Cross-Cultural Differences in Applicant Faking on Personality Tests: A 43-Nation Study
We thank Johannes Schult for his helpful comments on our statistical analyses, and the countless questioners and experts in R online forums. We thank Nida Bajwa for his helpful feedback. We thank the participants of the 9. Nachwuchsworkshop der Fachgruppe Arbeits-, Organisations- und Wirtschaftspsychologie der Deutschen Gesellschaft für Psychologie for their feedback on the research idea. We thank Boguslaw Pawlowski, Claire Würger, Corina Möller, Douglas P. Fry, Gahyun Youn, Ha-Na Seo, Matthias Stadler, Matti Keloharju, Michel Raymond, Miriam Junghans, and Ying Zhang for their help in collecting the available language translations of the Big Five Inventory. A previous version of this manuscript was presented at the 17th European Congress of Work and Organizational Psychology in Oslo, Norway.
Abstract
In a globalised world, more and more organisations have to select from pools of applicants from different cultures, often by using personality tests. If applicants from different cultures were to differ in the amount of faking on personality tests, this could threaten their validity: Applicants who engage in faking will have an advantage, and will put those who do not fake at a disadvantage. This is the first study to systematically examine and explain cross-cultural differences in actual faking behavior. In N = 3,678 employees from 43 countries, a scenario-based repeated measures design (faking vs. honest condition) was applied. Results showed that faking differed significantly across countries, and that it was systematically related to countries’ cultural characteristics (e.g. GLOBE's uncertainty avoidance, future orientation, humane orientation, and in-group collectivism), but in an unexpected way. The study discusses these findings and their implications for research and practitioners.
Because sometimes the truth isn't good enough. Sometimes people deserve more. (Batman in Nolan's (2008) The Dark Knight)
Introduction
Organisations all over the world use personality tests to select appropriate candidates (e.g. Bartram, 2004). At the same time, applicant faking is considered one of the major threats to personality tests in personnel selection (Morgeson et al., 2007a, 2007b). Despite this, cross-cultural differences in faking on personality tests during personnel selection have been widely ignored (Frei, Yoshita, & Isaacson, 2006).
If there are cross-cultural differences in applicant faking behavior, international personnel selection has to be aware of them, or it will otherwise run the risk of systematically disadvantaging applicants from cultures which are rather more reluctant to fake. Although some groundwork has been laid in this regard (e.g. Fell, König, & Kammerhoff, 2016; König, Hafsteinsson, Jansen, & Stadelmann, 2011; König, Wong, & Cen, 2012b; Sandal et al., 2014), cross-cultural differences in actual faking behavior have not yet been examined. This may seem particularly surprising considering the increasing practical relevance of faking as globalisation and international personnel selection progress (Chand & Tung, 2014; P. Evans, Pucik, & Bjorkman, 2011; Javidan, Dorfman, de Luque, & House, 2006; Lievens, 2008; Ryan, Leong, & Oswald, 2012). Applicants’ cultural backgrounds provide a promising opportunity to gain an understanding of differences in their faking behavior during personnel selection across countries (see also Fell et al., 2016). Using a within-subjects faking design (Burns & Christiansen, 2011; Peterson & Griffith, 2006), we examine cross-cultural differences in actual faking behavior across 43 countries. The rationale of this study reflects that of a previous study on cross-cultural differences in applicant faking (Fell et al., 2016). However, this study significantly extends the understanding of cross-cultural differences in faking by relying on direct behavioral measures instead of measuring attitudes.
Theoretical Background
The theoretical background of our study is divided into two sections. First, we will present a short introduction to applicant faking. We will argue that it is a real, common phenomenon which differs systematically and is often detrimental. Second, we will propose nine hypotheses that may explain why there are cross-cultural differences in applicant faking (cf. Fell et al., 2016; House, Hanges, Javidan, Dorfman, & Gupta, 2004).
Prevalence Rates, Differences, and Consequences of Applicant Faking
We define applicant faking behavior as applicants’ conscious distortion of responses in order to achieve better scores (e.g. on a personality test) and to increase their chances of being hired (Levashina & Campion, 2007; McFarland & Ryan, 2000). There is strong reason to believe that faking in non-cognitive personnel selection methods is a prevalent behavior among applicants (e.g. Ellingson, Sackett, & Connelly, 2007; Levashina & Campion, 2007; McFarland & Ryan, 2000; Robie, Brown, & Beaty, 2007; Uziel, 2010). Donovan, Dwight, and Hurtz (2003) revealed an average faking base rate of 29 per cent. Using the same survey technique of maximum anonymity, König and his colleagues (König et al., 2011, 2012b) presented faking as a class of behaviors (e.g. downplaying negative attributes, exaggerating agreeableness) in which the majority of applicants engage in one way or another. This picture also applies to personality tests, in which about half of applicants have been found to engage in faking (Griffith, Chmielowski, & Yoshita, 2007), and is supported by personnel experts’ view that up to 50 per cent of applicants fake (Rees & Metcalfe, 2003; Robie, Tuzinski, & Bly, 2006).
Systematic individual variance exists in applicant faking, and can be considered as more problematic than the high prevalence rates per se. If all applicants distorted their answers to the same extent, there would be no shifts in applicant rank orders, no matter how prevalent the faking (Tett et al., 2006). In such a “perfect world”, faking would not be a serious problem for personnel selection. But according to several studies, applicants differ systematically in terms of faking, depending on various individual characteristics such as personality, mental ability, or demographic characteristics (e.g. Griffith, Malm, English, Yoshita, & Gujar, 2006; Hogue, Levashina, & Hang, 2013; Levashina, Morgeson, & Campion, 2009; McFarland & Ryan, 2000; for an overview see Tett et al., 2006). High prevalence rates and systematic differences leave the door open for considerable and detrimental consequences of faking.
Faking continues to be a hotly debated topic (e.g. Morgeson et al., 2007b). Some studies have led researchers as well as practitioners (Rees & Metcalfe, 2003; Robie et al., 2006) to claim that faking impairs the utility of non-cognitive personnel selection methods such as personality tests, as these studies suggest that applicants can successfully fake their way into jobs (e.g. Berry & Sackett, 2009; Donovan, Dwight, & Schneider, 2014; Komar, Brown, Komar, & Robie, 2008; Mueller-Hanson, Heggestad, & Thornton, 2003). However, other researchers do not share this conviction because they do not consider faking as a threat (e.g. Ones, Viswesvaran, & Reiss, 1996)—some even suggest that faking has positive effects on job performance (e.g. Ispas et al., 2014). Yet other researchers hold a more balanced view on faking and its consequences, after showing that relationships between faking and job performance vary across selection methods and operationalisations (e.g. Ingold, Kleinmann, König, & Melchers, 2015; O'Connell, Kung, & Tristan, 2011; see also Peterson & Griffith, 2006; Peterson, Griffith, Isaacson, O'Connell, & Mangos, 2011).
Cross-Cultural Differences in International Personnel Selection
Applicant faking has received much monocultural but surprisingly little cross-cultural research attention (P. Evans et al., 2011; Javidan et al., 2006; Lievens, 2008; Ryan et al., 2012). Cross-cultural differences in actual faking behavior have been widely ignored (cf. Dipboye, Macan, & Shahani-Denning, 2012; Frei et al., 2006), which seems surprising in view of practical necessities and given the existing links between theories on applicant faking and on cross-cultural behavioral differences. When deciding whether or not to behave in a certain way, people think about whether or not such behavior would be common in their culture: If it is (not) very common, cultural members will feel pressured to (not) engage in it (Frese, 2015; see also Eriksson, Strimling, & Coultas, 2015). In the context of applicant faking, applicants can accordingly be assumed to engage more in distorting responses to increase their chances of being hired if they assume this to be common among applicants of their culture.
In various theoretical faking models (e.g. McFarland & Ryan, 2000, 2006; Mueller-Hanson, Heggestad, & Thornton, 2006; Snell, Sydell, & Lueke, 1999), norms assume important roles for the emergence or absence of applicants’ faking (Ziegler, MacCann, & Roberts, 2012). Snell et al. (1999) considered subjective pervasiveness and acceptance of faking as crucial for applicants’ motivation to actually fake. Similarly, McFarland and Ryan (2000) suggested that values and morals influence applicants’ intention to fake (for empirical evidence, see McFarland & Ryan, 2006). Mueller-Hanson et al. (2006) linked subjective norms (What will others think of me faking?) theoretically and empirically to applicants’ willingness and intention to fake.
Considering cultural practices as “shared perceptions of how people routinely behave in a culture” (Frese, 2015, p. 1327), it seems rather surprising that empirical studies have rarely addressed cross-cultural differences in faking behavior. There are only a small number of exceptions. Comparing faking prevalence rates (e.g. exaggerating skills, de-emphasising negative attributes, portraying oneself as more agreeable) among applicants in four countries, König et al. (2011, 2012b) showed that prevalence rates in Iceland and Switzerland were substantially lower than those in China and the US. Sandal et al. (2014) surveyed students from 10 countries in terms of the importance they assigned to several self-presentation tactics (expressing excellence, mentioning external obstacles, and stressing assertiveness and accommodation), and found indications of culturally and economically caused differences. In a survey of employees from 31 countries, Fell et al. (2016) revealed substantial country-level relationships between attitudes toward (mild/severe forms of) applicant faking and cultural dimensions of the Global Leadership and Organisational Behavior Effectiveness project (GLOBE; House et al., 2004). Despite their individual limitations (e.g. in part no measures of actual faking behavior, small number of countries), these studies have created awareness of cross-cultural differences in applicant faking and have taken first steps to explain them.
However, three crucial questions have been left unanswered: (a) Are there systematic cross-cultural differences in actual, directly measured faking behavior? (b) If there are cross-cultural differences in directly measured faking behavior, can they be explained by applicants’ cultures? (c) If these cross-cultural differences can be explained by applicants’ cultures, are the country-level relationships between culture and faking behavior consistent with those established by Fell et al. (2016) for attitudes toward applicant faking? Relying on the cultural framework of GLOBE (House et al., 2004) and Fell et al.'s (2016) research, we will put forward nine hypotheses on why systematic cross-cultural differences in applicant faking behavior should emerge.
Thus, this study aims to explain the phenomenon of cross-cultural differences in applicant faking by embedding it into the cultural framework of GLOBE (House et al., 2004).
