BACKGROUND
People are not very good at detecting others’ lies. Their accuracy in lab-based veracity judgment tasks (judging whether given information is truthful or not) is often at chance level (Bond & DePaulo, 2006). It is because humans have a default cognition that believes others are honest (i.e., truth-default cognition; Levine, 2014; Levine et al., 2021), which makes them reluctant to judge others as lying (i.e., making lie judgments). However, making lie judgments is essential to detect deception. So, the present research investigated individual and situational characteristics that lead people to make lie judgments.
Past studies, especially those related to an influential theory on deception, the truth-default theory (TDT) by Levine (2014), have identified that how suspicious people are is a crucial factor for making lie judgments. For example, professionals of lie detection (e.g., police detectives and judges) who often deal with suspicious information and individuals make more lie judgments than laypeople (Hurst & Oswald, 2012; Masip et al., 2016). The suspicion highlighted by the TDT is called state suspicion, and it is often elicited by specific cues available to the observer. These cues can be embedded in the information contents or the surrounding context. They can also be provided to the observer directly (e.g., observers are told the probability of particular information being a lie). People in situations eliciting high state suspicion tend to make more lie judgments (Kim & Levine, 2011). Numerous studies have investigated the effects of state suspicion on lie judgments (Burgoon et al., 1996; Hubbell et al., 2001; Kim & Levine, 2011; Stiff et al., 1992). For example, Stiff et al. (1992) asked their participants to assess the speaker’s credibility. They compared the results for the condition where the participants were told that the speaker was likely to lie and when they were not.
However, state suspicion cannot fully explain people’s veracity judgments because individual differences related to suspicion also play a role. Levine and McCornack (1991) conducted the first major research focused on an individual’s stable characteristic of being suspicious of others, referred to as dispositional suspicion. To measure this tendency, they developed the Generalized Communicative Suspicion (GCS) scale, which has been used in various research. The GCS appears to affect a wide range of behaviors, such as judgments of honesty (Levine & McCornack, 1991), voting decisions (Zhou et al., 2017), and responses to persuasive messages (Vishwanath et al., 2018). Generally, individuals with high GCS are expected to make more lie judgments than those with low GCS (Levine & McCornack, 1991). Indeed, this has been found in some studies (Bond et al., 2005), but it is not always replicated (Bond & Lee, 2005). Hurst and Oswald (2012) observed a significant relationship between GCS and deceptiveness perception for undergraduates and novice police officers but not for experienced officers. These inconsistent findings may stem from the varying extent of state suspicion across studies. Veracity judgments may result from the interaction between the observer’s level of dispositional suspicion and the degree of state suspicion aroused by the surrounding contexts. Therefore, it is important to separate the two types of suspicion.
To our knowledge, the study by Levine and McCornack (1991) is the only one that has examined the effect of suspicion on veracity judgments by clearly distinguishing the two types of suspicion. They reported that participants with higher GCS made more lie judgments when state suspicion was low or moderate. However, such a relationship was weaker under high state suspicion. Based on this finding, the present study distinguished two types of suspicion to examine their associations with veracity judgments. Some methodological limitations of Levine and McCornack’s study (1991), such as their deceptive messages not being spontaneously created by the message sender, were more strictly controlled in the present study.
In addition, the current study investigated whether the amount of information individuals gather before reaching a veracity judgment differs depending on the state or dispositional suspicion. A high level of state suspicion is expected to encourage people to gather more information (Sinaceur, 2010). However, the relationship between dispositional suspicion and information gathering is unclear. Whereas some studies have suggested that high levels of dispositional suspicion are related to more information gathering, others have found the opposite pattern. Professionals seem to make more rapid judgments rather than gathering much information, and more attention is paid to any information confirming the suspect’s lying (Nahari, 2012). It could be related to the fact that the information experts deal with is often embedded in a highly suspicious context. Thus, people with high dispositional suspicion may employ different strategies to gather information depending on the level of state suspicion. They may focus on a small amount of information under a situation with high state suspicion while trying to gather more information under a situation with low state suspicion. Alternatively, the pattern can be reversed.
