“If Others Are Honest, I Will Be Too”: Effects of Social Norms on Willingness to Fake During Employment Interviews

Applicant faking in employment interviews is a pressing concern for organizations. It has previously been suggested that subjective norms may be an important antecedent of faking, but experimental studies are lacking. We report a preregistered experiment (N = 307) where effects of conveying descriptive social norms (information about what most applicants do) on self-reported willingness to fake were examined. Although we observed no difference between the faking norm condition and the control condition, in which no norm was signaled, participants in the honesty norm condition reported lower willingness to fake compared to those in both the faking norm condition and the control condition. The latter supports the idea that conveying honesty norms may be an effective means of reducing faking, although future research needs to evaluate its usefulness in real employment interviews.

Personnel Assessment And decisions sociAl norms And FAking interviews on personality (e.g., Buehl & Melchers, 2017). Recently, there has been a call for more research on situational and contextual factors that determine whether applicants will choose to fake (Melchers et al., 2020). One important factor might be the perceived norm. Social norms lie at the core of human functioning (House et al., 2020). Whereas some social norms can support social order and cooperation, others can encourage harmful behaviors. Indeed, social norms have been found to affect a wide variety of behaviors, such as drinking alcohol (Walters & Neighbors, 2005), smoking cigarettes or sharing needles (Reid et al., 2010), exercising (Okun et al., 2002), environmental conservation (Goldstein et al., 2008), and charity donations (Agerström et al., 2016). People are often motivated to pay attention to norms, as failing to do so may elicit social disapproval (Cialdini et al., 1990). Whereas injunctive social norms inform us about what we ought to be doing; for example, that we should tell the truth because lying is morally wrong, descriptive social norms inform us about how most people behave. Especially in situations when people are uncertain about how they should act, they tend to observe what others do to determine an appropriate course of action. When applicants learn that most people fake, or conversely, that most people are truthful during employment interviews, they may come to perceive this to be a suitable and effective behavior in this particular situation (e.g., "If everybody else chooses to exaggerate their qualifications during job interviews, it must be the most effective way of coming across as a good applicant") and act accordingly (Chung & Rimal, 2016;Cialdini et al., 1990;Legros & Cislaghi, 2020).
Few studies have examined the role of subjective norms in the context of applicant interview faking (Bourdage et al., 2020;Dürr & Klehe, 2018;Lester et al., 2015). The primary purpose of these previous studies has been to test the usefulness of the theory of planned behavior (Ajzen, 1991), and they have therefore examined subjective norms as a predictor of faking along with attitudes and perceived behavioral control. In contrast to the present study, their focus has mainly been on injunctive norms, more specifically the extent to which one thinks that close others (e.g., family, friends) find faking behavior acceptable (although the norm measure used in the study by Bourdage and colleagues [2020] also included one item asking about the descriptive norm). In general, these past studies suggest the existence of a relationship between social norms and self-reported interview faking. However, they have used correlational designs. This means that although the observed relationship between subjective norms and faking could indeed be due to subjective norms affecting faking behavior, it could also be that those who fake are motivated to think that other people fake, because this belief serves to justify their own behavior (Daumiller & Janke, 2019;O'Rourke et al., 2010). Thus, studying the relationship between perceptions of others' faking and own self-reported faking behaviors in correlational studies can yield inflated associations. The present study is to our knowledge the first to test the causal effect of social norms in the context of employment interview faking.

Aims and Hypotheses
Descriptive social norms offer information that people can use to guide their actions (Legros & Cislaghi, 2020). We experimentally tested whether descriptive norms affect people's willingness to fake during job interviews. We expected willingness to fake to shift in the direction of the norm (i.e., what most people do), hypothesizing that: Hypothesis 1: People will become more prone to faking when they learn that most people fake, relative to an honesty norm condition and a control condition.
Hypothesis 2: People will become less prone to faking when they learn that most people are honest during job interviews, relative to the faking norm condition and the control condition.
Hypothesis 3: Perceived prevalence of faking will be positively correlated to willingness to fake.

METHOD
For reasons of transparency, we report how we determined the sample size, all data exclusions, all manipulations, and all measures in the study. The preregistration can be found on the Open Science Framework project page: [https://osf.io/m6hrd/?view_only=d9eab08f9eec43b2b0f8b-42c999c30e2].

