Economic Predictors of Differences in Interview Faking Between Countries: Economic Inequality Matters, Not the State of Economy

Many companies recruit employees from different parts of the globe, and fak- ing behavior by potential employees is a ubiquitous phenomenon. It seems that applicants from some countries are more prone to faking compared to others, but the reasons for these differences are largely unexplored. This study relates country-level economic variables to faking behavior in hiring processes. In a cross-national study across 20 countries, participants ( N = 3,839) reported their faking behavior in their last job interview. This study used the random response technique (RRT) to ensure participants’ anonymity and to foster honest answers regarding faking behavior. Results indicate that general economic indicators (gross domestic product per capita [GDP] and unemployment rate) show negli-gible correlations with faking across the countries, whereas economic inequality is positively related to the extent of applicant faking to a substantial extent. These findings imply that people are sensitive to inequality within countries and that inequality relates to faking, because inequality might actuate other psychological processes (e.g., envy) which in turn increase the probability for unethical behavior in many forms.


INtRODUCtION
When applicants are interviewed for a job, they do not always represent their true abilities and skills, some opt to fake-they intentionally distort or even falsify their responses to create a specific impression (Levashina & Campion, 2006). Faking occurs not only in interviews but also in personality tests (e.g., Birkeland, Manson, Kisamore, Brannick, & Smith, 2006), and it is a phenomenon about which many practitioners are concerned (e.g., Robie, Tuzinski, & Bly, 2006) because applicants who fake gain an unfair advantage over non-faking applicants.
Faking has already been established as a phenomenon whose extent varies between countries (e.g., Bye et al., 2011;Fell, König, & Kammerhoff, 2016;Frei, Yoshita, & Isaacson, 2006;König, Wong, & Cen, 2012). These differences between countries are relevant for all organizations that recruit in more than one country, ranging from small companies that are situated at the border of two countries and thus have employees from both countries, to large organizations such as the European Union that try to attract applicants from all member states (see Christensen, 2015).
Despite the importance of these differences in faking tendencies between countries, attempts to explain these differences are rare. In this paper, we argue that economic variables (e.g., unemployment rates) are correlated with these differences in interview faking. Therefore, we link economic predictors to prevalence rates of interview faking in 20 countries. We measure faking prevalence with the randomized response technique (RRT) that has been developed for the measurement of sensitive topics. The RRT prevents researchers from being able to identify whether a response by a participant was due to a randomization device (in our case: dice) or due to their actual behavior (Fox & Tracy, 1986). Such conditions increase the likelihood of honest answers and, thus, have been applied to research applicants' faking behavior in the past (e.g., Donovan, Dwight, & Hurtz, 2003).

economic Predictors of Faking Differences Between Countries
National State of the Economy. Several theories (e.g., Marcus, 2009;McFarland & Ryan, 2000) argue that situational factors influence faking, with the state of the economy being an important situational aspect-the argument being that a healthier economy reduces the motivation to fake. When a country's economy is doing well, applicants may assume that many organizations are hiring because there are many job advertisements. This shifts the hiring market power towards the applicants. Consequently, applicants may have a reduced need to exaggerate their abilities. In other words, in countries that have a healthy economy and many vacancies (possibly even more vacancies than applicants), applicants should have fewer reasons to fake than in countries where people are struggling to find a job.
There are a few possible indicators for the state of the economy of a country. In particular, many cross-cultural studies have used the gross domestic product per capita (GDP) as their operationalization of the state of the economy of a country (see, e.g., Heath, Richards, & de Graaf, 2016). GDP per capita can be defined as the value of all final goods and services of a country divided by its population in a year in current US dollars, and it allows for the comparison of living conditions across countries (World Bank, n.d.). Prior research has found the GDP to be negatively related to several kinds of unethical behaviors (e.g., corruption, You &Khagram, 2005, andacademic cheating, Orosz et al., 2018). Even more importantly, Robie, Emmons, Tuzinski, and Kantrowitz (2011)  One could also argue that it is the unemployment rate, another indicator for the state of the economy of a country that might affect applicants' faking behavior. Although applicants might be more or less aware of the general state of the economy of their country, they primarily care about getting a job, and a high unemployment rate increases the importance of finding a job (König, Hafsteinsson, Jansen, & Stadelmann, 2011;Marcus, 2009; but see . Thus, we hypothesize: The lower the GDP per capita (H1a) and the higher the unemployment rate (H1b) of a country, the more people in a country engage in interview faking.
Economic Inequality. It might also be possible that it is not the state of the economy per se, but economic inequality that triggers unethical behavior such as faking: The differences between the rich and the poor within a country might psychologically matter more than the average state of the economy of a country. In particular, less affluent people in high-inequality countries have more to win if they are successful in the world of work, for example by getting well-paid jobs, whereas affluent people in high-inequality countries might have more to lose if they are not successful (Buttrick & Oishi, 2017;You & Khagram, 2005). These arguments are bolstered by the empirical finding of country-level differences in economic inequality being positively related to country-level differences in corruption (Zhang, Cao, & Vaughn, 2009), software piracy (Husted, 2000), and crime in general (Pratt & Cullen, 2005). Furthermore, economic inequality might also trigger the people's perception that they live in a very competitive world, and competitiveness has been argued (Roulin, Krings, & Binggeli, 2016) and shown to be related to interview faking (e.g., Schilling, Roulin, Obschonka, & König, 2020). Consequently, living in a country with high inequality might make faking a particularly attractive strategy to get ahead.
Thus, we hypothesize: The higher the economic inequality of a country, the more people in a country engage in interview faking (H2).