Participants in this study watched a video in which a person talked about their everyday personal interactions and then determined whether the speaker was telling the truth or not. After dispositional suspicion was measured, participants were divided into two groups to alter their level of state suspicion. Participants in a suspicious condition were told that the speakers in the video may have been involved in a crime, whereas those in a non-suspicious condition were not. The participants assessed the speakers twice for each video; the first was an assessment of the perceived deceptiveness, which took place just after viewing the first video, measuring the initial deceptiveness of the speaker. Then, participants made a dichotomous truth-lie judgment. Regarding the relationship between state suspicions and veracity judgments, it is assumed that the participants in the suspicious condition make more lie judgments than those in the non-suspicious condition. The supposed relationship between the GCS and lie judgments predicts that people with higher GCS make more lie judgments than those with lower GCS. If the two types of suspicion interact, as found by Levine and McCornack (1991), the relationship between GCS and lie judgments will be more prominent when the state suspicion is low. So, such a relationship was expected to be observed in the non-suspicious condition but not in the suspicious condition.
Although participants had to make a truth-lie judgment in the end, they were allowed to watch several other videos showing the same speaker before making the judgment. The number of viewed videos signified the amount of information participants gathered. The relationship between the amount of information gathering and GCS level is still unclear, so participants with higher GCS may watch more videos, or the results can be the opposite. If dispositional and state suspicions interact, people with higher GCS may prefer quick decision-making in the suspicious condition while gathering more information in the non-suspicious condition. On the other hand, if people with high GCS are generally more careful in gathering information, they will seek more information in the suspicious condition because the necessity to make lie judgments is high. Such an expectation is weaker in the non-suspicious condition; thus, they may reach a decision relatively quickly.
In summary, the present study aimed to reveal the relationship between dispositional and state suspicion as well as their impacts on lie judgments and information gathering. Although past research on deception underscored the significance of suspicion, little attention has been paid to the distinction between the two types of suspicion. Also, their effects on information-gathering behavior have rarely been investigated. Since the two types of suspicion should influence real-life situations involving lie judgments, the findings of this research may have useful practical applications.
PARTICIPANTS AND PROCEDURE
PARTICIPANTS
This study was conducted online. Two hundred and sixty respondents registered in a research company (Macromill Inc.) participated in the study (148 males, 111 females, and 1 other, mean age = 45.90, SD = 10.80). All participants gave informed consent. The required sample size calculated based on previous similar studies (Hurst & Oswald, 2012; Levine & McCornack, 1991) was 222 to obtain a moderate effect size (α = 0.05, power = 0.80, effect size = 0.15).
MEASURES
Three sets of measures were used in the study. The first was the Generalized Communicative Suspicion (GCS) scale developed by Levine and McCornack (1991). The original English version was translated into Japanese with permission. There were fourteen items measuring skeptical and distrusting attitudes toward others (e.g., “I often feel as if people are not completely truthful with me”), which were answered on a 5-point scale (1 – strongly disagree, 5 – strongly agree). The reliability of this scale in the present study was high (α = .81). For the analysis, all items were averaged to produce a GCS score for each participant, with higher scores indicating a higher level of dispositional suspicion.
The second set of measures contained manipulation check items, asking participants to indicate how guilty someone is of committing theft (1 – not guilty at all, 7 – very guilty) and how likely the suspects of theft are to lie when they are interviewed (1 – very unlikely, 7 – very likely). Since this study manipulated participants’ state suspicion using a story about theft, these items measured whether committing theft is viewed negatively and suspiciously.
The third set of measures was about potential factors affecting veracity judgments. The first item asked how important it is to make accurate veracity judgments in daily life (1 – not at all, 7 – very important). The second one asked how often they make veracity judgments (1 – not at all, 7 – very often). The third one asked whether they had been deceived in the past (once, many times, or not at all).
STIMULUS VIDEOS
The study used video clips in which a person talked about personal interactions. There was one male and one female speaker, and five videos were created for each of them (see Supplementary materials for the creation process). The videos’ contents were mundane and unrelated to crime. It was because emotion-provoking crime-related content can easily raise suspicion for anyone, possibly eliminating the effect of dispositional and state suspicion.
STATE SUSPICION MANIPULATION
There were suspicious and non-suspicious conditions, distinguished by informing (or not informing) participants that the speaker in the video might have committed a crime. This procedure is based on the study by Strömwall et al. (2003), who suggested that it is better than simply telling participants that the speaker might be deceptive. It is because the crime story infers the speaker’s motivation to lie, easily convincing participants that deception might be involved in what the speaker says.