Uncertainty Avoidance: Faking Can Be an Adventure
The risk of getting caught is one of the costs of faking (Bangerter, Roulin, & König, 2012). After all, even applicants who are aware of the low risk of actually getting caught (Robie et al., 2006) cannot be absolutely certain that their faking will not be exposed: Someday, they might be unable to master a task that they were supposed to be able to master (e.g. fluency in a foreign language, stress resilience). Therefore, cultures’ avoidance of uncertainty due to an emphasised social order (i.e. “societal corset” of norms, rituals, and bureaucracy; House & Javidan, 2004) should contradict applicant faking. Moreover, the disadvantageous effects on others caused by faking should contradict high uncertainty avoidance cultures’ tendency to follow rules and their low tolerance for breaking them (de Luque & Javidan, 2004; Venaik & Brewer, 2010). In line with this argument and with previous research on the justification of ethically questionable behavior (Parboteeah, Bronson, & Cullen, 2005) and corruption (Seleim & Bontis, 2009), Fell et al. (2016) also showed a strong negative correlation between a positive attitude toward faking and uncertainty avoidance. Thus, we hypothesise lower levels of faking in high uncertainty avoidance countries (H1).
Future Orientation: Faking Can Be Shortsighted
Engaging in faking behavior means that long-term behavioral consequences have been traded off for short-term ones (Mead, Baumeister, Gino, Schweitzer, & Ariely, 2009). In the short term, faking can provide applicants with better chances of being hired. In the long term, faking can have negative effects, such as a poorer “person–job fit” (Pace & Borman, 2006) or difficulties “to maintain a positive moral self-image and future relationships” (Gino, Schweitzer, Mead, & Ariely, 2011, p. 192). Therefore, applicants from high future orientation cultures, who “engage … in planning, investing in the future, and delaying individual or collective gratification” (House & Javidan, 2004, p. 12), should rather prefer long-term over short-term advantages, i.e. refrain from faking. In this context, Fell et al. (2016) showed a nominal negative correlation between positive attitudes toward mild applicant faking and future orientation on a country level. In line with this, Seleim and Bontis (2009) showed that corruption is barely compatible with a strong future orientation. Thus, we hypothesise lower levels of faking in high future orientation countries (H2).
Power Distance: Faking Can Be Poetic Justice
It has been argued that a perceived lack of fairness in personnel selection makes applicants engage in faking behavior (Heggestad, 2012; Snell et al., 1999; see also Dalal, 2005). In line with such a suggestion, participants of an applicant scenario reported less faking when they considered the selection process to be fair (McFarland, 2003). In high power distance cultures, people accept that power is not fairly distributed across organisational or societal levels, but is located at the top (House & Javidan, 2004). In such cultures, not everyone has the same rights, and the powerless are dominated by the powerful (Carl, Vipin, & Mansour, 2004). People in these cultures report higher corruption, lower general satisfaction, lower civic engagement, lower gender equality, and lower civil liberties (Carl et al., 2004; Seleim & Bontis, 2009).1 Therefore, they might try to find alternative ways to achieve “some degree of protection against abuse of power” (Carl et al., 2004, p. 557) and adapt their own behavior to the relatively high prevalence of ethically questionable behavior by engaging in faking behavior (cf. Snell et al., 1999). Accordingly, Fell et al. (2016) showed a strong positive correlation between a positive attitude toward faking and power distance. Thus, we hypothesise higher levels of faking in high power distance countries (H3).
Institutional Collectivism: Faking Can Put Others at a Disadvantage
In their endeavor to improve their chances of being hired, applicants who fake put themselves at an advantage by putting others—the organisations that are selecting and the other applicants who refrain from faking—at a disadvantage (Tett et al., 2006). Applicants who fake might consider their chance of gaining a competitive edge over the other applicants (see also Roulin, Krings, & Binggeli, 2016) to be more important than that of others (i.e. giving the job to the best applicant). However, disregarding the interests of others should be considered particularly problematic in cultures high in institutional collectivism, because these cultures “encourage and reward collective distribution of resources and collective action” (House & Javidan, 2004, p. 12) and put group interests before those of the individual (Gelfand, Bhawuk, Nishii, & Bechtold, 2004). This argument (from Fell et al., 2016) also fits with the findings that institutional collectivism is related to stronger mutual trust (Realo, Allik, & Greenfield, 2008), less justification of unethical behavior (Parboteeah et al., 2005), and less corruption (Seleim & Bontis, 2009). Thus, we hypothesise lower levels of faking in high institutional collectivism countries (H4).
Humane Orientation: Faking Can Be Just a Peccadillo
Applicants and recruiters often agree that a certain faking behavior can be a peccadillo, an excusable mistake (Jansen, König, Stadelmann, & Kleinmann, 2012). Some cultures are characterised more than others by people who are “fair, altruistic, friendly, generous, caring, and kind to others” (House & Javidan, 2004, p. 13) and tolerant of mistakes (GLOBE Project Team, 2006). Such highly humane-oriented cultures seem to demonstrate more understanding for ethically questionable behavior (Seleim & Bontis, 2009). Fell et al. (2016) showed that more humane-oriented cultures had more positive attitudes toward applicant faking in job interviews. Therefore, in such mistake-tolerant humane-oriented cultures, faking should be more common because it is intermittently reinforced (i.e. possibly receiving the job) and rarely punished (Mazur, 2002). Thus, we hypothesise higher levels of faking in highly humane-oriented countries (H5).
Performance Orientation: Faking Can Undermine Progress
Several studies have suggested that faking can undermine organisational performance (e.g. Kristof-Brown, Zimmerman, & Johnson, 2005; Peterson & Griffith, 2006; Peterson et al., 2011; Rosse, Levin, & Nowicki, 1999): This occurs directly when faking is negatively correlated with job performance or positively correlated with counterproductive work behavior. It occurs indirectly when applicants who fake crowd out applicants with higher true scores, who might possibly have a better person–job fit. Therefore, faking is a potential obstacle for people in high performance orientated cultures because they are characterised by a striving for actual improvement and excellence (House & Javidan, 2004). For example, high performance oriented cultures typically reward people (e.g. students) for achieving great things (e.g. excellent exam results) and encourage them to become better and better (GLOBE Project Team, 2006). However, such rewards and encouragement are based on the assumption that students deserve them—that they have not faked. Accordingly, a culture can only have high performance orientation if excellence is achieved by engaging in honest and hard work (and having the appropriate abilities), and faking in selection contexts should therefore be considered a negative act (but see also Fell et al., 2016). Thus, we hypothesise lower levels of faking in high performance orientation countries (H6).
In-Group Collectivism: Faking Can Help the Next of Kin
Several studies have shown that in cultures where people feel much pride in and loyalty to their families (House & Javidan, 2004), where family ties are especially strong, and where family and friends are particularly respected (Gelfand et al., 2004), ethically questionable behavior (e.g. bribing) is more prevalent (Mazar & Aggarwal, 2011; Seleim & Bontis, 2009). As employment can provide social security, often not only for the job holder but also for the next of kin (Bracking, 2003), applicants’ faking can be justified by their obligation to help their next of kin and make them proud in high in-group collectivism cultures (cf. Kohlberg, 2008; Thackray, Tryba, & Griffith, 2013). Furthermore, if applicants fake and consequently achieve a potentially well-paid job, they can use their earnings to support their families, and job prestige might even radiate to other family members. This is why people in high in-group collectivism cultures have more positive attitudes toward faking (Fell et al., 2016). Thus, we hypothesise higher levels of faking in high in-group collectivism countries (H7).
Gender Egalitarianism: Faking Can Help to Overcome Barriers
In some situations, applicants may fake not to have greater chances of being hired than their competitors, but to have merely the same chances: Women might be recommended to “[c]onform with male stereotypic behaviour” in order to be successful at work (Hayes, Allinson, & Armstrong, 2004, p. 413; Powell, Bagilhole, & Dainty, 2009; van Vianen & van Schie, 1995), at least in cultures that do not value gender equality a great deal and do not try to minimise gender role differences (House & Javidan, 2004). Similarly, men living in such low gender egalitarianism cultures might be recommended to expend extra effort when they wish to work in fields that are culturally considered as female (Schilt & Williams, 2008; Williams, 1992, 1995) or when their own career paths are interrupted (Schneer & Reitman, 1990). Faking might therefore be a way to overcome the barriers caused by low gender egalitarianism, and this has also been mentioned as the reason why people in high gender egalitarianism cultures have less positive attitudes toward faking (Fell et al., 2016). Thus, we hypothesise lower levels of faking in high gender egalitarian countries (H8).
Assertiveness: Faking Can be Trying to Win at All Costs
Applicants engage in faking behavior in order to achieve their goal of being hired, despite obstacles or costs (e.g. better competitors, poorer person–job fit, risk of being caught, ethical concerns). Therefore, faking behavior should be more rewarded—or at least more appreciated—in those cultures where goal-oriented aggressive behavior has quite positive connotations, where modesty is less valued (Den Hartog, 2004; König et al., 2011), where people are “assertive, confrontational, and aggressive in social relationships” (House & Javidan, 2004, p. 12), where results are placed above relationships, and where people “act and think of others as opportunistic” (Den Hartog, 2004, p. 405). In cultural assertiveness, there might even be a notion of Machiavellianism, which has been defined as “a strategy of social conduct that involves manipulating others for personal gain, often against the other's self-interest” (Wilson, Near, & Miller, 1996, p. 285) and includes behavioral tendencies “toward self-promotion, emotional coldness, duplicity, and aggressiveness” (Paulhus & Williams, 2002, p. 557). Accordingly, cultural assertiveness (Parboteeah et al., 2005; but see also Fell et al., 2016) and Machiavellianism (for an overview see Furnham, Richards, & Paulhus, 2013) have been linked to several ethically questionable behaviors—the latter even directly to applicant faking and impression management (Becker & Dan O'Hair, 2007; Hogue et al., 2013; Lopes & Fletcher, 2004; Mueller-Hanson et al., 2006; Sherry, Hewitt, Besser, Flett, & Klein, 2006; Snell et al., 1999). Thus, we hypothesise higher levels of faking in highly assertive countries (H9).
