Abstract
Behavioral integrity (BI)—a perception that a person acts in ways that are consistent with their words—has been shown to have an impact on many areas of work life. However, there have been few studies of BI in Eastern cultural contexts. Differences in communication style and the nature of hierarchical relationships suggest that spoken commitments are interpreted differently in the East and the West. We performed three scenario-based experiments that look at response to word–deed inconsistency in different cultures. The experiments show that Indians, Koreans, and Taiwanese do not as readily revise BI downward following a broken promise as do Americans (Study 1), that the U.S.–Indian difference is especially pronounced when the speaker is a boss rather than a subordinate (Study 2), and that people exposed to both cultures adjust perceptions of BI based on the cultural context of where the speaking occurs (Study 3).
Do managers act in ways that are consistent with what they say? When managers are seen to be consistent, they are thought to have high “behavioral integrity” or “BI.” This concept was first introduced by Simons (1999, 2002), and the construct has since been validated and shown to be conceptually distinct from related constructs such as moral integrity, trust, psychological contracts, and justice (Simons, 2002), and these distinctions have been supported empirically (Simons, Friedman, Liu, & McLean-Parks, 2007; Simons, McLean Parks, & Tomlinson, 2014). Moreover, its impact on subordinates’ attitudes and performance has been documented in 21 studies with 35 independent samples (literature reviewed in Simons, Tomlinson, & Leroy, 2011; meta-analyzed in Simons, Leroy, Collewaert, & Masschelein, 2015). Although the importance of BI is clear, there has been little research assessing what makes some people “especially sensitive or insensitive to leaders’ behavioral inconsistencies” (Simons et al., 2011, p. 338). Because BI is a perception, it is important for managers to understand what the typical impact of word–deed inconsistencies are on subordinates. Moreover, managers need to know which audiences might be more or less reactive to these inconsistencies.
The only research to our knowledge that addresses this question empirically is a study of Black versus White perceptions of word–deed inconsistencies within the United States (Simons et al., 2007). The argument made in Simons et al. (2007) is that Americans of different races observe others through different historical lenses, such that Blacks are more likely than Whites to notice word–deed inconsistencies. In this article, we similarly address the question of social context, but widen the focus to look at cross-national differences in reactions to word–deed consistency. Because BI is a perception, and national culture and context have been shown to shape perceptions (e.g., Chiu, Morris, Hong, & Menon, 2000; Morris & Peng, 1994), we expect national culture and context to impact reactions to work–deed inconsistency.
The aspect of national cultures that we focus on is the role of words and language. We focus on this aspect of culture because there is a rich literature on cultural differences in approaches to speech and language, and because the core of BI is an evaluation of words that are said. If spoken words are interpreted differently across cultures, then we expect that the degree to which people make judgments about a speaker based on those words will also be different. Differences in use and interpretation of words are especially visible in Eastern versus Western contexts. Therefore, in this article, we explore research that informs East–West differences in interpretation of words, and hypothesize that word–deed inconsistency will affect perceptions of BI in the West more strongly than in the East. We also examine the role of hierarchy, suggesting that East–West differences in reactions to word–deed inconsistency will be especially strong when the speaker is a boss, rather than a subordinate. Thus, when looking at the role of words in leadership, there are different nuances in the East and West—expectations drawn from one cultural context may not apply to the other cultural context.
These ideas are tested using two scenario studies with subjects in India, Taiwan, Korea, and the United States. These scenario studies are the first to experimentally manipulate word–deed inconsistency, and compare in a carefully controlled way differences in reactions with word–deed inconsistency. 1 This method allowed us to examine how Eastern and Western respondents form BI and trust judgments of superiors and subordinates given identical events. Thus, Studies 1 and 2 look at how subjects, within their national cultural context (the United States for Americans, Indian for Indians, Taiwan for Taiwanese, and Korea for Koreans) judge the behaviors of others.