A preliminary study (N = 17) was conducted, in which participants were asked to rate how likely a person involved in theft/fraud would tell lies, using a 7-point scale. The result (M = 4.94, SD = 1.18) was significantly higher than the midpoint on the scale, t(16) = 3.17, p = .006, confirming that people tend to believe that such a person would tell lies. Based on this result, we created a news article describing a theft/fraud crime against an older woman (see Supplementary materials) and showed it to the participants in the suspicious condition. After reading the article, those participants were told that some young people were captured by a security camera placed near the victim’s apartment, and the police interviewed them. The participants were informed that the videos they were about to see were part of the interview where the suspects discussed something unrelated to the criminal case. The story and the information about the video were not given to the participants in the non-suspicious condition.
PROCEDURE
Participants first answered the GCS scale and then received a brief explanation of the whole study procedure. They were told that each video depicted a person describing their thoughts about a common interpersonal situation and then presented with four practice trials in which participants viewed videos similar to those used in the main study (the practice videos were unrelated to the videos used in the main study). They indicated whether the speaker was truthful or deceptive after watching each video. Next, participants were randomly divided into two state suspicion conditions and received relevant instructions.
In the main part of the study, participants watched the first video and rated how deceptive he/she looked (1 – definitely telling the truth, 6 – definitely lying). Participants were then asked to decide whether to watch another video of the same speaker or make a final veracity judgment. When participants chose to watch another video, a new video appeared on the next page. Participants were allowed to view a maximum of five videos per speaker. Once they decided to make a final judgment or watched the fifth video, participants were asked to make truth-lie judgment. The number of videos each participant watched before reaching the truth-lie judgment was recorded and used to indicate the information-gathering behavior. The same procedure was then performed for the other speaker. After all the veracity judgments had been made, participants reported their age and gender. Also, they answered questions about the manipulation check, the possible factors affecting veracity judgments, and whether all videos played without problems. Finally, all participants were debriefed. This procedure was approved by the ethics committee of the author’s university.
Table 1
Descriptive statistics for the two conditions
RESULTS
THE ROLE OF DISPOSITIONAL SUSPICION IN THE DIFFERENT LEVELS OF STATE SUSPICION
Table 1 shows the mean and standard deviations of all the measurements. Some preliminary analyses were performed to check whether the state suspicion manipulation worked and whether the level of GCS differed between the two conditions (see Supplemental analysis 1 in Supplementary materials). None of the results should undermine the following main results.
A multiple regression analysis was performed to examine the association between dispositional and state suspicion and the deceptiveness ratings. The mean deceptiveness rating for each participant was the dependent variable, and the GCS score, the state suspicion condition, and their interaction were predictor variables. The model included age, the importance of accurate judgments, the frequency of making veracity judgments, the guiltiness of theft, the possibility of a suspect lying, and the experience of being deceived (a dummy variable) as control variables. This model was significant (R2 = .08, p = .012), and the two types of suspicion explained a small variance in deceptive ratings (∆R2 = .03). The result is illustrated in Figure 1A. There was a significant main effect of GCS (∆R2 = .03, β = .16, t = 2.51, p = .013), indicating that participants with higher GCS perceived the speakers as more deceptive than those with lower GCS. The effect of condition (∆R2 = .00, β = –.02, t = –0.38, p = .705) and the interaction were not significant (∆R2 = .01, β = –.10, t = –1.58, p = .116). Among the control variables, the experience of being deceived (β = .18, t = 2.47, p = .014) and the possibility of a suspect lying (β = .18, t = 2.15, p = .033) were positively associated with the deceptiveness ratings.
Although Figure 1A suggests a potential interaction, it was not statistically significant. Also, it is recommended that tests of the simple slope effect be performed in any regression analysis involving binary variables (Robinson et al., 2013). Therefore, a simple slope test was conducted to evaluate the effect of GCS for each condition. The effect was significant for the non-suspicious condition (β = .26, t = 3.04, p = .003) but not for the suspicious condition (β = .07, t = 0.72, p = .473).
An identical regression analysis was conducted for truth-lie judgments (see Figure 1B). The dependent variable was the number of lie judgments, ranging from 0 to 2. The mean number of total lie judgments was 0.53 (SD = 0.61). Most participants (n = 138, 53.1%) judged both speakers as truthful, and only 16 (6.2%) made lie judgments for both speakers. Due to the lower frequency of lie judgments, Poisson regression was used. The model itself (R2 = .03, p = .320), the effect of condition (∆R2 = .00, β = .07, Z = 1.11, p = .269), GCS (∆R2 = .00, β = –.03, Z = –0.45, p = .653), and the interaction were not significant (∆R2 = .00, β = –.05, Z = –0.92, p = .359). The only significant variable was the importance of accurate veracity judgments (β = –.11, Z = –2.13, p = .034).
