Introduction

Honest reporting is essential for maintaining trust and efficiency across various sectors, including volunteer work, where organizations depend on self-reported data to assess contributions, allocate resources, and measure impact. However, individuals may sometimes exaggerate or misrepresent their skills—either intentionally or subconsciously—due to social pressures, personal incentives, or a lack of oversight. This raises a critical question for organizations and policymakers: How can honesty in self-reporting be encouraged without resorting to costly monitoring and enforcement mechanisms? One promising solution is the use of honesty pledges, which prompt individuals to commit to truthfulness before reporting.

Honesty pledges have been tested in various contexts, demonstrating their potential to encourage ethical behavior. For instance, Jacquemet et al. (2013) found that signing an oath strengthened intrinsic motivation to tell the truth. Similarly, McDonald, Scott, and Hanmer (2017) reported that individuals who pledged honesty in voting surveys were less likely to misrepresent their past voting behavior. Tsipursky et al. (2018) introduced the Pro-Truth Pledge, which effectively reduced the spread of misinformation online by committing signees to truthfulness. Other studies have examined the effectiveness of pledges in more specific settings. Jacquemet et al. (2019) found that oaths were particularly effective in morally loaded environments, while Heinicke et al. (2019) demonstrated that truth pledges significantly reduced dishonesty, particularly among habitual liars. More recently, Dunaiev & Kadjavi (2021) showed that honesty pledges foster truthful reporting in both individual and team-based settings, and Peer & Feldman (2024) found that pledges consistently reduced dishonesty—even in the presence of fines—affecting both frequent cheaters and those less inclined to follow rules.

Despite these promising findings, honesty pledges are not universally effective, and their success often depends on the context and implementation. Dimant et al. (2020) and Martuza et al. (2022) found no significant effects of honesty nudges in certain scenarios, suggesting their impact is highly context-dependent. Kristal et al. (2020) tested a widely cited finding on honesty pledges but failed to replicate its effects across five conceptual replications, concluding that signing an honesty pledge at the outset does not necessarily increase truthful reporting. Similarly, Jacquemet et al. (2020) demonstrated that while oaths reduced dishonesty among partial liars, they had little impact on chronic liars in an income/tax-reporting experiment.

Additional studies highlight the limitations of honesty pledges in comparison to alternative interventions. Cagala et al. (2024) tested commitment requests in both online and offline settings but found no significant effect on truthfulness. Prima et al. (2020) compared honesty pledges to verification threats in a welfare benefits application process in Indonesia. Their framed field experiments with 599 participants revealed that verification threats were significantly more effective than honesty pledges in deterring dishonest reporting. Notably, they also found gender differences, with men being more prone to dishonesty and more responsive to interventions, suggesting that demographic factors influence the effectiveness of honesty nudges. These mixed findings highlight the complexity of designing and implementing successful honesty interventions, as their effectiveness depends on contextual factors, intervention design, and individual differences.

Building on this foundation, recent studies have examined conditions that enhance the effectiveness of honesty nudges. LeMaux and Necker (2023) found that the timing of moral reminders is critical, with interventions before or during decision-making proving most effective. Peer et al. (2024) demonstrated that high-involvement pledges, such as written commitments, have a greater impact on honesty compared to passive nudges, particularly when repeated over time. These findings suggest that engagement and reinforcement play crucial roles in the success of honesty interventions.

Despite these insights, the effectiveness of honesty pledges in voluntary settings remains underexplored. In particular, there is limited research on whether honesty pledges can improve the accuracy of self-reported volunteer application data, where applicants may have incentives to misrepresent their qualifications or intentions. This study seeks to fill that gap by investigating whether honesty pledges can serve as a viable tool for promoting truthful reporting in a high-stakes volunteer recruitment process.

This study examines the effectiveness of an honesty pledge integrated into an online volunteer application for the FIFA World Cup 2022™ in Qatar. To qualify, applicants were required to demonstrate proficiency in English, the official language of the FIFA World Cup Qatar 2022™ Volunteer Program, while Arabic proficiency was considered an advantage. The application process consisted of three stages: (1) Online application form (self-reported qualifications and skills), (2) Online gamified assessment, and (3) Interview and final selection.

Our research focuses on the first phase—the online application form—to evaluate whether an honesty pledge can influence applicants’ self-evaluations and reduce misreporting, potentially limiting the number of unqualified candidates advancing to later stages.

Methods

Study design and participants

The current large-scale randomized controlled trial was conducted online. It included a convenience sample of 12,388 participants who were 18 years or older (mean = 33.72, SD = 9.84). Ethical approval was obtained for the study. To ensure data quality and protection, all data was anonymized and stored securely on a password-protected, encrypted computer, accessible only to the research team.