Method
Overview
Data were collected through an online survey. A within-subjects faking design (Burns & Christiansen, 2011; Peterson & Griffith, 2006) was applied. At two measurement times (faking vs. honest condition), employees from 43 countries (i.e. nine cultural clusters,2 N = 3,678; Table 1) were presented with the Big Five Inventory (BFI; e.g. Benet-Martínez & John, 1998; John, Naumann, & Soto, 2008)—a common Big Five personality test with 44 items.
| GLOBE Cluster | n | Language(s) | Male | Female | MAge | SDAge |
|---|---|---|---|---|---|---|
| Anglo | ||||||
| Australia | 88 | English (n = 88) | 48% | 52% | 47.59 | 13.06 |
| Canada | 92 |
English (n = 69) French (n = 23) |
47% | 51% | 44.80 | 11.89 |
| Ireland | 89 | English (n = 89) | 43% | 56% | 44.40 | 10.87 |
| New Zealand | 96 | English (n = 96) | 53% | 46% | 51.15 | 12.09 |
| South Africa | 97 | English (n = 97) | 55% | 43% | 41.66 | 12.34 |
| UK | 92 | English (n = 92) | 50% | 50% | 45.11 | 12.19 |
| US | 91 | English (n = 91) | 52% | 48% | 49.03 | 12.30 |
| Confucian Asia | ||||||
| China | 73 | (Modern) Chinese (n = 73) | 52% | 48% | 36.22 | 9.58 |
| Hong Kong | 65 |
(Traditional) Chinese (n = 61) English (n = 4) |
46% | 52% | 33.31 | 9.89 |
| Japan | 77 | Japanese (n = 77) | 49% | 51% | 46.78 | 10.05 |
| Singapore | 62 | English (n = 62) | 55% | 42% | 38.31 | 11.08 |
| South Korea | 71 | Korean (n = 71) | 65% | 35% | 40.04 | 9.90 |
| Taiwan | 73 | (Traditional) Chinese (n = 73) | 47% | 51% | 37.12 | 9.83 |
| Eastern Europe | ||||||
| Czech Republic | 87 | Czech (n = 87) | 53% | 46% | 38.71 | 10.35 |
| Greece | 97 | Greek (n = 97) | 51% | 48% | 36.56 | 7.65 |
| Hungary | 84 | Hungarian (n = 84) | 48% | 52% | 39.93 | 8.59 |
| Kazakhstan | 79 | Russian (n = 79) | 56% | 42% | 34.48 | 9.51 |
| Poland | 90 | Polish (n = 90) | 57% | 43% | 38.94 | 10.93 |
| Russia | 89 | Russian (n = 89) | 53% | 46% | 40.17 | 9.79 |
| Germanic Europe | ||||||
| Austria | 84 | German (n = 84) | 56% | 44% | 42.31 | 10.12 |
| Germany | 88 | German (n = 85) | 55% | 44% | 44.10 | 11.47 |
| Netherlands | 94 | Dutch (n = 94) | 49% | 50% | 44.98 | 12.64 |
| Switzerland | 95 |
German (n = 61) French (n = 28) Italian (n = 6) |
49% | 51% | 42.72 | 11.95 |
| Latin America | ||||||
| Argentina | 92 | Latin Spanish (n = 92) | 50% | 50% | 39.73 | 8.91 |
| Brazil | 88 | (Brazilian) Portuguese (n = 88) | 55% | 45% | 41.15 | 11.50 |
| Colombia | 97 | Latin Spanish (n = 97) | 56% | 44% | 38.72 | 9.67 |
| Costa Rica | 89 | Latin Spanish (n = 89) | 54% | 42% | 40.73 | 10.37 |
| Mexico | 90 | Latin Spanish (n = 90) | 51% | 49% | 37.53 | 10.05 |
| Venezuela | 90 | Latin Spanish (n = 90) | 58% | 42% | 42.76 | 10.98 |
| Latin Europe | ||||||
| France | 93 | French (n = 93) | 46% | 54% | 42.16 | 10.28 |
| Israel | 84 | Hebrew (n = 84) | 58% | 42% | 43.35 | 13.03 |
| Italy | 81 | Italian (n = 81) | 56% | 43% | 40.49 | 9.46 |
| Portugal | 94 | Portuguese (n = 94) | 47% | 53% | 36.41 | 9.07 |
| Spain | 92 | Spanish (n = 92) | 50% | 50% | 40.49 | 9.15 |
| Middle East | ||||||
| Turkey | 84 | Turkish (n = 84) | 48% | 52% | 33.95 | 8.14 |
| Nordic Europe | ||||||
| Denmark | 98 | Danish (n = 98) | 53% | 46% | 46.67 | 10.29 |
| Finland | 90 | Finnish (n = 90) | 52% | 48% | 43.96 | 10.63 |
| Sweden | 89 | Swedish (n = 89) | 51% | 45% | 47.40 | 11.94 |
| Southern Asia | ||||||
| India | 64 | English (n = 64) | 59% | 39% | 35.17 | 10.50 |
| Indonesia | 77 |
Indonesian (n = 75) English (n = 2) |
53% | 47% | 32.94 | 7.32 |
| Malaysia | 79 |
English (n = 46) Malay (n = 33) |
53% | 46% | 35.33 | 9.34 |
| Philippines | 74 |
English (n = 73) Tagalog (n = 1) |
53% | 47% | 35.47 | 10.26 |
| Thailand | 80 | Thai (n = 80) | 43% | 56% | 32.56 | 6.90 |
| Total | 3,678 | 52% | 47% | 40.85 | 11.40 |
Statistical analyses were conducted using R 3.1.3 (R Core Team, 2015) and several additional R packages: descr (Aquino, 2014), plyr (Wickham, 2011), psych (Revelle, 2015), reshape (Wickham, 2007), miceadds (Robitzsch, 2015), car (Fox & Sanford, 2011), vegan (Oksanen et al., 2015a), stringr (Wickham, 2012), semTools (Pornprasertmanit, Miller, Schoemann, & Rosseel, 2014), compute.es (del Re, 2013), and heplots (Fox, Friendly, & Monette, 2014; Friendly, 2007, 2010).
Data Quality Checks
We took several steps to ensure data quality and to exclude careless responders (Meade & Craig, 2012). First, we ensured that no participant's ID appeared more than once. Although we configured the online survey such that participants were not able to skip any BFI item, one person had two missing values in two different personality dimensions of the honest condition—probably due to a technical error. We applied a two-way imputation to these two missing values: Limiting data to the relevant country subsample, two separate simple two-way imputations were conducted for each scale (Robitzsch, 2015; Sijtsma & van der Ark, 2003; van Ginkel, van der Ark, Sijtsma, & Vermunt, 2007). Second, only those participants who had completed the faking condition without interruption and answered a self-reported single insufficient effort item (“You will receive credit for this study no matter what. However, in your honest opinion, should we use your data in our analyses in this study?”; cf. Meade & Craig, 2012, p. 441) in the affirmative (nfaking = 6,981) were invited to the honest condition. Third, in the honest condition, the survey was completed by N = 4,658 participants, who again had completed this second survey without interruption and had answered the insufficient effort item (Meade & Craig, 2012, p. 441) in the affirmative (67% response rate).3 Fourth, using all BFI items of the honest condition, we excluded 27 multivariate outliers (Tabachnick & Fidell, 2013, p. 99) by their individual Mahalanobis distances (df = 44, p < .001); this step was conducted separately for each country. Fifth, we further excluded 337 participants who belonged to the fastest 5 per cent of responders in the faking or in the honest condition; this step was also conducted separately for each country (Fell et al., 2016; Meade & Craig, 2012). Sixth, we followed Meade and Craig's (2012, p. 452) recommendation of applying the Even-Odd Consistency as individual reliability index (only data of honest condition). After applying the Spearman-Brown split-half prophecy formula (rwithin, corrected = [2*rwithin]/[1+rwithin]), we excluded 616 participants with individual reliability coefficients of less than .30 (following Johnson, 2005).
The final sample consisted of N = 3,678 participants from 43 countries (Mcountry n = 85.53, Mdncountry n = 89, SDcountry n = 9.36).
Sample
Participants were recruited by an international market research firm, which was also responsible for providing incentives (i.e. tokens).4 Participants had to be 18–75 years old (Min = 18, Max = 74, M = 40.85, Mdn = 40, SD = 11.40) and employed (Table 1). Only people meeting these criteria were invited, and we additionally checked both criteria at the beginning of the online survey. Overall, the gender ratio was well balanced (52% male, 47% female, 1% did not indicate their gender); gender was also balanced across countries (Table 1). Of the participants, 63 per cent indicated that they had answered a self-description questionnaire as an applicant at least once (M = 3.44, Mdn = 1, SD = 19.02).5
Procedure
Please imagine you are in the following situation: You have searched for a job that matches your preferences for a long time. You nearly gave up hope of finding such a job at all. But now you have found a very attractive job ad! It is an interesting position. The job matches your abilities exactly and provides several opportunities to work independently. The job provides a high income and a broad range of career opportunities. The job features long-term job security. Imagine what it would mean to you in this situation to really get this job!
You have already cleared the first hurdles of the application process. To get the job, you are now asked to answer the following 2-part self-description questionnaire.
There was no explicit instruction to fake (Mesmer-Magnus & Viswesvaran, 2006) in order to avoid “exaggerate[ing] the effects of response distortion beyond what would usually be seen in actual motivating contexts” (Smith & Ellingson, 2002, p. 216).
Subsequently, participants answered the 44 items of the BFI (e.g. Benet-Martínez & John, 1998). To visually illustrate the scenario, we changed the logo of our university into a fictitious company's logo on the instruction and BFI pages. Having completed the personality questionnaire, participants were informed that the scenario had ended, and the fictitious company logo thus changed back to the university logo.
The survey drew to a close with general questions (e.g. manipulation checks) and the insufficient effort item (Meade & Craig, 2012). Participants received incentives irrespective of their answers to this item. Finally, all participants were thanked and directed back to the surveying company's website.
Since it was important that participants had stayed in the applicant scenario (faking condition) and that their data were suitable for data analysis, only those participants who had completed the first questionnaire without interruption and answered the insufficient effort item in the affirmative were invited to complete the second questionnaire (honest condition) by the surveying company.
Some days ago we asked you to answer a self-description questionnaire. You were supposed to imagine yourself doing that within an application situation. Now, it is your task to answer the questionnaire again, this time as honestly as possible.
At the end of the survey in the honest condition, participants were thanked and directed back to the website of the surveying company.