Study 3 takes a very different, but related, approach. It asks whether subjects who are steeped in both Eastern and Western culture anticipate the kind of East–West differences in behavior that we find in Studies 1 and 2. That is, do they hold “implicit theories” (Hong, 2009) about cultural differences in the value of word–deed consistency that match the theories presented in this article? We do this by asking subjects from Singapore (which—like Hong Kong—is heavily influenced by both Western and Eastern cultural systems; Bishop, 1998; Tavassoli & Lee, 2003; Wharton, 2000) to report how they think Indians in India will react to word–deed inconsistency, and how they think Americans in the United States will react to word–deed inconsistency. The purpose here is not to show that American and Indians make different perceptions of BI (which was our goal in Studies 1 and 2), but to test whether people who are culturally informed have a mental model of East–West cultural systems that includes expectations of different reactions to word–deed inconsistency. This is important because this bicultural perspective is structurally similar to that of experienced expat managers, such as Americans managing in India, or Indians managing in the United States. Those expat managers typically need to operate effectively in two cultural worlds—that of their home company and that of the employees they manage in the foreign country. If they truly gain deep global experience and awareness, they should themselves become bicultural, and thus, able to shift their expectations about the impact of word–deed consistency to the specific cultural context—either East or West. Figure 1 shows the baseline model, and the structure of the three studies.
Visual summary of effects.
This work offers two contributions to theory and practice. First, it helps clarify how the behaviors and/or perceptions that drive interpersonal trust may differ across cultures—an issue of vital importance for international managers and scholars of trust and cultural difference. Second, it expands our understanding of perceiver effects in BI—an as-yet under-explored area for research (Simons, Tomlinson, & Leroy, 2011).
Behavioral Integrity
BI is a perception that a person acts in ways that are consistent with what he or she says. This concept was developed by Simons (1999, 2002), because related constructs that existed at the time missed the simple idea that people care deeply about how reliable someone’s word is. Trust, at least in some forms (e.g., Cummings & Bromiley, 1995), implies a belief that the other would be benevolent and caring. Moral integrity (Mayer, Davis, & Schoorman, 1995) implies that the other will act in ways consistent with the observers’ morals. The theory of BI says that, although benevolence and similarity of values may be good, people also judge leaders in a much simpler way—do those leaders do what they say, even if what they say is not necessarily caring or moral? They make this judgment because it is important that the person be reliable, which provides certainty. People may hope that a leader shares one’s morals, but it is also important that a leader simply be reliable.
Another construct that is seemingly related to BI is the psychological contract (Rousseau & McLean Parks, 1993), where people react to whether employment-related expectations are met. The key difference is that psych contract breaches are typically event focused, centering on critical terms of employment, and words need not have been spoken to set those contractual expectations. BI is, by contrast, a perception of the speaker’s character, based on what was said and done, and people construct their judgments of speakers not just based on statements about important employment issues, but in response to commitments both proximal and distal, important and trivial (Simons, 2002). In sum, BI is about the simple, but powerful idea that people are judged on whether they act in ways that are consistent with what they say, as sometimes distinct from whether they act morally, are benevolent, or keep to promises specifically about major issues. Although perceptions of BI are related to trust, and some definitions of trust consider BI as a sub-dimension (e.g., Cummings & Bromiley, 1995), the most widely accepted definition of trust, as a willingness to accept vulnerability to another party (Mayer et al., 1995), renders it distinct. This distinction was confirmed through Confirmatory Factor Analysis (CFA) in Simons et al. (2007) for individual-level BI and through Partial Least Squares (PLS) analysis in Palanski and Yammarino (2011) and Palanski, Kahai, and Yammarino (2011) for both individual- and group-level perceptions of BI.
