The Truth-Default is Ubiquitous, but Some People Experience It More than Others

Abstract

Truth-default theory posits that most people are normatively honest and are believed by others. Do people think that others consider them honest? This paper explores how social perceptions of deception align with the truth-default. Using data from the All of Us dataset (N = 116,914 total respondents), we observed that most people feel trusted. While the predicted long-tail distribution was universal across subsamples, people who self-identified as male, are minorities in the US, had less education, and had less income were less likely to feel trusted by others. Connections to and implications for truth-default theory are discussed.

The functionality of human communication rests on the truth-default (Levine, 2020), a core premise underlying truth-default theory (TDT; Levine, 2014, 2020). According to the theory, most communication is honest and passively accepted as honest. That is, people are (mostly) truthful and people (mostly) believe others. But, to what extent do people think others believe them? Such social perceptions of trust and deception have rarely been examined in the literature (c.f., Markowitz, 2024; Zvi & Elaad, 2018), yet they have strong theoretical roots in truth-default theory. Against this backdrop, we extend deception theory and research by presenting one of the first large-scale evaluations of how the truth-default, as indicated by how much people think others trust them, operates within the general population.

Asserting the ubiquity of honesty and belief may seem challenging in a post-truth world (McIntyre, 2018). Nevertheless, truth-default theory holds that honesty and belief are—and must be—the sine qua non of efficient and effective communication. Without honesty and belief, humans cannot share knowledge, coordinate with groups, maintain social and personal relationships, and perform other functional communication tasks (Levine et al., 2024).

Prior data are remarkably consistent with truth-default theory’s predictions regarding the predominance of honesty and belief. Recent work on the prevalence of lying documents two key findings: (1) honesty is normative, and (2) there are a few prolific “outliars” in a population who skew the distribution of deception to the right (Markowitz, 2023; Serota et al., 2010, 2022, 2024). Specifically, most people are normatively honest in that they tell zero-to-two lies per day, with the vast majority of those lies being white lies of little importance. In contrast, there is a small percentage of the population facilitating a skewed distribution. Results also show that thoughts of lies and deception do not come to mind unless actively triggered (Clare & Levine, 2019; Levine et al., 2021; Markowitz & Hancock, 2024). Even in deception detection experiments, people are nevertheless truth-biased (Bond & DePaulo, 2006).

Do people realize that others are truth-biased and default to the truth? Do people think others trust them? To address these understudied questions, we used a large and researcher-available dataset to provide an initial answer. As described below, participants were asked a question that we formally describe as the social perceptions of deception: “In your day-to-day life, how often do people act as if they think you are dishonest?” Answers scaled from 0 = never to 5 = almost every day. The findings of seminal diary studies on deception prevalence suggest the answer is seldom (DePaulo et al., 1996). Truth-default theory further suggests that the answers will likely be skewed, fitting the same long-tailed pattern as the prevalence of deception.

A long-tailed distribution is expected to the extent that perceptions have some basis in reality. Since most people are normatively honest (Serota et al., 2010, 2022, 2024) and most people are also truth-biased (Bond & DePaulo, 2006), most people should be trusted by others. Some proportion of lies, however, are detected over time (Park et al., 2002). People who lie frequently are, over time, less likely to be trusted. The net result is a skewed distribution in which most people are trusted but a few are not.

An implication following from TDT is that the predicted skewed distribution will be robust across most if not all sub-samples in any human population, and observing a ubiquitous skew would be consistent with the theory.ⁱ The extent to which this is the case, however, is an empirical question. For example, in addition to prolific liars, others may feel distrusted because they experience discrimination by belonging to an outgroup or a stigmatized group (e.g., Engstrom et al., 2024). This suggests that the length and thickness of the long right-hand tail might vary by demographic category. Thus, we propose the following two research questions:

RQ₁: How are social perceptions of deception distributed in the population?

RQ₂: What is the relationship between the social perceptions of deception and demographic characteristics?

Method

All of Us Data

We examined the relationship between demographics and social perceptions of deception using the All of Us dataset (All of Us, 2019). This archive contains data from over 100,000 US residents who responded to various surveys with the goal of using such data to better evaluate, understand, and improve health. Given that in certain settings, lie production is linked to aversive personality traits and well-being concerns (Daiku et al., 2021; Jones & Paulhus, 2017; Markowitz, 2023), we analyzed these data in accordance with All of Us principles.

Basic demographic data were collected on each participant, along with responses to Social Determinants of Health items, which were particularly relevant to our research aims. One survey question, drawn from prior work on everyday discrimination (Williams et al., 1997), related to deception and asked, “In your day-to-day life, how often do people act as if they think you are dishonest?” This question contained seven response options and represented our primary dependent variable of interest: (1) Never, (2) Less than once a year, (3) A few times a year, (4) A few times a month, (5) At least once a week, (6) Almost every day, and (7) Skip. According to All of Us, less than 2% of participants responded with “Skip” to this question.