Translation Process
Participants from every country were able to answer both questionnaires in the official language(s) of their country (Table 1). Due to the limitations of translation-back-translation procedures (cf. Fell et al., 2016), a collaborative translation approach was established (Douglas & Craig, 2007). Items were taken from English sources and translated into German, or were written directly in English and German. The German version of the questionnaires was the origin for all translations; translators were provided with the English version in order to support their work. There were two experienced professional translators for every language version, who independently translated the questionnaires, discussed their versions, and created a final collaborative version. We were often able to provide the translator dyads with an existing version of the BFI because it has been widely used in several languages (Andrade, 2008; Benet-Martínez & John, 1998; Denissen, Geenen, van Aken, Gosling, & Potter, 2008; John, 2007; John, Donahue, & Kentle, 1991a, 1991b, 1991c; John et al., 2008; Jouhikainen, 2010; Lang, Lüdtke, & Asendorpf, 2001; Oshio & Urakawa, 2012; Rammstedt, 1997; Schmitt, Allik, McCrae, & Benet-Martínez, 2007; Steenkamp, de Jong, & Baumgartner, 2010; Tsaousis & Georgiades, 2009).
Measures
Personality Traits
The BFI comprises 44 short phrases (Table 3), which participants rate on a 5-point scale from 1 (disagree strongly) to 5 (agree strongly). The BFI is a relatively short instrument for measuring five dimensions of personality: Agreeableness, Conscientiousness, Extraversion, Neuroticism (frequently negatively recoded as Emotional Stability), and Openness. These five dimensions have often been accepted as a “general taxonomy of personality traits” (John et al., 2008, p. 116) and have been intensively studied with regard to their relationships with job performance (e.g. Barrick, Mount, & Judge, 2001). For personnel selection, Conscientiousness and Emotional Stability seem to be of special relevance (Barrick et al., 2001).
Faking
On the country level, countries’ mean faking was operationalised as their means of the five standardised differences (i.e. five personality dimensions).
Cultural Dimensions
Countries’ cultural characteristics were operationalised as GLOBE's societal practices scores (for details see GLOBE Project Team, 2006; House et al., 2004). Countries’ cultural scores are available from GLOBE Project Team (2004). The GLOBE project has established nine intercorrelated cultural dimensions (uncertainty avoidance, future orientation, power distance, institutional collectivism, humane orientation, performance orientation, in-group collectivism, gender egalitarianism, and assertiveness) by surveying 17,370 middle managers (House & Hanges, 2004). The nine scales have a mean reliability of .77; the country-level variance according to ICC(1) is .25 (Hanges & Dickson, 2004).
Results
Manipulation Checks
The external validity of this scenario-based, experimental study was ensured in several ways. Participants revealed their impressions of the applicant scenario (cf. Ingold et al., 2015; Jansen et al., 2013) and perceived attractiveness of the described open job position (Ellingson, 2012) on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). As expected, participants indicated that they “read the description of the advertised job exactly” (M = 4.27, SD = 0.80), that they “perceived the scenario as realistic” (M = 3.94, SD = 0.85), that they were “easily able to imagine myself [themselves] in the application situation” (M = 4.07, SD = 0.81), and that “filling out the self-description questionnaire, I [they] behaved like I [they] would do in a real application” (M = 4.33, SD = 0.77). Participants also indicated that the “described job was attractive” for them (M = 4.15, SD = 0.80). Finally, comparisons of the Big Five personality scores (unit-weighted mean scores) in both conditions (faking vs. honest) also confirmed that participants understood the application scenario (Table 2).
| M (SD) | t-test | |||
|---|---|---|---|---|
| Faking | Honest | t | d | |
| Agreeableness | 3.92 | 3.80 | 16.98 | 0.27 |
| (0.56) | (0.57) | |||
| Conscientiousness | 4.13 | 3.95 | 25.73 | 0.42 |
| (0.58) | (0.59) | |||
| Extraversion | 3.45 | 3.33 | 17.76 | 0.28 |
| (0.66) | (0.68) | |||
| Neuroticism | 2.31 | 2.49 | −22.29 | −0.36 |
| (0.69) | (0.70) | |||
| Openness | 3.75 | 3.66 | 13.92 | 0.21 |
| (0.58) | (0.60) | |||
| Agreeableness (9 items) | 3.92 | 3.80 | 16.98 | 0.27 |
| (0.56) | (0.57) | |||
| Conscientiousness (8 items) | 4.08 | 3.89 | 25.77 | 0.42 |
| (0.61) | (0.61) | |||
| Extraversion (5 items) | 3.18 | 3.09 | 11.02 | 0.18 |
| (0.79) | (0.80) | |||
| Emotional Stability (7 items) | 3.71 | 3.54 | 21.33 | 0.34 |
| (0.69) | (0.71) | |||
| Openness (8 items) | 3.88 | 3.78 | 14.23 | 0.22 |
| (0.62) | (0.64) | |||
-
Note: All p < .01, one-tailed. We used the effect size d for paired observations (Cohen, 1977, p. 48, formulas 2.3.5 and 2.3.6):
.
Invariance of Five-Factor Structure
Since we were interested in country-level relationships between GLOBE's cultural dimensions and faking, we first had to establish the invariance of the five-factor structure of the BFI (honest condition only). Following the suggestion of Byrne and van de Vijver (2010), we grouped the 43 countries in our sample into a smaller sample of cultural clusters in order to make it accessible for inspections for measurement invariance. Instead of adopting regional clustering (as in Schmitt et al., 2007), we relied on GLOBE's cultural country clustering (Gupta & Hanges, 2004, p. 190), which seemed more appropriate since our study relied on the GLOBE framework. Thus, each country was assigned to its cultural cluster (Table 1) for the inspection of the invariance of the five-factor structure (cf. Schmitt et al., 2007): (a) inspection of the reliability of the personality dimensions across the cultural clusters, (b) inspection of the congruence between the worldwide factor structure (without US data) and the US factor structure, and (c) inspection of the congruence between the factor structures of the cultural clusters (without US data) and the US factor structure.
Reliability of the Personality Dimensions across the Cultural Clusters
Within the honest condition, the Kaiser-Meyer-Olkin measure of .94 revealed “marvelous” (Dziuban & Shirkey, 1974, p. 359) sampling adequacy. Parallel analysis suggested a maximum of nine factors (vs. six components), scree test six factors/components. Acknowledging the intercorrelations among the Big Five dimensions (L. Chang, Connelly, & Geeza, 2012), we used promax rotation (but see also Schmitt et al., 2007). In the total sample as well in the US sample, principal factor analysis (cf. Russell, 2002) revealed a relatively clear five-factor structure7 (Table 3) with good scale reliabilities for Agreeableness (αtotal sample = .77, αUS sample = . 86), Conscientiousness (αtotal sample = .82, αUS sample = .82), Extraversion (αtotal sample = .80, αUS sample = .89), Neuroticism (αtotal sample = .82, αUS sample = .88), and Openness (αtotal sample = .82, αUS sample = .83). Revealing internal reliability across cultural clusters (cf. Schmitt et al., 2007), reliability coefficients were acceptable and never fell below .62 (Table 4).
| N = 3,678 | nUS = 91 | nnon-US = 3,587 | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| I see myself as someone who … | A | C | E | N | O | A | C | E | N | O | A | C | E | N | O |
| Agreeableness (A, 9 items) | |||||||||||||||
| 1. tends to find fault with others.R | .35 | −.03 | −.15 | −.36 | −.16 | .66 | −.10 | −.01 | −.10 | −.07 | .34 | −.02 | −.16 | −.36 | −.16 |
| 2. is helpful and unselfish with others. | .57 | .08 | .01 | .13 | .11 | .48 | .17 | −.14 | −.13 | .08 | .57 | .07 | .01 | .12 | .11 |
| 3. starts quarrels with others.R | .30 | .13 | −.20 | −.32 | −.19 | .32 | .05 | −.09 | −.14 | .18 | .30 | .13 | −.20 | −.33 | −.19 |
| 4. has a forgiving nature. | .59 | −.20 | −.02 | −.08 | .03 | .77 | −.17 | −.07 | −.11 | .08 | .59 | −.20 | −.02 | −.08 | .02 |
| 5. is generally trusting. | .58 | −.16 | −.03 | .02 | −.02 | .58 | −.30 | .05 | −.21 | .04 | .57 | −.15 | −.03 | .03 | −.02 |
| 6. can be cold and aloof.R | .40 | .09 | .28 | −.09 | −.31 | .73 | .07 | .13 | .11 | −.03 | .39 | .09 | .28 | −.11 | −.31 |
| 7. is considerate and kind to almost everyone. | .66 | .13 | −.05 | .11 | .09 | .74 | .22 | −.03 | .27 | .11 | .65 | .12 | −.05 | .10 | .09 |
| 8. is sometimes rude to others.R | .38 | .22 | −.03 | −.20 | −.17 | .53 | .15 | .14 | .15 | .05 | .37 | .22 | −.04 | −.22 | −.17 |
| 9. likes to cooperate with others. | .54 | .10 | .13 | .04 | .12 | .75 | −.24 | −.02 | −.17 | −.07 | .53 | .11 | .13 | .03 | .12 |
| Conscientiousness (C, 9 items) | |||||||||||||||
| 10. does a thorough job. | .02 | .56 | −.04 | .12 | .14 | −.29 | .53 | −.32 | −.44 | .26 | .03 | .55 | −.04 | .13 | .13 |
| 11. can be somewhat careless.R | −.10 | .62 | .04 | −.08 | −.17 | −.01 | .72 | .07 | .22 | −.17 | −.11 | .61 | .03 | −.09 | −.17 |
| 12. is a reliable worker.D | .21 | .47 | −.08 | .11 | .09 | .07 | .17 | −.08 | −.14 | −.03 | .21 | .47 | −.08 | .11 | .10 |