Having established that BI is distinct from trust and psychological contracts conceptually and empirically, we now turn to measuring perceived BI. The perception of consistency has most often been measured using Simons’ scale (Simons et al., 2007). This scale includes two dimensions (alignment between enacted and espoused values, and follow through on promises), but those two dimensions are highly intercorrelated (r = .94; Simons et al., 2007). The reason is that someone observed not acting consistent with their words (as expressed through value statements) might have their words (expressed as promises) also doubted, and vice versa. Word–deed inconsistencies are likely to lead to global judgments of the person’s BI. Another BI scale (Dineen, Lewicki, & Tomlinson, 2006) adds the dimension of enacting rules consistent with what is espoused, but this scale is also highly correlated with the Simons’ scale, for the same reasons just mentioned—observed word–deed inconsistency in any one area is likely to lead to global judgments of a person’s BI. Despite some differences in these measures, and the existence of some sub-dimensions, they all provide valid, stable measures of perceived global, trait-level BI. This contrasts with measures, such as the one developed by Leroy et al. (2012), that focus on BI applied to specific domains (such as BI about safety values espoused by nurses). Both conceptually and methodologically, we treat BI as a global assessment of a person’s trait, without focusing on specific sub-dimensions or domains. This is consistent with most work on BI, which uses the global BI measure (including promissory BI and values BI) rather than searching for differential effects on each dimension.
The impact of perceptions of a person’s BI has been studied extensively, and can now be treated as fairly well established. Perceptions of perceived BI of managers have been associated with trust in managers, absenteeism, organizational commitment, employee retention, customer service, and company profitability (Dineen et al., 2006; Greenbaum, Mawritz, & Piccolo, 2015; Kannan-Narasimhan & Lawrence, 2012; Palanski & Vogelgesang, 2011; Palanski & Yammarino, 2011; Simons, 2008; Simons, Tomlinson & Leroy, 2011). The reason why BI perceptions have such strong effects is that perceived word–action alignment can be expected to enhance trust, as it reduces uncertainty, and sinister attributions are often made for broken promises (Simons, 2002).
A recent meta-analysis (Simons et al., 2015) shows that the most strong and proximal effect of perception of a leader’s perceived BI is follower trust (r = .72), which produces secondary effects of enhanced follower organizational commitment, follower performance, and follower Organizational Citizenship Behavior (OCB). Trust can be defined as “a psychological state comprising the intention to accept vulnerability based on positive expectations of the intentions or behavior of another” (Rousseau, Sitkin, Burt, & Camerer, 1998, p. 395). The idea that perceived BI would affect trust is consistent with Dirks and Ferrin’s (2002) “character-based” perspective of trust, which suggests that followers’ judgment of a leader’s character (as opposed to their relationship) impacts how vulnerable that follower will allow themselves to be. To verify that our results are consistent with past studies, we test for the effects of BI on follower trust. We look at this particular outcome variable because it is the most proximal effect of perceived BI (Simons et al., 2015) and because, in an experimental study, it is very difficult to produce a plausible measure of OCB, performance, or organizational commitment. To summarize predictions from prior work on BI, we state our baseline hypotheses:
Although we will test for effects of perceived BI on trust as a validity check, our main contribution will be to understand culturally driven differences in the effect of word–deed consistency on perceived BI, which is both new in the literature and can be reliably measured in a scenario experimental design. Or, as Dirks and Ferrin (2002) would put it in the trust literature, we will be looking at one “follower attribute” (cultural context) and how that affects perceived BI, and—indirectly—trust.
Given that BI is a perception, little attention has been paid to actual word–deed consistency (with the exception of Cording, Simons, & Smith, 2009). This has likely occurred for two reasons: First, BI’s impact is thought to occur through perception, and BI is by definition a perceived attribute rather than an “objective” one. Second, in any field context, it is extremely challenging to objectively measure global word–action consistency, as such a measurement would require sampling from the entirety of a manager’s words and actions. However, we know from research on perceptions and biases (Fiske & Taylor, 1991), that even when facing the same target, people vary in their perceptions of that target. Thus, although there is an “objective” alignment between a person’s words and deeds, this pattern of alignment is then sampled, perceived, and interpreted by one or more observers, which introduces a set of social and cognitively based biasing factors. It is only this subjectively perceived and understood pattern of alignment that constitutes BI; actual word–deed alignment as an antecedent may more or less strongly be reflected in BI perceptions. The present study is one of a very few that, because of a vignette methodology, holds “actual” behavior constant 2 to allow empirical study across groups of people of how strongly BI perceptions respond to actual word–deed consistency.