THE EFFECT OF DISPOSITIONAL AND STATE SUSPICION ON THE AMOUNT OF INFORMATION GATHERING
We used the number of videos each participant watched before the final veracity judgment as an indicator of information-gathering behavior. The mean number of watched videos was 1.29 (SD = 0.66) across the two conditions. Only 28.8% of participants watched more than one video.
Figure 1
Interaction effect of GCS score and condition on deceptive ratings (A) and veracity judgments (B)
Note. GCS – Generalized Communicative Suspicion scale
A Poisson regression analysis was conducted on the number of watched videos. The main effect of GCS was marginally significant (∆R2 = .01, β = –.12, Z = –1.83, p = .068), suggesting that participants with higher GCS watched fewer videos than those with lower GCS. The effect of condition (∆R2 = .01, β = .12, Z = 1.63, p = .104) and the interaction (∆R2 = .01, β = –.07, Z = –1.22, p = .221) were not significant. Although the interaction was insignificant (see Figure 2), a simple slope analysis was conducted to further explore the association between GCS and information gathering in each state suspicion condition. The test revealed a marginally significant effect of GCS for the suspicious condition (β = –.19, Z = –1.83, p = .068), suggesting that participants with higher GCS tended to make lie judgments based on fewer videos than those with lower GCS in this condition. This was not true for the non-suspicious condition (β = –.04, Z = –0.63, p = .532).
Figure 2
Interaction effect of GCS score and condition on information gathering
Note. GCS – Generalized Communicative Suspicion scale
Additionally, two supplementary analyses were performed, and the results are reported in Supplemental analysis 2 and 3 (see Supplementary materials). The first analysis explored the association between the amount of information gathering and truth and lie judgments, and the second one examined the mediation effect of information gathering on the relationship between GCS and lie judgments. The results did not provide any robust findings with theoretical implications.
DISCUSSION
The present study investigated the association between dispositional suspicion and veracity judgments at different levels of state suspicion. The study found that individuals with a high level of GCS tended to hold stronger suspicion toward the speakers in the videos. However, the impact of GCS was present only for deceptiveness ratings and not for veracity judgments. Several past studies have also failed to detect an association between GCS and lie judgments (Bond & Lee, 2005; Masip et al., 2016), contradicting other studies that found that high GCS is associated with increased lie judgments (Bond et al., 2005; Levine & McCornack, 1991). This discrepancy reflects the importance of distinguishing deceptive ratings and veracity judgments, meaning distinguishing an early phase of suspicion formation about some information and a phase of making a final lie judgment about it. Hence, it is highly likely that the relationship between the initial suspicion and the final decision may change depending on how (e.g., asking about deceptiveness or the overall impression of the person) and when (e.g., a long time before the judgments or soon before them) the initial suspicion is measured. Indeed, it has been suggested that the initial suspicion is not maintained throughout the deception detection process (Street & Richardson, 2015). When exposed to information that initially raises suspicion, people seem to gradually return to their truth-default cognition if there is no additional reason to be suspicious. It is necessary to investigate how suspicious impressions change over time and how they lead to the final decision (Novotny et al., 2018).
Although the relationship between dispositional suspicion and lie judgments warrants further investigation, the present findings are valuable. In everyday life, suspicion towards others can be utilized for many behaviors (e.g., purchasing behaviors and voting choices) other than truthfulness judgments (see O’Sullivan, 2003). In a broader context, the psychological construct of suspicion is considered a determinant factor of decision-making in uncertain situations (Fein, 1996). Therefore, exploring the relationship between dispositional suspicion and various behaviors beyond veracity judgments is relevant to many areas of psychology.
Regarding the relationship between dispositional and state suspicion, the hypothesis was that dispositional suspicion exhibits a stronger association with deceptive perception and lie judgments when state suspicion is low than when it is high. Our statistical analysis did not reveal a significant interaction between the two types of suspicion, but the simple slope analysis of the data in Figure 1A supported the hypothesis for the deceptiveness ratings. It was found that participants with high GCS rated the speakers as more deceptive than those with low GCS in the non-suspicious condition, but this trend diminished under the suspicious condition. This interaction pattern is consistent with the findings of Levine and McCornack (1991). However, this effect was weak and found only for deceptiveness perception but not for lie judgments, again suggesting that deception perception and veracity judgments are two distinct processes. The TDT (Levine, 2014) supposes separate thresholds for deceptiveness perception and lie judgments. It means that information that is perceived as deceptive is not necessarily judged as a lie. In other words, extra evidence for deception is often required to exceed the threshold for lie judgments because people tend to return to their default cognition of believing others’ innocence over time. In the present study, participants likely crossed the first threshold to arouse suspicion, but the evidence was not strong enough for the second threshold to label the speakers as liars.