Participants and Design
The data for this study were collected on Prolific. Prescreening was done for UK residents who are native English speakers, 18-65 years of age. The participants were randomly assigned to a faking norm condition where they received information saying that most people fake their responses during job interviews, an honesty norm condition saying that most people give honest responses, or a control (no norm) condition where nothing was mentioned about what other people do during job interviews. Participants were paid an hourly rate of 7.08 British pounds, and they all received payment for 5 minutes (i.e. 0.59 pounds) even though most finished in less time.
Our plan was to recruit 330 participants, which would mean about 85 % power to detect small-to-medium effects of Cohen's d = 0.4 for the follow-up planned comparisons between the experimental conditions. We collected 331 participants, and after excluding three individuals who failed to answer an attention check correctly and an additional 21 who failed a manipulation check, the final sample consisted ReseaRch aRticles of 307 participants (107 in the faking norm, 94 in the honesty norm, and 106 in the control condition).

Materials and Procedure
The materials were prepared with Qualtrics online survey software. Participants were introduced to a study on job interview behavior. After the participants had given their informed consent, they were exposed to the experimental manipulation, which is similar to several previous norm experiments (e.g., Agerström et al., 2016;Goldstein et al. 2008). In the faking norm condition, participants learned that: It is estimated that on average 75% of job applicants choose to polish their responses during job interviews, in order to appear more suitable for the job. This could include, for example, exaggerating one's qualifications or abilities, or omitting/covering up previous failures or negative events.
In the honesty norm condition, the participants learned that: It is estimated that on average 75% of job applicants choose to be honest in their responses during job interviews. That is, they avoid to exaggerate their qualifications or abilities, or to omit/cover up previous failures or negative events.
In the control (no norm) condition, the participants read the following: This study is part of a research project that examines behaviors during job interviews; in particular, whether people choose to polish their responses during hiring interviews, in order to appear more suitable for the job, or if they prefer to be honest. Polishing one's responses could include for example exaggerating one's qualifications or abilities, or omitting/covering up previous failures or negative events.
All participants then read that "This study is part of a research project that looks further into people's attitudes to job interviews." To ensure that all participants payed attention to the manipulation, they had to wait 15 seconds before they could proceed to the next page, and they were exposed to a reminder of the importance of reading the text thoroughly. Next, they were asked to imagine that they were applying for a job in which they were very interested and to indicate how they would act during the job interview. The shortened version (11 items; previously used by Ingold et al., 2015) of the Interview Faking Scale (Levashina & Campion, 2007) constituted the dependent variable. Items include: "I would make something up in order to be able to give better interview responses," "I would present other people's experiences or achievements as my own," "I would omit something to improve the interview responses," "I would cover something up in order to be able to give better interview responses" (1 = does not apply at all, 5 = fully applies). Reliability was acceptable (Cronbach's α = .81, McDonald's ω = .82). We also included an attention check: "It's important that you pay attention to this study. Please tick 'fully applies.'" All participants answered demographic questions about age, gender, work experience, interview experience (yes/ no), and occupation. In the two norm conditions, there was also a manipulation check where we asked which information was presented in the beginning of this survey (with the options "A majority of applicants choose to polish their responses during hiring interviews"; "A majority of job applicants choose to be honest in their responses during hiring interviews"; and "None of the above").
We were also interested in the perceived faking prevalence. In the control condition, the participants were thus asked: How common do you think it is that people polish their responses during job interviews, in order to appear more suitable for the job? Please answer in percentages, where 0% means that no one polishes their responses, and 100% means that everyone does it. (Polishing one's responses could include for example exaggerating one's qualifications or abilities, or omitting/covering up previous failures or negative events.).
Finally, all participants were debriefed about the full purpose of the study.

RESULTS
The data were analyzed with IBM SPSS version 26.0 and jamovi 0.9.5.12 (jamovi project, 2019). To test our main hypothesis, we conducted an ANOVA with experimental condition (3: faking norm, honesty norm, and control; between groups) as the independent variable and the interview faking scale as the dependent variable. This revealed a statistically significant effect of experimental condition that was small to moderate, F(2, 304) = 8.14, p < Personnel Assessment And decisions sociAl norms And FAking interviews .001, η p ² = .051. We proceeded by performing pairwise comparisons between the different experimental conditions. Reported p-values were adjusted using the Benjamini-Hochberg false discovery rate methodology (Benjamini & Hochberg, 1995). Finally, we found that among the 106 participants (i.e., the control condition) who reported perceived prevalence of faking, nobody answered that 0-20% of applicants fake, whereas 10.4% answered 21-40%, 22.6% answered 41-60%, 38.7% answered 61-80%, and 28.3% answered that 81-100% fake during interviews. Mean on this five-point scale was 3.85 (SD = 0.96). As hypothesized (Hypothesis 3), perceived prevalence of faking was positively correlated with faking intentions, r = .30, p = .002 [.12, .47], suggesting that perceived norms matter for willingness to fake: The more one believes that others fake, the more willing one is to fake.