Although perceived BI is expected to be a universally relevant construct (indeed, as the GLOBE study points out, “integrity” is considered a universal facilitator of leadership effectiveness; Dorfman, Hanges, & Brodbeck, 2003, p. 677), there is evidence that some populations may react differently to word–deed inconsistency than others. Simons et al. (2007) found that Black employees tend to notice word–action inconsistencies more than non-Black employees, so that inconsistent manager behavior undermines Black employees’ BI perceptions even more than it does non-Black employees’ BI perceptions. The reason why this occurs, they argue, is due to historic exposure in the Black community to noticeable acts of word–deed inconsistency, such as the Tuskegee experiments where the U.S. government told Black men one thing about receiving medical treatment, but in actuality did not provide real medicine (Jones, 1993). These events can lead to perceptual priming, so that managerial inconsistencies are noticed more clearly by Blacks than Whites (Fiske & Taylor, 1991).
As we look at the role of perceived BI more globally, we need to consider cultural factors that might affect reactions to actual word–deed inconsistency. The next section looks at cultural differences in the role of language and speech, because that is central to the construct of BI. In particular, we will argue, Eastern employees may be less reactive to word–deed inconsistency than Western employees.
Word–Deed Consistency and Eastern Cultures 3
Implicit in the idea of evaluating someone’s word–deed consistency are the ideas that (a) words spoken about future actions are understood as explicit signals of what actions will be taken in the future, and (b) actions taken are under the control of the speaker. These two elements of judging a speaker/actor may not exist in the same way in Eastern and Western cultures. First, below, we will discuss literature showing that language in some cultures is understood literally, whereas in other cultures that is not the case.
Literal Meaning Versus Context
Hall (1976), in his seminal work on culture and language, coined the term “High Context” (HC) and “Low Context” (LC) communication. For HC communications “most of the information is either in the physical context or internalized in the person” (p. 79) and there is very little information in the “coded, explicit, transmitted part of the message” (p. 79). As Triandis (1994) put it, “words are not taken at face value” (p. 185). By contrast, in LC communication, most of the information is in the actual transmitted message, and people watch and evaluate what is said very carefully. According to Triandis (1996), in LC cultures “people distrust what is not said clearly” (p. 184). They look for “explicit logic, proofs, linear organization of the argument” and emphasize “what is said” so that “precision in word usage” is highly valued.
On the continuum from HC to LC, Hall argued, Americans are on the low end, whereas Chinese are on the high end. The underlying reason that Americans are LC is that “Western thinking” is founded on Greek philosophers (Socrates, Plato, and Aristotle) who emphasize logic and linear thinking (this same point is made by Nisbett, Peng, Choi, & Norenzayan, 2001). In Hall’s view, LC thinking is a “shortcoming” of Western culture, because it misses many key elements of communication.
Triandis (1994) and Gudykunst (1993) tie HC and LC communications to cultural differences in individualism/collectivism (which was confirmed by Holtgraves’s [1997] study of conversational indirectness). The reason is that collectivists care more than individualists about their relationship with other people, so they avoid direct verbal expressions that might harm those relationships; there is concern in such cultures that direct disagreements might cause the other party to lose “face” (Ting-Toomey, 1988). To communicate effectively without explicit verbal statements, people in collectivist cultures know to watch carefully for subtle clues about meaning (such as emotions, body language, voice level, and what is not said). Individualism–collectivism does not line up completely with East versus West, but all Eastern countries are above the mid-point on collectivism, and all Anglo, Germanic European, and Nordic European countries are below the mid-point on collectivism (Hofstede, 1980). Sanchez-Burks (2002) takes a different angle, suggesting that Americans ignore relational cues, not due to their greater individualism, or Greek-influenced focus on logic, but due to Protestant Relational Ideology (PRI), which gives work a religious significance and emphasizes “self-reliance” and “limited personal indulgence” (p. 920) in the workplace. PRI theory, though, still identifies the same parts of the “West” as being less focused on relations: Anglo, Germanic European, and Nordic European countries, which are all heavily Protestant.