MetHOD Sample
Students and recent graduates in 20 countries were asked about their faking behavior during their most recent application (for country details see Table 1). Data were mostly collected directly by the collaborators using online surveys (see Table 1). The data for the People's Republic of China have been published in König et al. (2012); the data for Switzerland in König et al. (2011), and the data for the United Arab Emirates (   Note. Exclusion because of inconsistencies means that participants stated that they had a job but indicated in the RRT question "I am currently" employed that they do not have a job. A "-" indicates that the data were missing or questions were not asked during the study (see in particular Donovan et al., 2003;König et al., 2011;König et al., 2012). a Participants threw real dice (provided by the research team).
b Participants were given a randomly generated list of "1" and "0" and asked to mark the response option "true" if there was a "1" on their list and to give the correct (truthful) answer if the list showed a "0". Langer, and König (2018). For these three countries, the raw data were available and re-analyzed for the current publication. The data for Iceland and the US were directly taken from König et al. (2011) and Donovan et al. (2003), respectively. The final sample (see Table 1 for exclusion of participants) consisted of N = 3,839 applicants from 20 countries. Across countries, the mean age was 23.28 (SD = 6.07), 59.9 percent were female; 20.7 percent indicated that they had already experienced one or two job interviews in their life, and 61.4 percent indicated that they already had three or more job interviews in their life. Most of the participants were undergraduates (36.1%), nearly as many were graduate students (31.8%), and a smaller percentage already had their master's degree, a comparable or a higher degree (10.2%). Further details about all samples can be found in Table 2.

Procedure
To introduce the participants to the topic of job interviews, the questionnaire started with five items about their experience with their most recent job interview. Two sample items for this phase were "When did your last job interview take place?" and "How strongly did you wish to get the job?" Next, participants were introduced to the RRT technique. We explained that the technique ensures anonymity because the answers to the RRT items depend on a randomization device (dice in our case) and that no researcher is able to identify if the answer participants give are caused by the randomization device or by answering truthfully. In our case, participants were instructed to mark the response option "true" if the dice showed 1 or 2 regardless of their own behavior and to give the correct (truthful) answer if the dice showed the other four faces. Furthermore, we explained to them that researchers can only use these RRT data at the group level because they know that one third of all dice throws results on average on participants crossing "true." After these introductions, participants provided answers to 14 items about their behavior in the last interview. To ensure comparability across papers (Donovan et al., 2003;Husain et al., 2018;König et al., 2011;König et al., 2012), this paper focuses on the 11 items that were asked in all countries; these items can be found in Table 3. The questionnaire ended with several demographic questions. (The procedure was slightly different in Iceland and in the US, see Donovan et al., 2003, andKönig et al., 2011.) economic Indicators