Procedure

Participants received an email invitation to register for an account on an online portal. Upon registration, they were required to complete a basic application form to proceed to the next stage of the volunteer program. The sample was randomized by stratified random sampling based on gender. Participants in the treatment group were exposed to the honesty nudge intervention. Participants in the control group did not receive the honesty pledge. They proceeded directly to the application questions without any prior statement about honesty. Answering some of the application questions was optional, allowing participants to choose whether or not to respond. Participants in the treatment group were presented with an honesty pledge at the beginning of the application form (Fig. 1). The statement read: “By clicking this box, I agree to tell the truth during this application.” Participants were required to confirm they had read the statement by clicking a checkbox before proceeding. This honesty pledge remained visible, hovering at the top of the application form as they scrolled through the questions, serving as a continuous reminder. Participants in both groups answered the same set of questions, which included demographic information, employment status, education level, preferences for volunteering, relevant volunteering experience, language skills, physical activity levels, the importance of health, and medical details.

Fig. 1
figure 1

The pledge is displayed and remains at the top of the form as participants complete the questionnaire.

Full size image

The primary outcome in our study was participants’ responses to the items about proficiency in the Arabic and English languages: Please indicate your proficiency in the English/Arabic language (response options: 0 (None), 1 (Beginner), 2 (Intermediate), 3 (Advanced), 4 (Native)). The other outcome variables of interest are: an item about their physical activity: Do you make sure that you walk or do other moderate physical activity for at least 30 min every day? (response options: Yes/No), an item about their comfort with disability: Are you comfortable working with people with disabilities? (response options: Yes/No) and, an item about health importance: How much does your health matter to you? (response options: 0 (Not at all) to 4 (Very much).

Data analysis

The treatment condition was coded as 1 for the honesty pledge group and 0 for the control group. Descriptive statistics were conducted for the sample, grouped by treatment versus control. Responses to five questionnaire items were analyzed: language proficiency (Arabic and English), physical activity, comfort with disability and health importance. Each language proficiency response and the health importance response were analyzed using ordinal logistic regression.

For further analysis, English and Arabic language proficiency were categorized into two groups: Non-proficient, which included responses of None and Beginner, and Proficient, which encompassed Intermediate, Advanced, and Native. This categorization was designed to reflect a practical distinction between participants who lack functional English or Arabic language skills (Non-proficient) and those who demonstrate at least a moderate level of communication ability (Proficient). Such proficiency is considered necessary for effective interaction in Arabic-speaking environments and in the English-speaking FIFA World Cup setting, as specified by the volunteer program. Additionally, this categorization considered the applicant’s perception of the level of proficiency necessary to increase their chances of acceptance. It was assumed that marking Advanced or Native on the Arabic proficiency question would be perceived as advantageous, particularly since Arabic is listed as a preferred skill for volunteers. Similarly, marking Advanced or Native on the English proficiency question was likely considered sufficient to meet the mandatory English requirement.

For Arabic language proficiency, the sample was further divided based on respondents’ nationality—Arab-speaking versus non-Arab-speaking—since the primary focus was on the self-evaluation of Arabic language skills by non-Arab-speaking participants, who are more likely to be motivated to misreport. Finally, physical activity and comfort with disability were analyzed using binary logistic regression.

Results

Participant characteristics

The sample consisted of 12,388 participants who were 18 years or older (mean = 33.72, SD = 9.84), with 20.5% identifying as females. 57.6% were having a Bachelor’s degree, following by 20.8% with a Master’s degree or secondary education (18.2%). Geographically, the majority of respondents were from Africa (34%), followed by Asia-Pacific (31.3%) and the Middle East (24.5%). Approximately half of the participants identified as Arab-speaking (52.7%), while the remaining 47.3% were non-Arab-speaking. The percentage of missing values for demographic characteristics ranged between 0.2% and 0.5%. Specifically, there were 64 missing cases for education (0.5%) and 27 missing cases for nationality and residence (0.2%). In total, 6255 participants were randomized to the control group and 6133 were randomized to the treatment group. Descriptive statistics of the sample are presented in Table 1.

Table 1 Demographics: descriptive statistics (N = 12,388).
Full size table

Arabic language proficiency

Among participants from non-Arabic-speaking countries, 80.8% reported non-proficient Arabic language proficiency (Table 2). The results of the ordinal regression analysis for the non-Arab-speaking sample indicated that participants who did not receive an honesty pledge had 11.3% higher odds of reporting higher levels of Arabic proficiency compared to those who received the pledge (OR = 1.113, 95% CI [1.011, 1.225], P = 0.029). Receiving the pledge (the reference category) was associated with a lower likelihood of reporting greater proficiency. Table 3 shows the results from the regression analysis.

Table 2 Descriptive statistics for English and Arabic proficiency.
Full size table
Table 3 Results of ordinal regression analysis for language proficiency.
Full size table

English language proficiency

91.8% of all participants reported proficient English language skills (scoring either advanced or native) (Table 2). The results of the ordinal regression analysis indicated that participants who did not receive an honesty pledge had 9.2% lower odds of reporting higher levels of English proficiency compared to those who received the pledge (OR = 0.908, 95% CI [0.850, 0.971], P  = 0.004). Interestingly, receiving the pledge (the reference category) was associated with a higher likelihood of reporting greater proficiency. Table 3 shows the results from the regression analysis.