| 13. tends to be disorganised.R | −.08 | .68 | .01 | −.02 | −.12 | −.22 | .79 | .22 | .01 | −.08 | −.08 | .68 | .00 | −.02 | −.12 |
| 14. tends to be lazy.R | .09 | .58 | .09 | −.06 | −.01 | .04 | .48 | −.06 | −.27 | .05 | .09 | .58 | .09 | −.07 | −.02 |
| 15. perseveres until the task is finished. | .08 | .49 | −.03 | .08 | .29 | .02 | .55 | −.29 | −.19 | .25 | .08 | .48 | −.02 | .08 | .29 |
| 16. does things efficiently. | .11 | .49 | −.03 | .03 | .29 | −.07 | .63 | −.08 | −.09 | .20 | .12 | .49 | −.02 | .03 | .29 |
| 17. makes plans and follows through with them. | .05 | .38 | −.01 | .01 | .28 | .17 | .31 | −.01 | −.18 | .02 | .05 | .37 | −.01 | .01 | .29 |
| 18. is easily distracted.R | −.14 | .49 | .00 | −.34 | −.01 | .17 | .53 | −.03 | −.09 | .10 | −.16 | .49 | .00 | −.35 | .00 |
| Extraversion (E, 8 items) | |||||||||||||||
| 19. is talkative. | .10 | −.08 | .60 | .16 | .18 | −.05 | −.11 | .73 | −.03 | .03 | .11 | −.07 | .60 | .16 | .18 |
| 20. is reserved.R | −.05 | −.09 | .58 | −.12 | −.10 | .05 | −.04 | .75 | .09 | .10 | −.05 | −.09 | .58 | −.12 | −.11 |
| 21. is full of energy.D | .15 | .10 | .21 | −.13 | .41 | −.11 | −.10 | .10 | −.64 | .18 | .17 | .11 | .21 | −.12 | .40 |
| 22. generates a lot of enthusiasm.D | .22 | .03 | .25 | −.03 | .46 | .06 | .09 | .23 | −.28 | .38 | .24 | .03 | .24 | −.02 | .46 |
| 23. tends to be quiet.R | −.06 | .04 | .77 | .07 | −.07 | −.02 | .00 | .87 | .08 | .07 | −.06 | .04 | .76 | .06 | −.07 |
| 24. has an assertive personality.D | −.07 | .09 | .17 | −.08 | .48 | −.16 | .05 | .40 | −.36 | .28 | −.07 | .10 | .15 | −.07 | .49 |
| 25. is sometimes shy, inhibited.R | −.15 | .15 | .56 | −.23 | −.01 | .05 | .15 | .78 | −.02 | .00 | −.16 | .15 | .55 | −.23 | −.01 |
| 26. is outgoing, sociable. | .29 | −.05 | .57 | .03 | .21 | .08 | −.06 | .63 | −.22 | .15 | .30 | −.05 | .56 | .03 | .20 |
| Neuroticism (N, 8 items) | |||||||||||||||
| 27. is depressed, blue. | −.01 | −.05 | −.13 | .54 | −.06 | −.14 | .08 | −.08 | .74 | .26 | .00 | −.05 | −.13 | .54 | −.07 |
| 28. is relaxed, handles stress well.R | −.06 | .16 | .16 | .65 | −.29 | −.02 | .04 | .00 | .76 | −.02 | −.07 | .16 | .16 | .64 | −.29 |
| 29. can be tense. | .04 | −.03 | −.06 | .65 | .03 | −.12 | −.21 | −.18 | .38 | .24 | .04 | −.04 | −.06 | .65 | .02 |
| 30. worries a lot. | .14 | .14 | −.11 | .54 | .07 | .04 | −.08 | −.07 | .63 | .07 | .14 | .14 | −.12 | .53 | .07 |
| 31. is emotionally stable, not easily upset.R | −.07 | .05 | .11 | .57 | −.27 | −.08 | .05 | .12 | .84 | .13 | −.07 | .05 | .10 | .56 | −.28 |
| 32. can be moody.D | −.10 | −.16 | −.06 | .46 | .10 | −.39 | −.05 | −.18 | .18 | .16 | −.09 | −.17 | −.06 | .47 | .10 |
| 33. remains calm in tense situations.R | −.08 | −.07 | .15 | .49 | −.33 | −.03 | −.20 | .10 | .71 | .05 | −.09 | −.07 | .15 | .49 | −.34 |
| 34. gets nervous easily. | .14 | −.07 | −.05 | .69 | −.07 | −.05 | −.27 | −.29 | .43 | .24 | .15 | −.07 | −.04 | .69 | −.09 |
| Openness (O, 10 items) | |||||||||||||||
| 35. is original, comes up with new ideas. | −.04 | .05 | .04 | −.07 | .70 | −.05 | .13 | .25 | .14 | .71 | −.04 | .04 | .03 | −.07 | .70 |
| 36. is sophisticated in art, music, or literature. | .03 | −.10 | .00 | .02 | .52 | .22 | −.11 | −.09 | −.08 | .60 | .02 | −.10 | −.01 | .02 | .52 |
| 37. is curious about many different things. | .08 | −.04 | .04 | −.03 | .50 | .07 | −.26 | .00 | −.14 | .64 | .08 | −.04 | .05 | −.04 | .50 |
| 38. is ingenious, a deep thinker. | −.03 | .11 | −.17 | .03 | .66 | −.07 | .00 | −.11 | −.07 | .79 | −.02 | .12 | −.18 | .03 | .65 |
| 39. has an active imagination. | .04 | −.08 | −.01 | .03 | .68 | .07 | .28 | .07 | .25 | .68 | .04 | −.09 | −.01 | .03 | .68 |
| 40. is inventive. | −.07 | −.01 | .00 | −.17 | .74 | −.12 | .13 | .19 | .10 | .74 | −.07 | −.02 | .00 | −.17 | .74 |
| 41. values artistic, aesthetic experiences. | .09 | −.09 | −.06 | .02 | .61 | .20 | −.05 | −.06 | .21 | .61 | .08 | −.08 | −.06 | .02 | .61 |
| 42. prefers work that is routine.R, D | −.23 | −.01 | .15 | −.17 | .25 | −.22 | .01 | .16 | −.26 | .18 | −.23 | −.01 | .15 | −.17 | .26 |
| 43. likes to reflect, play with ideas. | .00 | .03 | −.09 | −.07 | .68 | .04 | −.09 | .07 | .10 | .78 | .00 | .03 | −.09 | −.07 | .68 |
| 44. has few artistic interests.R, D | −.09 | .01 | .05 | −.01 | .28 | .09 | .43 | .14 | .27 | −.08 | −.09 | .00 | .04 | −.02 | .29 |
- Note: Factor loadings > .30 are in bold face.
- RItem is recoded. D Item dropped from later analyses.
| Cronbach's α | |||||
|---|---|---|---|---|---|
| GLOBE Cluster | A | C | E | N | O |
| Anglo | .82 | .82 | .86 | .87 | .81 |
| Confucian Asia | .75 | .81 | .79 | .82 | .84 |
| Eastern Europe | .76 | .83 | .79 | .83 | .80 |
| Germanic Europe | .76 | .80 | .86 | .87 | .83 |
| Latin America | .68 | .75 | .66 | .72 | .77 |
| Latin Europe | .74 | .81 | .79 | .82 | .83 |
| Middle East | .73 | .82 | .88 | .66 | .85 |
| Nordic Europe | .77 | .78 | .85 | .83 | .81 |
| Southern Asia | .77 | .80 | .71 | .78 | .62 |
| Total Sample | .77 | .82 | .80 | .82 | .82 |
- Note: A = Agreeableness, C = Conscientiousness, E = Extraversion, N = Neuroticism, O = Openness.
- RItem is recoded.
Congruence between the Worldwide Factor Structure and the US Factor Structure
Within the honest condition, we inspected the congruence between the worldwide (non-US) and the US standardised loadings. Visual inspection revealed seven items with problematic loadings (Table 3): item 12 (Conscientiousness, “is a reliable worker”), item 21 (Extraversion, “is full of energy”), item 22 (Extraversion, “generates a lot of enthusiasm”), item 24 (Extraversion, “has an assertive personality”), item 32 (Neuroticism, “can be moody”), item 42 (Openness, “prefers work that is routine”), and item 44 (Openness, “has few artistic interests”). These seven items were dropped from further analysis. Following Schmitt et al. (2007), we also used the US structure as target matrix for Procrustes analysis due to the US origin of the BFI (see also McCrae, Zonderman, Costa, Bond, & Paunonen, 1996); Table 5 shows the Procrustes rotated factor solution. The two factor loading patterns were highly correlated (
= 0.21, rsymmetric Procrustes rotation = .89, p < .01);8 indices in the diagonal of the orthogonal rotation matrix (not shown here) were a minimum of .97.
| nnon-US = 3,587 | |||||
|---|---|---|---|---|---|
| I see myself as someone who … | A | C | E | N | O |
| Agreeableness (A) | |||||
| 1. tends to find fault with others.R | .26 | −.17 | −.21 | −.38 | −.31 |
| 2. is helpful and unselfish with others. | .45 | −.06 | −.11 | .04 | .01 |
| 3. starts quarrels with others.R | .22 | −.06 | −.26 | −.36 | −.36 |
| 4. has a forgiving nature. | .45 | −.32 | −.11 | −.17 | −.04 |
| 5. is generally trusting. | .42 | −.29 | −.13 | −.08 | −.08 |
| 6. can be cold and aloof.R | .33 | −.08 | .18 | −.05 | −.43 |
| 7. is considerate and kind to almost everyone. | .56 | .00 | −.17 | .05 | −.02 |
| 8. is sometimes rude to others.R | .34 | .07 | −.13 | −.16 | −.36 |
| 9. likes to cooperate with others. | .41 | .00 | .03 | −.04 | .02 |
| Conscientiousness (C) | |||||
| 10. does a thorough job. | −.15 | .38 | −.13 | −.01 | −.02 |
| 11. can be somewhat careless.R | −.20 | .43 | −.04 | −.09 | −.40 |
| 13. tends to be disorganised.R | −.21 | .50 | −.09 | −.06 | −.34 |
| 14. tends to be lazy.R | −.02 | .43 | .00 | −.09 | −.23 |
| 15. perseveres until the task is finished. | −.09 | .35 | −.10 | −.06 | .13 |
| 16. does things efficiently. | −.04 | .36 | −.10 | −.11 | .12 |
| 17. makes plans and follows through with them. | −.14 | .27 | −.07 | −.13 | .14 |
| 18. is easily distracted.R | −.26 | .34 | −.03 | −.38 | −.24 |
| Extraversion (E) | |||||
| 19. is talkative. | −.03 | −.18 | .52 | .08 | .12 |
| 20. is reserved.R | −.15 | −.24 | .54 | −.15 | −.19 |
| 23. tends to be quiet.R | −.15 | −.11 | .70 | .07 | −.16 |
| 25. is sometimes shy, inhibited.R | −.24 | .02 | .52 | −.23 | −.16 |
| 26. is outgoing, sociable. | .16 | −.13 | .48 | −.05 | .13 |
| Neuroticism (N) | |||||
| 27. is depressed, blue. | −.17 | −.24 | −.25 | .42 | −.11 |
| 28. is relaxed, handles stress well.R | −.18 | −.04 | .01 | .66 | −.38 |
| 29. can be tense. | −.16 | −.23 | −.21 | .47 | −.01 |
| 30. worries a lot. | .01 | −.01 | −.26 | .46 | −.02 |
| 31. is emotionally stable, not easily upset.R | −.18 | −.16 | −.04 | .54 | −.35 |
| 33. remains calm in tense situations.R | −.20 | −.28 | .02 | .48 | −.40 |
| 34. gets nervous easily. | −.04 | −.26 | −.21 | .56 | −.10 |
| Openness (O) | |||||
| 35. is original, comes up with new ideas. | −.18 | .04 | .03 | −.18 | .56 |
| 36. is sophisticated in art, music, or literature. | −.08 | −.13 | −.02 | −.06 | .39 |
| 37. is curious about many different things. | −.07 | −.10 | .02 | −.18 | .40 |
| 38. is ingenious, a deep thinker. | −.18 | .09 | −.20 | −.11 | .51 |
| 39. has an active imagination. | −.10 | −.10 | −.02 | −.10 | .58 |
| 40. is inventive. | −.20 | −.01 | .02 | −.29 | .60 |
| 41. values artistic, aesthetic experiences. | −.02 | −.10 | −.08 | −.07 | .48 |
| 43. likes to reflect, play with ideas. | −.15 | .02 | −.09 | −.20 | .55 |
- Note: Factor loadings > .30 are in bold face.