Looking at the East, Yum (1988) also sees religion as the basis for the high focus on relationships in most of Asia. Confucianism, according to Yum, emphasizes the “five basic human relationships: loyalty between kind and subject, closeness between father and son, distinction in duty between husband and wife, obedience to orders between elders and youngers, and mutual faith between friends” (p. 376). These religious ideas drive East Asians toward particularism, long-term asymmetrical reciprocity, and in–out group distinctions, which in turn lead to process-oriented (rather than outcome-oriented) communications. As a result, meaning is in the interpretation, not the message, and communication is not simply the transference of messages.
Although these theories do differ somewhat in explanation for cultural differences in communication (Greek logic, individualism–collectivism, Protestant Relational Ideology, and Confucianism), they all tell a consistent story about East–West differences. A pattern of indirectness has been shown in China, Korea, Thailand, Japan, Singapore, Taiwan, and India, and contrasted to greater directness in the United States, Australia, Canada, Sweden, Germany, and the Netherlands (Merkin, Taras, & Steel, 2014; Sanchez-Burks et al., 2003).
One example of how words are understood more in context than based on literal meaning comes from East Asia. Han (2011) explains that people in Taiwan are likely to be modest in their verbal statements of personal accomplishments, but this does not mean that they believe they have accomplished little. Rather, a modest statement is made in the hope that the other party “would not just passively accept what was said” (p. 260) but rather would counter that the speaker actually did accomplish a great deal. The initial statement of modesty is not meant as a statement of the speaker’s modest beliefs about themselves, but as the start of an interaction script that is socially known and expected. Another example comes from Indonesia, which is not an East Asian (or Confucian) country, but still has the same focus on indirect communication. Triandis (1994) relates the following Indonesian parable (told by Hofstede):
A man has two sons. He went to the first and said: “Son, go and work in the vineyard today.” The son replied, “I will go sir,’ but he did not go. The man went to the second son and said the same thing, and the son replied “I will not go,” but later changed his mind and did go. Which of the two sons did the bidding of the father? The collectivist missionary argued that it was the first, because he did not contradict his father. (p. 185)
As Triandis put it “understanding attained without words is more precious than that attained through precise articulation” (p. 184). In Eastern cultures, how something is said is more important than what is said.
Given the degree to which communication is context dependent in the East, rather than linear and direct, in the East, people are less likely to focus so strongly on the literal words, and be less judgmental toward a speaker whose words are not consistent with their actions. Whereas words are understood in the West (or, at least, in Anglo, Germanic European, and Nordic European countries) to fully and accurately convey meaning, this expectation is not held in Eastern cultural contexts. Whereas people in Western cultures focus heavily on specific words stated, those in Eastern cultures are more likely to see words as just one piece of information among many. As Nisbett et al. (2001) explain, Eastern patterns of communication are more holistic, and as such there is more acceptance of inconsistency and even contradiction: “East Asians have always lived in a complex world in which many relevant factors are important” (p. 296).