The results from the regression analysis on the health importance variable, physical activity and comfort with disability showed that receiving the pledge does not significantly predict the outcome (P > 0.05).

Discussion

The analysis revealed that participants who did not receive an honesty pledge were less likely to report higher levels of English proficiency compared to those who received the pledge. Interestingly, receiving the pledge was associated with a greater likelihood of reporting higher proficiency.

Given the context of applying for a volunteer program at the FIFA World Cup, where English proficiency is mandatory, the honesty pledge may not have worked as expected due to several psychological factors. On the one hand, participants may have felt pressure to appear qualified. English proficiency being a mandatory criterion likely amplified the need for self-promotion, as participants may have believed that reporting lower proficiency could disqualify them from the opportunity. Research on impression management highlights that individuals are more likely to engage in deceptive self-presentation when there are clear incentives, such as gaining a competitive position (Weiss & Feldman, 2006; Bourdage et al., 2017).

Additionally, unlike low-risk environments such as online surveys (McDonald, Scott, & Hanmer, 2017) or social media pledges (Tsipursky et al., 2018), our study involved a competitive selection process where applicants had incentives to present themselves as ideal candidates. This context mirrors findings by Prima et al. (2020), who showed that in a high-stakes welfare application setting, honesty pledges were less effective than verification threats in reducing dishonesty. Similarly, Cagala et al. (2024) found no significant impact of commitment requests on truthfulness in an experimental setting, emphasizing the variability of pledge effectiveness across different contexts.

On the other hand, instead of deterring exaggeration, the honesty pledge might have reinforced participants’ need to rationalize their responses as truthful. Participants could have internally justified overestimating their English skills by convincing themselves that they possess the necessary potential, will present themselves better in the next selection phases or improve during the program. This resolution of cognitive dissonance is consistent with findings that individuals often adapt their honesty to align with their self-image in high-stakes settings (Leary, 1995; Schlenker & Weigold, 1992).

Finally, self-reported language proficiency is inherently subjective and prone to biases. Participants might have interpreted their proficiency more optimistically, especially under the belief that they were being honest. The honesty pledge may have amplified participants’ subjective self-assessment rather than encouraging them to critically evaluate their true skills (Schlenker & Weigold, 1992; Leary, 1995).

Conversely, the results showed that non-Arabic-speaking participants who did not receive an honesty pledge were more likely to report higher levels of Arabic proficiency compared to those who received the pledge. Receiving the pledge was linked to a reduced likelihood of reporting greater proficiency.

The difference in results between Arabic and English proficiency can be explained by their respective roles in the context of the FIFA World Cup volunteer application process. English proficiency is mandatory, while Arabic is an advantage but not required. These distinctions influence participants’ motivations, self-perception, and behaviors when reporting their language skills. The mandatory requirement for English proficiency creates a high-stakes scenario, where participants feel significant pressure to report a level of proficiency that ensures their eligibility. In contrast, reporting Arabic proficiency is a low-stakes decision, as it is not mandatory. Participants may not feel the same pressure to exaggerate their skills. For those who did not receive the honesty pledge, reporting higher proficiency in Arabic may still reflect a desire to stand out as a competitive candidate, leading to inflated reports. Those who received the pledge, however, may have felt less compelled to overstate their skills, instead providing more honest assessments.

Despite the valuable insights gained from this study, several limitations must be acknowledged. First, due to logistical constraints, we were unable to track participant engagement with the honesty pledge, which hovered throughout the questionnaire. Without data on whether participants actively noticed, read, or internalized the pledge, it is difficult to assess the precise mechanism through which it influenced self-reported behavior. Second, we had access only to the first phase of the application process, limiting our ability to validate language proficiency at this stage. While proficiency was assessed in later phases (such as interviews), we could not implement objective language tests during the initial self-report stage, potentially affecting the accuracy of our findings.

Third, the study relied on self-reported measures for complex constructs such as comfort with disability, health importance, and language proficiency, without additional validation methods such as attention checks or social desirability bias controls. This raises concerns about response reliability, as participants may have provided socially desirable answers rather than fully accurate self-assessments. In particular, the high self-reported health importance scores suggest that at least some responses may have been influenced by social desirability bias.

Finally, the study was conducted in a high-stakes setting, where applicants had incentives to misrepresent their qualifications to increase their chances of selection. While this reflects a real-world scenario, it also means that our findings may not generalize to lower-stakes honesty pledge interventions. Future research should explore honesty pledges in different contexts, incorporate engagement tracking, and include more robust validation measures to enhance data reliability.

Conclusion

Our findings indicate that honesty nudges may be less effective in high-stakes scenarios and could inadvertently increase the pressure to engage in self-promotion. In contexts where mandatory criteria amplify the stakes, such as with English proficiency, the honesty pledge may heighten the desire to present oneself as qualified. However, in lower-pressure situations, where impression management tactics are less critical—such as with non-mandatory Arabic proficiency—the honesty pledge appears to function as intended, encouraging more accurate self-assessments. This highlights the importance of considering the stakes and context when implementing honesty nudges.