- RItem is recoded.
Congruence between the Factor Structures of the Cultural Clusters and the US Factor Structure
Within the honest condition, the five Tucker Indexes of factor congruence (Table 6) revealed a fair overall similarity of the factor loadings of the US and the non-US sample (Lorenzo-Seva & ten Berge, 2006). Table 6 also shows Tucker Indexes of factor congruence between the US and the cultural clusters (Gupta & Hanges, 2004). Except for Neuroticism in the Middle East cluster (i.e. Turkey), Tucker Indexes revealed sufficient amounts of congruence (Min = .70, M = .85, SD = .05; Lorenzo-Seva & ten Berge, 2006).
| GLOBE Clusters (nnon-US = 3,587) | A | C | E | N | O |
|---|---|---|---|---|---|
| Anglo | .94 | .92 | .92 | .91 | .89 |
| Confucian Asia | .80 | .84 | .83 | .70 | .87 |
| Eastern Europe | .89 | .89 | .86 | .85 | .85 |
| Germanic Europe | .83 | .90 | .91 | .91 | .88 |
| Latin America | .85 | .80 | .85 | .82 | .79 |
| Latin Europe | .90 | .88 | .88 | .90 | .84 |
| Middle East | .67 | .27 | .71 | .16 | .85 |
| Nordic Europe | .81 | .88 | .90 | .87 | .86 |
| Southern Asia | .78 | .71 | .87 | .78 | .81 |
| Overall | .94 | .90 | .91 | .90 | .86 |
- Note: A = Agreeableness, C = Conscientiousness, E = Extraversion, N = Neuroticism, O = Openness.
- RItem is recoded.
Overall, inspections revealed sufficient invariance of the BFI's five-factor structure across cultural clusters to be able to conduct our main analysis (i.e. country-level correlation analysis; cf. Byrne & van de Vijver, 2010; Schmitt et al., 2007).
Country-Level Big Five Personality Scores and Faking Scores
In order to facilitate the interpretation of faking scores in the context of applicant selection, Neuroticism was recoded into Emotional Stability (Barrick et al., 2001; John et al., 2008). Using the slightly smaller item set, we again computed the individual-level Big Five personality scores in both experimental conditions; comparisons again revealed a substantial amount of faking (Table 2).
Confirming construct validity (Table 7) of the country-level Big Five personality scores in this study (only honest condition), correlation analysis revealed that they were highly correlated (mean r = .51) with those of Schmitt et al. (2007). Country-level standardised difference scores between the two experimental conditions also revealed substantial amounts of faking: Agreeableness (M = 1.49, SD = 1.00), Conscientiousness (M = 2.38, SD = 1.00), Extraversion (M = 1.29, SD = 1.00), Emotional Stability (M = 1.61, SD = 1.00), and Openness (M = 1.71, SD = 1.00). Faking scores also differed significantly on the country level.9 Across the five personality factors, mean country-level faking (M = 1.70, SD = 0.82) also showed good scale reliability (α = .88).
| Measures | M | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| BFI scores (n = 43) | |||||||||||
| 1. Agreeableness | 3.79 | 0.23 | |||||||||
| 2. Conscientiousness | 3.88 | 0.23 | .79** | ||||||||
| 3. Extraversion | 3.08 | 0.24 | .02 | .33* | |||||||
| 4. Emotional Stabilitya | 3.53 | 0.22 | .72** | .75** | .46** | ||||||
| 5. Openness | 3.78 | 0.24 | .57** | .67** | .30+ | .67** | |||||
| BFI T-scores (n = 31) | |||||||||||
| 6. Agreeableness | 47.05 | 2.49 | .35+ | .34+ | .41* | .42* | .28 | ||||
| 7. Conscientiousness | 46.32 | 3.18 | .60** | .66** | .07 | .53** | .43* | .24 | |||
| 8. Extraversion | 48.94 | 1.68 | .53** | .73** | .38* | .63** | .46** | .35+ | .65** | ||
| 9. Neuroticism | 51.06 | 2.28 | −.36* | −.44* | −.26 | −.50** | −.25 | −.60** | −.60** | −.51** | |
| 10. Openness | 48.87 | 2.70 | .52** | .68** | .48** | .70** | .62** | .55** | .57** | .60** | −.53** |
- Note: aRecoded from Neuroticism.
- + p < .10, two-tailed; * p < .05, two-tailed; ** p < .01, two-tailed.
Culture and Faking Behavior
For testing the nine country-level hypotheses on the relationship between faking behavior and culture, we relied on bivariate correlations (following Fell et al., 2016). Surprisingly, all but one (in the case of assertiveness, H9) pointed in the opposite direction than expected or were not significant (see Table 8). It should also be noted that an alternative approach, which was suggested by one of the reviewers, confirmed our results: We conducted separate, multi-level regression analyses with individual participants’ faking behavior (N = 3,678), which was operationalised as raw difference between the two conditions (i.e. applicant score–honest score), was successively regressed on country-level cultural predictors. As displayed in Table OSI3, unstandardised coefficients of fixed effects echoed country-level correlation coefficients.
| Measures | M | SD | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Fakinga | |||||||||||||||||||
| 1. Agreeableness | 1.49 | 1.00 | |||||||||||||||||
| 2. Conscientiousness | 2.38 | 1.00 | .83** | ||||||||||||||||
| 3. Extraversion | 1.29 | 1.00 | .47** | .34* | |||||||||||||||
| 4. Emotional Stability | 1.61 | 1.00 | .86** | .85** | .48** | ||||||||||||||
| 5. Openness | 1.71 | 1.00 | .60** | .64** | .20 | .59** | |||||||||||||
| 6. Mean | 1.70 | 0.82 | .92** | .90** | .61** | .92** | .74** | ||||||||||||
| GLOBE Cultural Dimensions | |||||||||||||||||||
| 7. Uncertainty avoidance | 4.22 | 0.62 | .38* | .18 | .46** | .39** | .20 | .39** | |||||||||||
| 8. Future orientation | 3.89 | 0.49 | .31* | .14 | .33* | .28+ | .18 | .31* | .76** | ||||||||||
| 9. Power distance | 5.10 | 0.45 | −.24 | −.13 | −.21 | −.29+ | −.07 | −.23 | −.54** | −.45** | |||||||||
| 10. Institutional collectivism | 4.28 | 0.47 | .05 | −.08 | .22 | −.07 | −.07 | .01 | .41** | .52** | −.22 | ||||||||
| 11. Humane orientation | 4.04 | 0.45 | −.31* | −.35* | .15 | −.28+ | −.09 | −.21 | .09 | .19 | −.16 | .43** | |||||||
| 12. Performance orientation | 4.12 | 0.41 | .20 | .02 | .30+ | .17 | .19 | .21 | .59** | .65** | −.36* | .50** | .29+ | ||||||
| 13. In-group collectivism | 4.98 | 0.79 | −.58** | −.47** | −.28+ | −.59** | −.31* | −.55** | −.56** | −.33* | .73** | −.09 | .13 | −.15 | |||||
| 14. Gender egalitarianism | 3.44 | 0.34 | −.03 | .05 | −.06 | .05 | −.02 | .00 | −.12 | −.19 | −.23 | −.14 | −.06 | −.38* | −.14 | ||||
| 15. Assertiveness | 4.10 | 0.36 | .22 | .32* | −.14 | .27+ | .17 | .21 | −.23 | −.02 | .20 | −.45** | −.54** | −.06 | .17 | −.10 | |||
| Attitude toward Faking (n = 31)b | |||||||||||||||||||
| 16. Severe faking | 0.00 | 0.33 | −.45* | −.46** | −.01 | −.38* | −.37* | −.41* | −.20 | .01 | .35+ | .14 | .45* | .18 | .46** | −.39* | −.25 | ||
| 17. Mild faking | 0.00 | 0.31 | −.56** | −.53** | −.12 | −.53** | −.45* | −.54** | −.36* | −.22 | .56** | −.05 | .22 | .12 | .68** | −.41* | −.16 | .86** | |
| 18. Unemployment ratec | 8.16 | 5.85 | .22 | .27+ | −.02 | .10 | .17 | .18 | −.25+ | −.25 | .09 | −.38* | −.36* | −.38* | .03 | .05 | .25 | −.36* | −.23 |
- Note: Intercorrelations on country level (N = 43). GLOBE scores are matched as follows: Scores for Germany are means of former East and former West Germany scores; Scores for South Africa are means of black South African and white South African sample scores; Scores for Switzerland are means of French- and German-speaking Switzerland scores.