DISCUSSION
In the personnel selection process, a candidate's goal is typically getting hired, and using deceptive impression management tactics such as faking can be a means to meet that end. Deceptive interviewees can bias the organization's decision toward less competent or suitable applicants (Weiss & Feldman, 2006), and if successful, such initial deception can potentially encourage future deceptive behavior at the workplace (Fleming & Zyglidopoulos, 2008). It is thus important to develop a better understanding of the antecedents of faking.
People look at how others act as a way to help them make decisions, especially in situations when they are uncertain about how to behave. The employment interview might very well be such a situation; is it best to be truthful at the cost of revealing one's flaws and risk losing the job to someone else, or to bend the truth and hope to get away with it? The present study provides new knowledge by being the first to examine the causal relationship between descriptive social norms and faking in employment interviews. Our findings suggest that descriptive social norms can serve as a cue for the appropriateness of faking versus FIGURE 1.
Effects of social norms on willingness to fake honest behaviors during interviews. Further strengthening this conclusion is the finding that perceived prevalence of faking predicted willingness to fake, which is consistent with the theory of planned behavior (Ajzen, 1991), and adds to the literature that has investigated subjective norms as a predictor of faking intentions (e.g., Bourdage et al., 2020) and intentions of other unethical behaviors in the workplace, such as theft (e.g., Bailey, 2006) and time theft (Henle et al., 2010).
Situational and contextual factors have been understudied in the faking literature (Melchers et al., 2020). Considering social norms as a motivator behind faking serves as a reminder that it may not only be applicants with problematic personality traits who choose to engage in such behavior. Situational explanations also have stronger implications for behavioral modification interventions, as personality is difficult to alter (Daumiller & Janke, 2019;Whitley, 1998). Studying norm compliance may suggest new ways for organizations to adjust their strategies to reduce bias and improve accuracy in the selection process.
However, our hypothesis was only partially supported as the faking norm condition did not differ from the control condition in faking intentions. Some social norms affect people's behavior through offering value-neutral information that the individual uses to determine an appropriate course of action, whereas other norms put pressure on individuals to act in a certain way (Legros & Cislaghi, 2020;Villatoro et al., 2010). When individuals feel this pressure, they tend to consider the reputational or emotional consequences (e.g., shame) that noncompliance may bring (Bell & Cox, 2015;Morris et al., 2015). The honesty norm manipulation described a prosocial behavior, whereas the faking norm manipulation described a more socially undesirable behavior. It is therefore possible that conveying the honesty norm put moral pressure on people to be honest, whereas the absence of pressure from the faking norm explains why we did not observe a difference between the faking norm and control condition; when a norm does not create external obligations, its effect may simply be weaker (Legros & Cislaghi, 2020). Another possible explanation is that people are not as surprised to hear that most people fake, compared to learning that most applicants are honest. If this is the case, the faking condition might serve to confirm the norm, whereas the honesty condition challenges the preconceived notion of how common faking is. After all, responses to the perceived faking prevalence question suggested that about 90% of respondents believed that more than 40% of all applicants fake. Nonetheless, the strength of the perceived norm of faking versus being honest in job interviews remains fairly speculative at this point and should be studied further. A third possibility concerns the subtleness of the faking norm manipulation. The term faking was not explicitly mentioned, and the examples of faking behavior that were provided (exaggerating one's qualifications or abilities, omitting/covering up previous failures or negative events) involved relatively mild forms of faking. Because the dependent variable included examples of more severe faking (extensive image creation), there was some misalignment between the independent variable and the dependent variable regarding severity. Future research aiming to replicate or extend this study should consider manipulations that involve more severe forms of faking, as long as this can be accomplished without compromising the credibility of the manipulation. After all, it is essential that people find the descriptive norm information referring to what most people do plausible in the first place.
We also explored whether demographic factors were of importance for faking. Applicants with more work experience are generally expected to be in less of a need to fake, as they have acquired more skills or qualifications compared to inexperienced applicants (Ellingson, 2012;Melchers et al., 2020). However, correlations between demographic variables and faking in past studies are often weak, and Melchers et al. (2020) conclude in their review of the literature that the most robust findings are that age tends to be negatively related to faking, and that men fake somewhat more than women. Other studies have also found men more likely than women to engage in heavier forms of impression management (Guadagno & Cialdini, 2007;Turnley & Bollino, 2001). Although we found that age and work experience were unrelated to faking intentions (despite ample variation in the sample), we did find that men reported higher willingness to fake than women. This is in line with traditional gender roles; whereas the feminine gender role does not encourage self-promotion, the masculine gender role does (Hogue et al., 2013;Wade, 2001). However, our study was not set up to examine gender differences, and it is therefore possible that for example differences in occupational sector (that the men work with different things than the women, and that they thus imagined different types of jobs when answering the faking measure) explains this finding.