10.
When applying for the job, I tried to portray myself as more agreeable (trusting, empathetic, cooperative) than I really am.  in which the data were collected) was not available in the World Bank Open Data bank, other sources were consulted (CEIC Data, n.d.;DeNavas-Walt, Cleveland, & Webster, 2003;Laenderdaten, n.d.). For three countries, no inequality data could be located for the same year as the data collection and we thus used the data available for any prior year closest to the data collection (i.e. for New Zealand [a three-year lag], India [a one-year lag], and Fiji [a fouryear lag]). Although inequality can be measured in several ways, researchers seem to have agreed that the best way to do so is to use the national Gini index (e.g., Bleidorn et al., 2016;Steptoe, Ardle, Tsuda, & Tanaka, 2007;You & Khagram, 2005). This index ranges from 0 (i.e. perfect income equality within a country) to 100 (i.e. perfect inequality).

Preliminary analyses
For all countries where we had raw data we used the same criteria to exclude participants with questionable data: (a) if participants were incomplete; (b) if they reported that they did not predominantly live in the country of data collection; (c) if they did not have at least one interview in the last year; (d) if they reported that their skill in the language of the questionnaire was lower than intermediate; (e) if they admitted that they did not or just partially followed the instructions of the RRT; (f) if they produced inconsistent data (e.g., if they mentioned that they had had a job interview in the demographics section, but, when they were asked the same question in the RRT section they answered "no") (see also Table 1). Table 3 reports the RRT corrected percentages of agreement to each item for each of the 20 countries (for calculating these percentages see the Appendix). To prevent having to correlate the economic predictors with single items with unknown reliability, we averaged all 11 items and thus created an overall faking score for each country. Reliability of this overall faking score at the country-level was Cronbach's α = 0.88. Table 4 shows means, standard deviations, and correlations of the variables. As can be seen, interview faking did not significantly correlate with GDP per capita (r = −0.10, p = .68) nor with unemployment (r = −0.16, p = .50), but is significantly and strongly correlated with inequality (r = 0.53, p < .05). These results do not provide support for Hypotheses 1a and 1b but do support Hypothesis 2.

DISCUSSION
This study investigated the correlation between country-level economic variables and interview faking tendencies for 20 countries from a wide range of world regions. Our findings show that indicators of economic wealth of a country (i.e. GDP, unemployment rates) were not related to faking, whereas an indicator of economic inequality within the countries was strongly associated with faking. On average, people from countries with high economic inequality also reported more faking in job interviews.
The results of this study highlight the importance of economic inequality for faking. In countries where there is more inequality, where the gap between rich and poor is wider, people seem to be more inclined to fake, possibly because they have more to gain or to lose depending on the income strata in which they are currently situated (You & Khagram, 2005). Similar to prior research (Pratt & Cullen, 2005;Zhang et al., 2009) our results indicate that inequality matters and that inequality can drive people to engage in unethical behavior. Our results are also consistent with the argument that inequality might trigger the perception that applicants have to compete against each other, because perceived competition is also known to be linked to interview faking . It seems also plausible that inequality and faking reciprocally influence each other-not only might inequality push people towards increasing their chances to get a job through questionable behavior, faking might give some people undeserved access to well-paid jobs, which subsequently leads to further inequality (see also You & Khagram, 2005). In addition, if inequality is large, less affluent people might particularly envy the affluent ones and conclude that it is not fair that the affluent are so wealthy, possibly by luck or due to their family background or even by engaging in unethical behavior, and that others are left behind (Ben-Ze'ev, 1992;de Vries, Pathak, van Gelder, & Singh, 2017;Hirschman & Rothschild, 1973). It is possible that the concentrated power by the wealthy in income unequal countries even leads to country-level norms of "anything goes," which would then lead to applicants feeling more free to do whatever it takes (including faking) to get a job (cf. Gino & Pierce, 2009). Interestingly, we found strong effects for inequality despite the fact that the students and recent graduates who participated in our study likely did not belong to the class of very poor in their countries because university education is often rather expensive. Although this fact operates against the hypothesized relationship (making for a more conservative test of our hypothesis), it should be noted that inequality also persists between university students: Some cannot afford to go to their favorite or the most prestigious university; others need to take student loans and pay off their student debt for the rest of their lives, whereas students who are better off start their professional lives without concerns about debt. Although we acknowledge that inequality between university students might be less salient than between the overall population of a country, perceived economic inequality still affects university students, which might lead to faking behavior. In contrast to the findings for the inequality index, no relationship was found for two of the most important national economic indicators-GDP per capita and unemployment (the correlation with unemployment even went in the wrong direction). This means that economic wealth might be unrelated to applicants' faking behavior, which could be explained by the following two factors. First, the GDP and the unemployment rate are rather coarse indicators of individual wealth within a country, and even if the GDP rises this might only be because the wealthy get even wealthier (which might especially be true in countries with low income equality; see Buttrick & Oishi, 2017). Second, even if, for example, Brazilian applicants are aware of the state of the Brazilian economy, they might not compare the wealth of Brazil with the wealth of, for example, Iceland and thus might not conclude that they are relatively poor. Unemployment rates did not affect faking behavior either. Thus, applicants might be aware of the challenges involved in getting a job without responding to these challenges through more faking. Perhaps unemployment is a less serious reason for people to envy other people (i.e. others who have a job) than economic inequality (Ben-Ze'ev, 1992;Blanchard & Summers, 1986;Hamilton, 1988).
To summarize, our findings indicate that people are more sensitive to inequality than to general economic indicators when it comes to faking-it is not the overall economic wealth of a country that is driving faking behavior but how that wealth is distributed (although it should be kept in mind that this interpretation of our results is based on correlational data). If people perceive that there is inequality in the country, this might evoke unethical behavior (see also Pratt &Cullen, 2005 andZhang et al., 2009). More generally, it seems that inequality is more likely to lead to other psychological processes (e.g., envy and competitiveness) that trigger faking than the general state of the economy within a country.