Individual Agency
To make a BI perception about a person based on word–deed consistency, it is necessary to not only assess the extent to which words and actions do or do not align but also believe that the relevant words and actions are under the control of the speaker. However, presumption of agency may also differ between Eastern and Western observers. Menon, Morris, Chiu, and Hong (1999) argued that people differ in their implicit theories of agency—that is, who or what controls behavior in situations. In the West, it is the individual that has agency, as the group is merely the context within which individuals act. In the East, it is the group that has agency. Menon et al. (1999) showed that, when describing several well-known business scandals, American newspapers referred more to the individual involved in each scandal, whereas Japanese newspapers referred more to the institution, implying a focus on the group rather than the individual as agent for Japanese. As Nisbett et al. (2001) put it, American individuals are more likely to maintain an “illusion of control” (p. 296). Similarly, Friedman, Liu, Chi, and Chen (2007) found that Chinese arbitrators were more likely to attribute control by a group actor (a company) than American arbitrators, and thus punish corporate wrongdoing more strongly; agency is more likely to exist in a group than an individual. Other examples come from South as well as East Asia. Miller (1984) found that Indians used situational attributions more than Americans, whereas Americans gave dispositional attributions more than Indians. Like East Asians, Hindu Indians (compared with Americans) are more likely to use contextual descriptions when describing acquaintances (Shweder & Bourne, 1982), such as “he is Anand’s son” or “she is from Mumbai,” rather than personal descriptions, such as “he is a good tennis player” or “she is outgoing.” In sum, in Eastern cultures that have a high focus on context, people may see actual speech as being driven by context (e.g., social obligations, formality, concern for relationship), not the speaker’s intention. 4 Likewise, actions that follow speech are more likely to be seen as driven by context (e.g., demands of others, changes of circumstances), not the person himself or herself. Thus, Americans are more likely than Indians to feel that the individual completely controls his or her words and actions, and so to form a trait ascription based on the perceived alignment or lack thereof.
Given these East–West differences in styles of communication and agency, a Western observer (compared with an observer from the East) is likely to accept a verbal commitment at face value, whereas an observer from the East (compared with Westerners) is more likely to consider the verbal commitment in a broader context of subtle relationship negotiations and face-saving gestures, and may understand it as meaning something very different than the apparent promise. And, a Western observer (compared with those from the East) is more likely to feel that later actions taken by the speaker were under his or her control, whereas an Eastern observer (compared with Westerners) is less likely to feel that later actions taken by the speaker were under his or her control. In sum, because BI is a dispositional judgment about the speaker, an Eastern observer may not make a negative trait attribution (low BI) for a spoken promise that is not fulfilled, whereas a Western observer would be more likely to do so. Thus, any “objective” difference in word and actions are likely to engender stronger negative BI perceptions of the speaker for Western than for Eastern observers. Based on our first set of hypotheses (above), differences between words and actions are, as a result, likely to engender a stronger reduction of trust in the speaker in Western than Eastern cultural contexts. 5
We tested the cross-cultural differences as stipulated in Hypotheses 1 and 2 first before proceeding to examine an additional cultural factor—the role of power distance and hierarchy. We used a scenario-based experimental design, to have the cleanest possible examination of cross-cultural differences in the impact of word–deed consistency. This research design is widely used in many areas of management research (e.g., pay, Kim, Park, & Suzuki, 1990; justice, Davidson & Friedman, 1998; deception, Triandis et al., 2001; and culture, Triandis, Chen, & Chan, 1998, just to name a few). The advantage of this approach is that it allows for the presentation of exactly the same stimulus to observers in different cultures to examine how different cultural lenses lead to different interpretations of a specific concrete behavior. In Study 1, we examined how participants from a Western culture, the United States, and those from Eastern cultures, India, Taiwan, and Korea, judged a target person who has kept or not kept the promise in a scenario. We should note that existing research confirms that there is more indirect communication in India than the United States (Kapoor, Hughes, Baldwin, & Blue, 2003), and that Indians are more likely than Americans to see external causes for behaviors (Miller, 1984).
Study 1
Research Participants
We collected data from American, Indian, and Korean adult, non-student subjects (n = 53 from India, n = 83 from Korea, and n = 66 from the United States) and from adult Taiwanese executive students who work full-time (n = 98 from Taiwan). Indian and U.S. data were collected through eLab, an online virtual lab of panelists around the globe based at one author’s university. Subjects were recruited by the lab managers through ongoing postings and advertisements. Due to the fact that English is spoken widely in India, this is a country that is heavily represented in the eLab sample, allowing us to do a cross-cultural study. Participation in an eLab study resulted in a chance to win a US$100 prize (one prize was provided for each 50 participants). Korean data were collected through an online survey, conducted by a Korean market research company. Subjects were all salaried, white-collar employees, who were paid about US$4 for completing the survey. Taiwanese data were collected through an executive training program. All participants were adults who were employed full-time, attending school part-time. Participants received a small gift, worth about US$1.50. For the Korean and Taiwanese sample, the survey was translated into the local language, and back-translated to ensure accuracy of translation (Brislin, 1970). Respondents were 49% male, the median age was 36 to 41 years, and the median education level was college graduate.