- aStandardised differences between faking and honest condition (Cohen, 1977, p. 48). bFactor scores for attitude toward faking in job interviews (Fell et al., 2016). cUnemployment rates (percentage of labor force) were taken from the World DataBank (2015); Taiwanese unemployment rates were taken from the Statistical System of the Republic of China (2014).
- + p < .10, two-tailed; * p < .05, two-tailed; ** p < .01, two-tailed.
The first hypothesis, which predicted lower levels of faking in high uncertainty avoidance countries (H1), had to be rejected. Instead, the results indicated significant positive relationships between uncertainty avoidance and faking on Agreeableness (r = .38), faking on Extraversion (r = .46), faking on Emotional Stability (r = .39), and mean faking (r = .39). The second hypothesis, which predicted lower levels of faking in high future orientation countries (H2), also had to be rejected. Again, the results indicated significant positive relationships for faking on Agreeableness (r = .31), faking on Extraversion (r = .33), faking on Emotional Stability (r = .28, p < .10), and mean faking (r = .31). The third hypothesis, which predicted higher levels of faking in high power distance countries (H3), was not confirmed. The results indicated an almost significant negative relationship between power distance and faking on Emotional Stability (r = −.29, p < .10). The fourth hypothesis, which predicted lower levels of faking in high institutional collectivism countries (H4), was not confirmed. All five correlations were insignificant; four were nominally close to zero. The fifth hypothesis, which predicted higher levels of faking in highly humane-oriented countries (H5), had to be rejected. The results indicated negative relationships for faking on Agreeableness (r = −.31), faking on Conscientiousness (r = −.35), and faking on Emotional Stability (r = −.28, p < .10). The sixth hypothesis, which predicted lower levels of faking in high performance orientation countries (H6), was not confirmed. The results indicated an almost positive relationship between performance orientation and faking on Extraversion (r = .30, p < .10). The seventh hypothesis, which predicted higher levels of faking in high in-group collectivism countries (H7), had to be rejected. The results indicated negative relationships between in-group collectivism and faking on Agreeableness (r = −.58), faking on Conscientiousness (r = −.47), faking on Extraversion (r = −.28, p < .10), faking on Emotional Stability (r = −.59), faking on Openness (r = −.31), and mean faking (r = −.55). The eighth hypothesis, which predicted lower levels of faking in high gender egalitarian countries (H8), was not confirmed. All five correlations were insignificant and also nominally close to zero. Finally, the ninth hypothesis, which predicted higher levels of faking in highly assertive countries (H9), was partially confirmed. The results indicated positive relationships between assertiveness and faking on Conscientiousness (r = .32) and faking on Emotional Stability (r = .27, p < .10).
Additional Analyses
Fell et al. (2016) argued that the attitude toward faking is an important precursor of actual faking behavior (cf. Ajzen, 1991; McFarland & Ryan, 2000, 2006; Mueller-Hanson et al., 2006; Snell et al., 1999; Taras, Kirkman, & Steel, 2010). Exploiting the fact that our sample includes all of the 31 countries previously examined by Fell et al. (2016), we correlated their country-level attitudes toward faking behavior in job interviews and our country-level scores of actual faking behavior in personality tests (N = 31). Following Fell et al. (2016), we expected a positive country-level relationship between faking attitudes (i.e. faking in job interviews is bad vs. good) and faking behavior (i.e. standardised difference scores between faking vs. honest condition). Again, the opposite was true (Table 8): For all personality factors but Extraversion (rsevere faking = −.01, ns, rmild faking = −.12, ns), attitudes were negatively correlated (Mr = −.47, ranger = [−.56; −.37]) with faking behavior.
Moreover, previous findings on the relationship between unemployment and faking behavior have been mixed. Several studies considered unemployment as fueling faking (e.g. Ellingson, 2012; König et al., 2011, 2012b; MacCann, Ziegler, & Roberts, 2012; Thackray et al., 2013). However, within the 31 countries examined, Fell et al. (2016) found negative correlations between countries’ unemployment rate and their attitude toward faking in job interviews. In the current study, the results were more heterogeneous (Table 8). Only the positive correlation for Conscientiousness was almost significant (r = .27, p < .10); all other correlations were insignificant and only nominally positive. On the one hand, the results could suggest, at best, that unemployment is rather positively related to actual faking behavior. On the other hand, we agree with one of the reviewers that the inconsistent results on the association between countries’ unemployment and faking could also reflect the necessity to consider more complex relationships: For example, countries’ social safety nets (cf. Debus, Probst, König, & Kleinmann, 2012) could moderate the association between unemployment and faking. Correlations might be stronger in countries where there is no social safety net available that reduces at least the worst consequences of losing during fierce job competition (e.g. lack of food, no roof over one's head).
Following a reviewer's suggestion we also found preliminary evidence (cf. Online Supporting Information) for the potentials of a second-order cultural factor that could be labeled “performance-basedness” (Stephan & Uhlaner, 2010) to explain cross-cultural differences in applicant faking.
Discussion
The main goal of this study was to examine cross-cultural differences in actual faking behavior across 43 countries. Results showed that faking was considerably related to culture—especially to uncertainty avoidance, future orientation, humane orientation, in-group collectivism, and assertiveness. Except for assertiveness, all of these relationships were consistently in the opposite direction from that hypothesised or were not significant. Furthermore, on the country level, faking behavior was negatively related to attitudes toward faking behavior, which is inconsistent with previous research on the individual level—for example, research on delinquency (Engels, Luijpers, Landsheer, & Meeus, 2004), on academic honesty (Payan, Reardon, & McCorkle, 2010), and even on faking behavior (McFarland & Ryan, 2000, 2006; Mueller-Hanson et al., 2006; Snell et al., 1999). Despite—or perhaps because of—the consistently contrary results (cf. Hollenbeck, 2008), our study calls for a detailed reflection on the interplay of culture, attitudes, and faking. First, there may be specific post-hoc explanations for the contradictory findings for the cultural dimensions uncertainty avoidance, future orientation, humane orientation, in-group collectivism, and assertiveness. Second, we discuss the explanatory power of social information processing (Salancik & Pfeffer, 1978).
Specific Explanations for Cultural Dimensions
Since uncertainty is inherent to applicant faking (Bangerter et al., 2012), we assumed that high uncertainty avoidance cultures would exhibit smaller amounts of faking. However, we found positive country-level correlations between uncertainty avoidance and faking. A positive relationship was also previously found in the case of corruption (Husted, 1999). Husted (2000) described this contradiction between high uncertainty avoidance cultures’ strong adherence to rules and the rare manifestation of these rules as ironic. Following Husted (1999), one could argue that applicant faking reduces uncertainty because it makes desired results more certain. In other words, faking increases the chances of being hired (i.e. reduces employment uncertainty) and thereby reduces applicants’ socioeconomic uncertainties.
Since deciding to engage in faking was assumed to be a decision in favor of short-term over long-term consequences (Mead et al., 2009), we assumed that high future orientation cultures would exhibit smaller amounts of faking. However, we found positive country-level correlations between future orientation and faking. As an explanation for this finding, future orientation might encourage people to focus on long-term consequences, but does not necessarily prevent them from optimistic forecasting of these long-term consequences (cf. Ashkanasy, Gupta, Mayfield, & Trevor-Roberts, 2004). On the individual level, future orientation is, in fact, accompanied by optimism (E.C. Chang et al., 2013). Thus, in the face of possible but unlikely negative consequences of faking (Robie et al., 2006; Roulin, Bangerter, & Levashina, 2014, 2015), optimistic beliefs (e.g. “it will never happen to me”; Hemenway, 2013, p. 290) might constitute such optimistic obstacles to refraining from faking. As one reviewer pointed out, applicants in cultures of high future orientation could optimistically believe that they will soon acquire certain desired characteristics. Indeed, recent research findings that volitional personality changes are possible (Hudson & Chris, 2015) suggest that such optimistic beliefs might be justified.
In the case of attitudes toward faking, Fell et al. (2016) were torn between the emphasis on fairness in highly humane-oriented cultures, which would rather contradict faking, and the simultaneous emphasis on leniency (i.e. tolerance of mistakes), which would rather facilitate faking. Relying on their finding of a positive correlation between positive attitudes toward faking and humane orientation, we believed that the forgiving nature of highly humane-oriented cultures would come to the fore in the case of applicant faking. However, the results of this study might rather be considered as evidence of the importance of aspects which preserve fairness in such cultures (see also Bond et al., 2004; Schlösser et al., 2013). The positive country-level correlation between faking and humane orientation also matches a more institutional (i.e. social care) manifestation of humane orientation rather than being a premise of interpersonal relationships (Brodbeck, Frese, Javidan, & Kroll, 2002).
Based on the conception of in-group collectivism as a cultural emphasis on being obliged to one's own in-group (i.e. family and organisation; Gelfand et al., 2004; House & Javidan, 2004, p. 12), and on the fact that applicants do not yet belong to the employer to which they are applying, we assumed a positive country-level relationship between in-group collectivism and faking. However, the relationship was negative. This result might be explained with reference to cultural differences (e.g. Mueller, Hattrup, Spiess, & Lin-Hi, 2012; Schyns, Paul, Mohr, & Blank, 2005) in applicants’ anticipated psychological contracts (e.g. Griffith, Lee, Peterson, & Zickar, 2011). Psychological contracts are “unwritten commitments made between workers and their employers” (Rousseau, Ridolfi, & Hater, 1996, p. 50). Although applicants do not yet have an actual psychological work contract, it can be assumed that they hold anticipated psychological contracts (Griffith et al., 2011), which presumably influence whether or not they engage in faking behavior (Griffith et al., 2011). In high in-group collectivism cultures, where anticipated psychological contracts (Gelfand et al., 2004; Gelfand, Erez, & Aycan, 2007; D.C. Thomas, Au, & Ravlin, 2003) are rather of a relational nature (i.e. exchange of workers’ loyalty for job security; Rousseau, 1990), applicants may be more inclined to decide against the potential disadvantages of faking and in favor of their potential future occupational in-group.
Our results partially confirm our hypothesis of a positive country-level relationship between assertiveness and faking. In assertive cultures, applicants might indeed engage more strongly in faking because dominant, competitive, aggressive, and opportunistic behavior is socially valued and expected (Den Hartog, 2004). Moreover, Den Hartog (2004) links assertiveness with Deal and Kennedy's organisational “tough-guy macho culture” (1982, as cited in Den Hartog, 2004, p. 425), which is basically a type of “survival of the fittest” culture, where risks are taken, fights are fought, and employees are either stars or losers. In such environments, faking is probably not strongly disapproved of (cf. Jacobson, Hood, & Buren, 2014). In the case of assertiveness, the results also confirm the positive country-level relationship between positive attitudes toward applicant faking in job interviews and actual faking behavior in personality tests.