limitations and Future Research
At least four limitations deserve being mentioned. First, the design of our study is correlational and thus does not allow causal conclusions. Second, more consistency in the way data were collected would have been preferable, but practical challenges in doing multi-national research makes consistency difficult to achieve, and for such practical reasons, many large surveys (e.g., Transparency International's Global Corruption Index, Hardoon & Heinrich, 2013) use, for instance, face-to-face plus online methods to collect data. Third, all our participants were either current students or recent graduates with only some experience in applying for jobs. It remains open, how much the results generalize to older job seekers and how well one sample can represent a country, and future research should thus try to collect data from older and more diverse samples. Fourth, the use of the RRT has clear benefits (i.e. higher anonymity), but it comes with the challenge that a considerable number of participants in every country did not or only partially followed the instructions. An additional disadvantage is that the RRT does not allow analyzing data at the individual level, which means that measurement equivalence cannot be statistically established (cf. Boer, Hanke, & He, 2018).
Future research should include additional variables beyond economic predictors that were the focus of this study. In particular, previous research has suggested that cross-country differences in faking correlates with three cultural dimensions of humane orientation, in-group collectivism, and gender egalitarianism ; see also Fell & König, 2020). Furthermore, future research should also study other kinds of ethically questionable behavior within the field of personnel selection (e.g., applicants unfairly using their influence and networks to get job interview invitations) and beyond and explore possible links to economic inequality. In addition, one could argue that the increased prevalence of fake news (at least in some countries) might make faking in personnel selection situations more acceptable, which could be empirically tested. Finally, future research could also examine how applicants adapt their application strategies when they are recruited in other countries than their home countries. This would allow exploring how much the home country has shaped their application behavior.  The following example explains how the RRT corrected percentages of agreement to each item was calculated. Imagine we collected data from 12 people in Avalon and asked the participating Avalonians to throw a dice before answering whether they used witchcraft to make a better impression in their last job interview. We instructed all of them to tick "yes" when the dice showed a 1 or a 2 and to answer truthfully otherwise, which means that 1/3 should tick "yes" just because of the dice. Looking at the data, we find that 8 Avalonians ticked "yes" and 4 "no." Because 1/3 equals the probability of being forced by the dice to tick "yes," 1/3 of the overall sample (i.e. 4 Avalonians) have to be subtracted from these 8 "yes" ticking Avalonians. The outcome of this is 4 Avalonians who honestly ticked "yes." In comparison with those 4 who ticked "no," this means: half of the Avalonians used witchcraft to make a better impression in their last job interview, and the other half did not. This is also illustrated by Figure A1.