Brett, Tinsley, Janssens, Barsness, and Lytle (1997) argued that when doing cross-cultural studies, it is necessary to confirm that the pool of subjects used in a study is culturally typical. To ensure that our samples also display the typical cross-cultural differences revealed previously, we measured their particularism 6 (Leung, 2007), vertical collectivism (Triandis & Gelfand, 1998), and internal (vs. external) locus of control (four items from Earley & Erez, 1997), using a 1 to 7 Likert-type scale, with 1 = low and 7 = high. Results showed the following: Mean levels of internal locus of control (Cronbach’s α = .75) were higher for Westerners (Americans; M = 5.58, SE = .11) than for those from the East (India, Taiwan, Korea; M = 4.33, SE = .08) in our sample (t = 8.13, p < .001). Mean levels of particularism (Cronbach’s α = .79) were higher for those from the East (M = 3.32, SE = .07) than Westerners (Americans; M = 2.17, SE = .11, t = 7.95, p < .001). However, mean levels of vertical collectivism (Cronbach’s α = .84) were no different for those from the East (M = 5.37, SE = .08) than the West (Americans; M = 5.54, SE = .14, t = 1.03, p > .05). 7 Overall, these results suggest that our samples showed responses that were typical of Eastern and Western culture.
Study Design
The study consisted of a set of scenarios where a business meeting is described at which a colleague (named “George” for American subjects) agrees to deliver data needed for a report. We made names “local” to not inadvertently trigger out-group dynamics in the relationship between the subject and the speaker in the scenario (Tajfel & Turner, 1979). In one condition (“promise kept”), the colleague delivers on the promise, whereas in the other condition (“promise not kept”) the colleague did not deliver on the promise (see the appendix for the full scenario). When choosing the name “Vijay” for Indians subjects, we consulted Indian faculty, to ensure that the name chosen was common and did not identify that the person as particularly high or low social status within India. Taiwanese and Korean names were similarly neutral. Thus, we had a 22 (Western/Eastern) × 2 (keep/not keep promise) between subjects experimental design. As a manipulation check, we asked subjects to identify which of the following two statements was true: “George delivered the material by Friday” or “George did not deliver the material by Friday.” All subjects responded correctly based on the experimental condition.
Measures
BI was measured using six items that have been previously developed and used by Simons et al. (2007), using a 1 to 7 Likert-type scale, with 7 = strongly agree and 1 = strongly disagree. This scale includes items such as “George delivers on promises,” “George practices what he preaches,” and “If George says he is going to do something, he will.” As in the past research using this scale, BI is considered to be an evaluation of a speaker—a judgment about his or her general tendency to act in a way that is consistent with his or her words. The alpha for this scale was .89 (.88 for Indians, .97 for Americans, .85 for Taiwanese, and .83 for Koreans). Trust was measured using a scale from Simons et al. (2007), adapted from Mayer et al. (1995), using a 1 to 7 Likert scale, with 7 = strongly agree and 1 = strongly disagree. This scale had three items: “I would be willing to let George have complete control over my future in this company,” “I would not mind putting my well-being in George’s hands,” and “I would feel good about letting George make decisions that seriously affect my life.” The alpha for this scale was .92 (.96 for Americans, .87 for Indians, .89 for Taiwanese, and .95 for Koreans). Means and standard deviations are shown in Table 1.
Means, Standard Deviations, Scale Reliabilities, and Correlations (Study 1).
Note. N = 300. Scale alphas are shown in parentheses. BI = behavioral integrity.
Correlation is significant at the .01 level (two-tailed).