Participants

Our sample of 116,914 respondents represented various demographic groups. Table 1 presents the full data disaggregated from All of Us across five demographic categories (i.e., gender, ethnicity, race, income, education level). Most participants self-identified as female, not Hispanic nor Latino, White, and were a college graduate/had an advanced degree. The modal response for annual income was between $100,000–150,000 in US dollars. However, the sample was sufficiently large to capture diverse subsamples of participants.

Table 1.

Demographic Breakdown of the All of Us Data

Demographic variable	Category	Count	%
Gender	Female	73,047	62.48
	Male	39,478	33.77
	Not man only, not woman only, prefer not to answer, or skipped	4,389	3.75
Ethnicity	Not Hispanic or Latino	101,618	86.92
	Hispanic or Latino	10,267	8.78
	Skip	3,674	3.14
	None Of These	1,032	0.88
	Prefer Not To Answer	323	0.28
Race	White	89,012	76.13
	Black or African American	9,126	7.81
	None Indicated	7,934	6.79
	Skip	3,674	3.14
	Asian	3,171	2.71
	More than one population	2,083	1.78
	None of these	1,032	0.88
	Another single population	559	0.48
	I prefer not to answer	323	0.28
Income	Annual Income: 100k-150k	18,557	15.87
	Annual Income: 50k-75k	16,513	14.12
	Annual Income: 75k-100k	14,230	12.17
	Annual Income: more 200k	12,154	10.40
	Annual Income: 35k-50k	10,191	8.72
	Annual Income: 10k-25k	9,361	8.01
	Prefer Not To Answer	8,975	7.68
	Annual Income: 150k-200k	8,952	7.66
	Annual Income: 25k-35k	7,466	6.39
	Annual Income: less-10k	5,412	4.63
	Skip	5,103	4.36
Education level	College graduate or advanced degree	73,061	62.49
	Highest Grade: College One to Three	27,261	23.32
	Highest Grade: Twelve Or GED	10,609	9.07
	Skip	3,142	2.69
	Less than a high school degree or equivalent	2,567	2.20
	Prefer Not To Answer	274	0.23

Note. As indicated in the main text, statistical models in the main text exclude those in categories related to “skip,” “none of these,” “prefer not to answer,” or “none indicated.” Levels within each variable are arranged in descending order by count.

Analytic Plan

To present the cleanest pattern of results, we excluded cases where participants responded to demographics with “skip,” “none of these,” “prefer not to answer,” or “none indicated” answers. This still retained a substantial sample (over 100,000 cases for each demographic category; Table 1) to examine links between social perceptions of deception and demographics. For the dependent variable, we transformed labels into their numeric equivalents when computing linear models: “Never” = 0, “Less than once a year” = 1, “A few times a year” = 2, “A few times a month” = 3, “At least once a week” = 4, “Almost every day” = 5.

Results

Density plots representing the social perceptions of deception across key demographic variables are in Figure 1. Addressing RQ₁, similar long-tailed distributions were observed in every sub-group. The modal response was invariably, and predictably, “Never.” Our second research question regarding differences is addressed below.

Figure 1.

Social Perceptions of Deception Density Plots Across Key Demographic Variables

Note. Numbers on the x-axis correspond to our primary dependent variable, the social perceptions of deception measure: “Never” = 0, “Less than once a year” = 1, “A few times a year” = 2, “A few times a month” = 3, “At least once a week” = 4, “Almost every day” = 5.

Gender

Gender was statistically linked to social perceptions of deception, χ²(5) = 120.47, p < .001, φ_c = 0.033. Specifically, male participants reported more social perceptions of deception than female participants (B = 0.045, SE = 0.006, t = 8.16, p < .001).¹

Ethnicity

Ethnicity was statistically linked to social perceptions of deception, χ²(5) = 359.79, p < .001, φ_c = 0.089. Hispanic or Latino participants reported more social perceptions of deception than non-Hispanic or Latino participants (B = 0.15, SE = 0.01, t = 16.04, p < .001).

Race

Race was statistically linked to social perceptions of deception, χ²(20) = 3981.2, p < .001, φ_c = 0.099. We chose White participants as the reference group solely because this category had the greatest representation in the dataset (Table 1). Specifically, Asian (B = 0.17, SE = 0.02, t = 10.81, p < .001), Black or African American (B = 0.55, SE = 0.01, t = 57.12, p < .001), participants representing more than one population (B = 0.42, SE = 0.02, t = 21.87, p < .001), and those of another single population (B = 0.17, SE = 0.04, t = 4.60, p < .001) reported more frequent social perceptions of deception than White participants.

Income

Income was statistically linked to social perceptions of deception, χ²(40) = 4921.6, p < .001, φ_c = 0.099. We observed that compared to participants making between $100,000–150,000 per year (the largest category in the dataset; Table 1), those with lower incomes reported more frequent social perceptions of deception while those making $150,000–200,000 (B = −0.023, SE = 0.01, t = −2.07, p = .039) and those making more than $200,000 (B = −0.04, SE = 0.01, t = −4.18, p < .001) reported fewer social perceptions of deception. As Figure 1 illustrates, there is a distinct increase in social perception of deception as income declines.