Practical Implications
Organizations may want to be attentive to the fact that faking behavior can be socially contagious in the sense that descriptive norms may influence applicants' motivation to fake. Our results suggest that applicants may be influenced particularly by honesty norms. This implies that if we can help applicants to get a less pessimistic view of how common faking is, perhaps faking prevalence will go down. Previous findings suggest that perceived norms can change when people receive accurate information about what others in their group or community do and approve of ("correcting misperceptions"; Chung & Rimal, 2016;Morris et al., 2015). Such strategies may also work when people are unaware of how common the behavior in question is (Legros & Cislaghi, 2020). However, if faking is more common Personnel Assessment And decisions sociAl norms And FAking interviews than people believe, this cannot be accomplished without using deception. Therefore, future work should establish the accuracy of perceptions before the potential benefit of such interventions can be determined. Perhaps a correcting misperceptions intervention should be targeted at extensive image creation (falsifying one's answers), as this might allow for correcting actual misperceptions without the need for deception.

Limitations and Suggestions for Research
The strengths of the present research include a theoretically supported research question, a preregistered experiment designed with sufficient statistical power, and manipulations and outcome measures that have been used in previous studies. However, there are also limitations. First, we used a general faking measure instead of focusing on previously proposed subfacets (e.g., inventing, borrowing) of faking behavior (Levashina & Campion, 2007;Melchers et al., 2020). Although the internal consistency was acceptable, future studies might want to take a more nuanced approach. For example, "high-stakes" lies are generally less socially accepted than "low-stakes" lies (Guthrie & Kunkel, 2014). It is possible that mild forms of faking (e.g., exaggerations) are seen as low-stakes lies and that compliments are regarded as social lubricant, whereas other types of faking such as extensive image creation are less normative and may thus be less affected by conveyed norms.
Regarding alternative explanations for the finding that the honesty norm condition reduced intentions to fake, the critical reader may argue that the honesty norm condition also served as a moral value prime. However, that is why we ensured that the concept of honesty was also mentioned in the instructions given to the participants in the control condition. Thus, we believe that it is unlikely that the observed difference between the honesty and control conditions is explained by moral value salience. Rather, it should be the information about what most other people do that explains the difference.
Regarding generalizability, we note that the sample was diverse in age, and that we did not rely on a student sample. Almost everyone had attended employment interviews and had work experience. However, the sample lacked ethnic diversity: Participants were British, and we cannot generalize the findings to all cultures. For example, intention to fake have been found positively related to power distance and ingroup collectivism, and U.S. applicants fake more than applicants from Iceland and Switzerland (Fell et al., 2016). It is thus possible that descriptive (and injunctive) norms related to faking behavior also differ between cultures and/ or between different sectors or occupations (Melchers et al., 2020). We recommend that future research examine these possibilities.
Another limitation is that our faking data are restricted to self-reports. Even though people's intentions toward work-related behaviors are often linked to performing those behaviors (Taylor & Small, 2002), anticipating an interaction is not the same as actually experiencing it (Dunbar et al., 2003). Willingness to fake may not translate directly into actual faking behavior, as contextual factors may strengthen or weaken the influence of the norm (Chung & Rimal, 2016). Previous research suggests that applicants are in part consistent in their level of faking across interviews  but also that interview faking is situation specific (Levashina & Campion, 2006). For example, the perceived costs of engaging in faking could be of importance, as could perceived competition for the job. According to Levashina and Campion's (2006) model of interview faking, willingness to fake must be accompanied by the capacity (e.g., verbal or social skills) and the opportunity (e.g., interview format) to do so in order for faking behavior to occur. Our findings should ideally be replicated with actual faking behavior during real interviews, but regrettably, this is difficult to accomplish in practice.
Furthermore, we focused on general effects of norms, and future research might want to examine interactions between effects of conveyed norms and theoretically relevant personality traits. Some individuals are in general more susceptible to social influence than others (Briñol & Petty, 2005). Moreover, an applicant's competitive worldview or perceived competition on the labor market (Ho et al., 2019;Schilling et al., 2020) might interact with sensitivity to conveyed norms. For example, applicants with a competitive worldview who learn that most applicants fake might interpret this as increased competition for the job, and conversely, if they learn that most applicants are honest, this might lessen their perceived competition and should make them less motivated to fake.
To conclude, this study provides initial experimental evidence that applicants' self-reported willingness to fake in employment interviews is influenced by their perceptions of how other applicants behave. We hope that the findings will spawn more research on this topic, with the ultimate goal of developing subtle, yet effective, social norm interventions that can reduce faking behavior during real employment interviews.