However, such culturally specific post-hoc explanations for contradictory findings might be criticised for arbitrariness and for ignoring the negative relationships between attitudes toward faking and actual faking behavior. Thus, we will also discuss the exploratory power of the concept of social information processing, which is a more general approach that also seems less prone to such limitations.
A Social Information Processing Explanation
Social information processing (Salancik & Pfeffer, 1978) proposes that “individuals, as adaptive organisms, adapt attitudes, behavior, and beliefs to their social context and to the reality of their own past and present behavior and situation” (Salancik & Pfeffer, 1978, p. 226). What we previously considered as contradictory country-level relationships might be explained by core ideas of social information processing (Salancik & Pfeffer, 1978). According to this conception, attitudes toward faking behavior are not only culturally influenced and themselves precursors of faking behavior. Rather, attitudes are also influenced by the environment, which is itself “created through individual and social processes” (Salancik & Pfeffer, 1978, p. 228), by social information about what attitudes are appropriate, and by attributions of past behavior. Assuming that a very high country-level faking score means that faking is very prevalent in a country, faking might accordingly be widely disapproved of (i.e. negative attitudes toward faking) because the negative consequences of faking behavior would be very apparent as well (cf. Fell et al., 2016). Despite the social disapproval of faking, applicants in selection or experimental selection research scenarios attributionally realise that a high prevalence of faking implies a certain necessity to engage in faking themselves—if only to overcome potential disadvantages. The environment which applicants perceive and process (Salancik & Pfeffer, 1978) is the background against which attitudes and behavior interact, but it is also shaped by the behavior. If collectively experienced and generationally processed, such an environment becomes culture (cf. Heine, 2012; House & Javidan, 2004). Thus, instead of being considered as contradictory, the relationships between culture, attitudes toward faking, and faking behavior that have been examined so far should be considered as probably more complex than previously thought.
Limitations
Our study is, however, not free of limitations. First, our operationalisation of faking might attract some criticism. The decision regarding which operationalisation to use depends on the particular study in question, because every method has specific limitations (Burns & Christiansen, 2011). For our study, faking behavior was measured as the standardised difference between the two experimental conditions. On the one hand, difference scores have been accused of constraining variance (but see Burns & Christiansen, 2011) and of bringing in autocorrelations (Burns & Christiansen, 2011). On the other hand, however, standardised faking difference scores are common, intuitive, and easy to interpret (Burns & Christiansen, 2011; but see also Kisbu-Sakarya, MacKinnon, & Aiken, 2013; D.R. Thomas & Zumbo, 2012).
Second, including 10 potential country-level predictors of faking behavior (nine cultural characteristics and unemployment) prevented us from applying more sophisticated data analysis, such as multi-level analysis (Tabachnick & Fidell, 2013; see also Fell et al., 2016; van Hemert, van de Vijver, & Poortinga, 2008). This would have been methodologically preferable (e.g. Hox, 2010) had there been fewer country-level predictors or a larger country sample size. Working within the GLOBE framework and not excluding single hypotheses (i.e. cultural dimensions), we instead had a predictor:country ratio of 10:43 for our 43-country study, which is too low for multi-level approaches. However, based on the often unexpected results of our study, we believe that limiting our study a priori to a smaller number of predictors, for example by only relying on the significant results of Fell et al. (2016), would have dramatically limited the potential knowledge gain.
Third, although this study clearly goes beyond previous research in this area, which often failed to measure actual faking behavior or only included a small number of countries, there is, of course, still additional room for improvement: We could have included even larger numbers of participants (or countries) in certain countries (or cultural clusters). However, part of this room for improvement is due to experimental design (e.g. repeated measures, availability of cultural predictors) and methodological rigor (e.g. completing the faking questionnaire without interruption, several post-hoc data quality criteria).
Fourth, the generalisability of our findings could be questioned since they are based on experimental online surveys among employees but not on actual applicants. Indeed, a meta-analysis by Hooper (2007) revealed that laboratory research overestimates faking effects unless the researchers take methodological precautions. Accordingly, to prevent an overestimation of faking, we realised three study design aspects in our study: (a) We applied a within-subjects design in which the faking (i.e. applicant) condition was followed by the honest condition, which has been shown to yield the most conservative faking effect sizes; (b) In the faking condition, participants were not directly asked to inflate their scores but to merely imagine themselves as being applicants. Although direct instructions have not been shown to significantly overestimate faking in lab studies, we agree with Hooper (2007, p. 41) that indirect instructions “create a setting more similar to that of the field in which participants are not instructed to self-present, but rather choose whether to do so”; (c) As one reviewer pointed out, describing the open position of the faking condition as especially suitable might have produced rather conservative estimates of faking because applicants might not have considered the need to fake to be particularly great. This conservative feature of our design, however, could also have strengthened the validity of our study.
Fifth, it may be objected that inspections of measurement invariance should have been examined across the 43 countries instead of the nine cultural clusters. However, conducting cross-cultural research based on broad country samples requires methodological alternatives to inspecting invariance across dozens of countries, as “testing for equivalence of a measuring instrument in large-scale cross-cultural studies can be fraught with difficulties” (Byrne & van de Vijver, 2010, p. 128). Relying on meaningful groups of countries (i.e. GLOBE clusters) constitutes one such methodological alternative (Byrne & van de Vijver, 2010; cf. Schmitt et al., 2007).
Sixth, it should be kept in mind that interpreting individual correlation coefficients raises the question of alpha error inflation. However, applying Bonferroni correction would have worsened the small power problem of country-level correlation analyses.
Future Research
This research illustrates the systematic and complex nature of cross-cultural differences in applicant faking behavior. In the hope that our study will pave the way for more to come, we outline some ideas for future research in more detail.
First, future research should examine culture, attitudes toward faking, and actual faking behavior altogether in one study in order to gain a deeper understanding of their interplay. With regard to cross-cultural differences, researchers should consider collecting cultural information themselves, because such primary cultural data have been shown to increase the chances of finding reliable cultural relationships for attitudes as well as behavior (Taras et al., 2010) on the country level. The availability of cultural predictor and behavioral dependent variables on both the individual and the country level would allow more complex analysis approaches to be applied (e.g. Cheung & Au, 2005) and further methodological idiosyncrasies to be inspected (e.g. Fontaine & Fischer, 2011; Wagner, 1982).
Second, future studies may consider examining the role of recruiters. Schmid Mast, Frauendorfer, and Popovic (2011) showed that recruiters’ cultural background (French-speaking Swiss vs. French-speaking Canadian) might moderate the relationship between applicants’ self-promoting behavior and their chances of being hired (see also Bye, Horverak, Sandal, Sam, & van de Vijver, 2014). Taken together with the results of this study, researchers could explore whether there are cross-cultural differences in recruiters’ attitudes toward applicant faking or in recruiters’ own faking behavior (e.g. exaggerating the degree of freedom in a position).
Third, future studies may encompass a broader view on faking and self-presentation. Is faking a behavioral pattern that is stable across contexts? By including faking behavior beyond personnel selection, for example, faking in the areas of dating (e.g. Ellison, Heino, & Gibbs, 2006; Gibbs, Ellison, & Heino, 2006), higher education (e.g. Payan et al., 2010), or school (e.g. A.D. Evans & Lee, 2014), future studies could provide a deeper understanding of the concept of faking and of the generalisability of results on behavior in particular contexts.
Fourth, we agree with previous authors (e.g. Burns & Christiansen, 2011; Frei et al., 2006) that faking is a complex phenomenon. Therefore, we hope that future research examines individuals’ cognitions during high-stakes test situations (e.g. personality tests) across countries in more detail. Post-hoc interviews (cf. König, Merz, & Trauffer, 2012a) or think-aloud procedures (cf. Robie et al., 2007) could provide insights into the process (e.g. relevant behavior, justifications, strategies) that leads to faking behavior and, in the end, to cross-cultural differences.
Implications
There are several theoretical and practical implications that can be derived from this study. Our study showed that cultural characteristics as well as attitudes toward faking are important precursors of faking behavior—but with unexpected results. In particular, the study questions previous models on the emergence of applicant faking, which considered a positive attitude toward faking to be a clear precursor of faking behavior (e.g. McFarland & Ryan, 2000). However, McFarland and Ryan's (2006) results have already hinted at this possible complexity: In their study, correlations on the individual level between attitude toward faking and faking personality scales were surprisingly modest or insignificant.
Our study exemplifies the importance of exercising caution when deriving implications for organisations (Bartunek & Rynes, 2010). We can be fairly sure that applicant faking behavior exists, that it exists across cultures, and that practitioners should take systematic cross-cultural differences seriously. In this realm, our study perhaps confirms what previous studies have shown step by step, country by country. Beyond increasing practitioners’ awareness (Bartunek & Rynes, 2010), our study shows how dangerous (e.g. due to potential lawsuits) it would be to rely on the scarce research conducted so far and to directly account for assumed cross-cultural differences in applicant faking. Until more research has been done, we advise organisations to be very cautious when attempting to deal with cross-cultural differences in applicant faking. Higher target values for applicants who belong to a certain cultural group could minimise discrimination, but it could increase or even cause discrimination just as well. However, there are alternatives (cf. Fell et al., 2016) that might also minimise discriminatory effects of cross-cultural differences in applicant faking, but which have side benefits rather than dangerous side effects. These include structured rather than unstructured job interviews, culturally heterogeneous rather than homogenous interviewer boards, and providing interviewers with cultural training. Such alternatives would surely be preferable to ignoring the international nature of the modern-day work environment.
Conclusion
In a globalised world, I-O psychology is supposed to support organisations. Being based on a relatively broad sample of 43 countries, examining actual faking behavior, and applying a conservative within-subjects design (e.g. faking condition before honest condition; large time-lag between conditions), this study makes an important contribution to explaining cross-cultural differences in applicant faking behavior. Although the results of our study may initially appear to be unexpected, they clearly support the idea of systematic cross-cultural differences in actual faking behavior.
Footnotes
, and also prints the sum of squares of the symmetric analysis, sometimes called
” (Oksanen et al., 2015b, p. 201).