Education Level

Education level was statistically linked to social perceptions of deception, χ²(15) = 2283.2, p < .001, φ_c = 0.083. Specifically, those whose highest level of education was one-to-three years of college (B = 0.15, SE = 0.01, t = 24.32, p < .001), completed grade twelve or those who had a G.E.D. (B = 0.22, SE = 0.01, t = 23.76, p < .001), or less than a high school degree (B = 0.33, SE = 0.02, t = 18.28, p < .001) reported more frequent social perceptions of deception than those who were college graduates (or those who had an advanced degree).

Note, the prior results were largely maintained in a global regression including all variables, simultaneously, as predictors (see Supplementary Table S1).

Discussion

Dozens of studies in deception research have noted a consistent distribution in lying rates: most people are honest, and there are a few prolific liars who skew the distribution to the right. To date, few studies have considered peoples’ subjective impressions of how others see them, called social perceptions of deception, and the degree to which people believe others might consider them to be dishonest. We observed a pattern of results that held across all demographic categories and was consistent with truth-default theory: most people think that others consider them to be honest. Only a few people think that others perceive them to be dishonest. Even then, almost no one answered “almost every day.” Perceptions of being seen as deceptive are an unusual event. However, the length and thickness of the long tail showed differences. Minorities, those with less education, and lower-income people were more likely to feel distrusted.

There are several contributions worth highlighting. First, it was a theoretical and empirical uncertainty if the truth-default would be present in terms of social perception, and across various groups that have often been understudied or difficult-to-reach in academic research (including deception scholarship). Our work leveraged researcher-available data, which were largely recruited for health-related research, to test our theoretical interests in the truth-default across such populations. That the truth-default held, in the form of perceptions of truthfulness, and was remarkably consistent across various demographic traits, underscores the ubiquity of the effect and the theory that underlies it. Other work has evaluated truth-default theory principles for their pancultural ubiquity (Levine et al., 2016; Serota et al., 2024), but large-scale findings within one cultural context and disaggregating across demographics are often difficult empirical tasks to undertake.

Second, our work potentially highlights deeply-rooted inequities in social perception that also include distrust. The item we examined from the All of Us dataset highlights perceptions of deceptiveness as a form of discrimination, which to the best of our knowledge has not been investigated with the depth or breadth that it requires. The findings from our work indicate how people from often socially ostracized, stereotyped, and oppressed groups (e.g., people of color, those with lower socio-economic standing) tend to believe that others perceive them as more dishonest actors in society relative to social counterparts. The consequences of such perceptions are likely socially and psychologically damaging for individuals within these groups. People who think that others believe they are dishonest may be less likely to engage in society, believe that others are working against them, and trust institutions less (for example, see Funk, 2022; Vij, 2020), contentions that may be addressable with additional All of Us data as well. We argue the evidence in this paper may be a starting point to recognize the pervasiveness of the social perceptions of deception, and hopefully use future work to mitigate its impact on historically undermined groups in society.

An interesting aspect of our findings is that two seemingly contradictory findings were observed. The theoretically anticipated ubiquitous skew was observed, but not all subsamples exhibited equally skinny tails. The former finding provides the primary theoretical contribution of these data while the latter shows the value of exploring data beyond a theory. In this case, truth-default theory is largely silent on demographic differences other than anticipating that any differences would be reflected in the thickness of the tails rather than the direction of the skew. Among the implications is that individual differences may moderate the propensity for honesty and the perceptions of others’ honesty, as well as contribute to situations where truthfulness does not always work.

Limitations and Future Directions

Our work relied on a single-item measure of social perceptions of dishonesty that was taken from a valid and reliable everyday discrimination scale. Still, future work might benefit from having or developing a multi-item measure of the social perceptions of dishonesty. We also evaluated just five demographic categories in this work, but there are many others (e.g., sexual orientation, family size) that might warrant future consideration.

The effect sizes in this study are admittedly small, yet statistically reliable. This is a consistent refrain in deception scholarship (Hauch et al., 2015; Luke, 2019), one that does not uniquely relate to the present study. Small effect sizes do not indicate trivial results, however, especially because they can have an important impact at the population level or at the magnitude that we report in the current work. We had access to one of the largest samples studying a deception phenomenon, which suggests even small relationships might have meaningful implications for how we think about perceptions of deception and social groups. Relatedly, we acknowledge that these data are correlational and not direct cause and effect. Future experimental research might help to augment these findings to assess the direction of the relationships in this study and reasons why these perceptions occur.

Finally, we only had access to data from US participants. Truth-default theory, which motivated the current project, is a pancultural theory that has demonstrated its robustness across many cultures and locales. We therefore believe such findings are likely to hold across cultures, but this deserves greater empirical attention. We encourage future research to extend our work by evaluating how social perceptions of deception might be moderated by culture and other context-related factors.