicted, the parameter estimate of the controllable process is susceptible to cognitive resources and levels of motivation to exert control. The parameter estimate of the uncontrollable process appears unaffected by these factors. We also demonstrate the external validity of our findings and their relevance to stimuli and instructions typical for consumer research. Finally, we find that controllable and uncontrollable processes are both predictive of product choice and consumption. We discuss implications for consumer protection.
Article
Public apologies struggle to communicate genuineness. Previous studies have shown that, in response to public apologies, perceptions of remorse and levels of forgiveness are often low, while skepticism about motive is high. Furthermore, attempts to reduce mistrust of public apologies by manipulating the verbal component of the message have had limited success. Across 6 studies (combined N = 3,818), we examined whether people respond more positively to public apologies if the apologies are accompanied by nonverbal demonstrations of remorse: operationalized as kneeling (Studies 1 and 6) and crying (Studies 2-5). Overall, embodied remorse had small-to-medium effects on perceived remorse, and through this relationship had reliable effects on perceived likelihood of reoffending, empathy, positive appraisals of the transgressor, and satisfaction with the apology. Positive effects of embodiment emerged regardless of whether transgressions were committed by a collective (Studies 1, 2, and 6) or an individual (Studies 3-5), and were equally strong regardless of whether or not the transgressor issued an apology (Studies 4 and 5). Furthermore, embodied remorse appeared to lie beyond suspicion: if anything, those low in dispositional trust were more positively influenced by embodied remorse than those high in dispositional trust. Despite all these positive effects, embodied remorse did not have a significant effect on forgiveness in any of the studies, and an internal meta-analysis revealed a significant effect that was of negligible size. (PsycINFO Database Record (c) 2019 APA, all rights reserved).
Article
When a person's language appears to be political-such as being politically correct or incorrect-it can influence fundamental impressions of him or her. Political correctness is "using language or behavior to seem sensitive to others' feelings, especially those others who seem socially disadvantaged." One pilot study, 6 experiments, and 3 supplemental experiments (N = 4,956) demonstrate that being politically incorrect makes communicators appear more authentic-specifically, less susceptible to external influence-albeit also less warm. These effects, however, are moderated by perceivers' political ideology and how sympathetic perceivers feel toward the target group being labeled politically correctly. In Experiments 1, 2, and 3 using politically incorrect language (e.g., calling undocumented immigrants illegals) made a communicator appear particularly authentic among conservative perceivers but particularly cold among liberal perceivers. However, in Experiment 4 these effects reversed when conservatives felt sympathetic toward the group that was being labeled politically correctly or incorrectly (e.g., calling poor Whites white trash). Experiment 5 tests why political incorrectness can boost authenticity, demonstrating that it makes communicators seem less strategic. Finally, Experiment 6 examines the use of political language in a meaningful field context: perceived persuasion in real political debates. Debaters instructed to be politically correct (vs. politically incorrect) were judged by their uninstructed conversation partners to be easier to persuade during the conversation, although they actually reported being similarly persuaded. Together, these findings demonstrate when and how using politically incorrect language can enhance a person's authenticity. (PsycINFO Database Record (c) 2019 APA, all rights reserved).
Article
Abundant research has shown that people fail to disregard to-be-ignored information (e.g., hindsight bias, curse of knowledge), which has contributed to the popular notion that people are unwillingly and unconsciously affected by information. Here we provide evidence that, instead, people simply do not want to ignore such information. The findings: In Studies 1 and 2, the majority of participants explicitly indicated a desire to use to-be-ignored information in classic paradigms. In Study 3, the effect of receiving to-be-ignored information was driven entirely by the subset of people who wanted to use it. In Study 4, persuading participants to ignore inadmissible evidence in a mock jury trial reduced the impact of such evidence.
Article
Evaluative conditioning refers to changes in the liking of a stimulus that are due to the fact that the stimulus has been paired with other, positive or negative stimuli. Although evaluative conditioning appears to be subjected to certain boundary conditions, significant evaluative conditioning effects have been obtained using a large variety of stimuli and procedures. Some data suggest that evaluative conditioning can occur under conditions that do not support other forms of Pavlovian conditioning, and several models have been proposed to account for these differences. In the present article, the authors summarize the available literature, draw conclusions where possible, and provide suggestions for future research.