The weak interaction can also be explained by insufficient manipulation of state suspicion. The manipulation-check items of the study indicated that receiving the crime story raised participants’ state suspicion. However, a main effect of condition on deceptiveness ratings or lie judgments was not found. The speakers in the videos being total strangers to the participants, and the speech contents being unrelated to the crime might have limited participants’ state suspicion from rising sufficiently (for a discussion of the relationship between closeness and truth bias, see Van Swol et al., 2012). The video contents can be improved by being more emotionally engaging and crime-related (e.g., capturing a scene of interrogation, emotional speech, or a person being known to participants). In addition, emphasizing the importance of correct lie detection (e.g., by telling participants that their performance in the study is related to real crime solving) may make participants more prone to state suspicion manipulations. Future studies must incorporate such improved scenarios.
Insufficient state manipulation can also be attributed to the experimental procedure. Answering the GCS scale would raise participants’ awareness of their current suspicion, and this might have raised their skepticism towards others regardless of the manipulation of state suspicion. Future research should avoid raising participants’ suspicion needlessly. For example, separating the administration of the GCS scale from the veracity judgment task or employing alternative approaches, such as thought-listing tasks (Levine et al., 2021) or the Implicit Association Test (Fan et al., 2022), could ensure the control of state suspicion.
The secondary aim of this study was to explore the association between dispositional suspicion and information-seeking behavior. The results revealed a borderline significant relationship in which individuals with high dispositional suspicion gathered less information before reaching final truth-lie judgments. Hurst and Oswald (2012) found that experienced police officers with high GCS reached lie decisions using less information gathering. The present research replicated this with laypeople. The reliance on less information by police officers may result from a biased strategy, where they selectively attend to information associated with deception (Nahari, 2012). Laypeople with high GCS may also use this strategy.
Nevertheless, the present findings on the information-gathering performance were far from robust, because only 30% of the participants watched more than one video. Notably, previous research also suffered from participants being reluctant to watch many videos (Hurst & Oswald, 2012), signifying that this is a common problem in the research field. It could be due to participants’ lack of motivation for information gathering or the task being too easy (the first video showed clear enough cues for the participants to make the decision confidently). Using videos with ambiguous veracity or applying penalties for wrong veracity judgments may encourage participants to view more videos. Examining the details of the information-gathering strategy is also crucial. Further research may investigate the factors underlying judgmental strategies, such as concerns about errors in judgment, the number of cues collected and used for the judgment, and why they were used (see Masip et al., 2006).
Although the current research has several limitations, the findings have a great scope for practical application, especially in training for detecting deception and prevention measures for fraud. Conventional training for deception detection has focused on informing people about the deceptive cues (e.g., others’ nonverbal behaviors or language use) and directing their attention to them. However, trainees often fail to detect deception in a reasonable time frame (Novotny et al., 2018) and do not even consider the possibility of being deceived (Wen et al., 2022). These findings point to the importance of dispositional suspicion. Those training methods may only have a limited effect on people with low levels of dispositional suspicion. Therefore, the training may need to focus on increasing general suspicion among trainees when processing information. Our findings on the relationship between dispositional suspicion and information gathering also have potential applications in fraud prevention. For instance, Vishwanath et al. (2018) found that individuals with high dispositional suspicion were less susceptible to phishing scams than those with low suspicion due to their extensive information gathering. While our study failed to replicate this finding, this line of research contributes to fraud prevention.
Our findings will also enhance the understanding of personality, since dispositional suspicion can be considered a personality trait. In personality research, investigating the impact of situations on behavioral patterns has been popular (Nasello et al., 2024). In the context of suspicion, dispositional suspicion and state (situational) suspicion are distinguishable relatively easily, and it is possible to observe how they interact. For example, Masip et al. (2016) found that novice police officers exhibited significantly higher GCS scores when assuming a context related to a police interrogation than when responding to an everyday interpersonal context, suggesting that supposedly dispositional characteristics are changeable depending on the situation. Therefore, future research similar to the present study will provide further insight into the interactional effects of personality traits and situations on human behaviors.
Supplementary materials are available on the journal’s website.