Results
To test Hypothesis 1a, we conducted a one-way ANOVA with perceived BI as the dependent variable and keep/not keep promise as the predictor, including data from all four countries. Keep/not keep promise was significant (F = 345.38, 1, 298, p < .001, ηp2 = .54). Mean perceived BI for promise kept was 5.48 (SE = .08) and for promise not kept, it was 3.23 (SE = .09). To test Hypothesis 1b we used the “mediation” package in R (Tingley, Yamamoto, Keele, & Imai, 2012). To obtain confidence intervals for estimation of the mediation effect, we used bootstrapping with 10,000 iterations. Bootstrapping allows for bias-corrected confidence intervals for significance tests of indirect effects (Mooney & Duval, 1993; Shrout & Bolger, 2002). Overall, the mediation effect of BI on the relationship between promise not kept and trust is −1.48, with a 95% confidence interval of [−1.82, −1.16], which is approximately 88% of the total effect of promise not kept on trust. Thus, Hypotheses 1a and 1b, representing prior known effects of word–deed consistency on BI and trust, were supported.
Before testing Hypotheses 2a and 2b, we checked whether there were significant differences in the effect of keep/not keep promise on perceived BI among the respondents from the three Eastern cultures. Two-way ANOVAs predicting perceived BI showed no significant interaction effects between keep/not keep promise and India/Taiwan (F = 0.26, 1, 147; p = .61), India/Korea (F = 2.13, 1, 132; p = .16), or Taiwan/Korea (F = 1.47, 1, 177; p = .23). Thus, our main analyses to test Hypotheses 2a and 2b combine data from India, Korea, and Taiwan, but for completeness, we also report comparisons with the United States separately for India, Korea, and Taiwan.
To test Hypothesis 2a, that the impact of not keeping a promise on perceived BI would be higher for Westerners than Easterners, we conducted a two-way ANOVA with BI as the DV and Easterner/Westerner and keep/not keep promise as predictors. The two-way interaction was significant (F = 14.14, 1, 296; p = .000, ηp2 = .05). Means for each cell are shown in Table 2, and the East/West interaction effect is shown in Figure 2. The same interaction effect was significant also when looking just at the United States/India (F = 3.80, 1, 115; p = .05, ηp2 = .03), the United State/Korea (F = 15.96, 1, 145; p = .00, ηp2 = .10), and the United States/Taiwan (F = 9.68, 1, 160; p = .002, ηp2 = .06), with the pattern of results closely matching those shown in Figure 2. These results showing that the impact of keep/not keep promise on BI was significantly stronger for Westerners and those from the East.
Cell Means and SD (Study 1).
Note. BI = behavioral integrity.
Effect of promise kept/not-kept on perceived BI (Study 1).
To test Hypothesis 2b, that the East/West differences just reported for BI would carry through to trust (a first-stage moderated mediation effect model), we used the “mediate” package in R. As shown in Table 3, the conditional indirect effect of promise not kept on trust was more strongly negative for Westerners than those from the East. The conditional indirect effect was also stronger for Americans than Indians, Koreas, and Taiwanese, analyzed separately.
Results of Conditional Indirect Relationships (Study 1).
Promise not kept → Perceived BI → Trust.
Note. N = 300 for East/West, 119 for United States/India, 149 for United States/Korea, and 164 for United States/Taiwan. The conditional indirect effects were based on 10,000 bootstrapping resamples. BI = behavioral integrity.
To check whether there are measurement biases across American participants and those from the three East Asian nations, we calculated three pairwise differential item functioning (DIF) 8 analyses of the items of BI and trust scales. That is, we conducted United States–India, United States–Korea, and United States–Taiwan DIF to check whether American and Indian/Korean/Taiwanese participants with the same latent traits showed different probabilities of giving a certain response on the survey. For the United States–India DIF, results show that the response probabilities across the two groups were different on three items: “When George promises something, I can be certain that it will happen,” total DIF: Pr(χ2, 1) = 4e-04, R2 = .0303; “If George says he is going to do something, he will,” total DIF: Pr(χ2, 1) = 0.0021, R2 = .0228; “I would feel good about letting George make decisions that seriously affect my life,” total DIF: Pr(χ2, 1) = 0.002, R2 = .0305. Specifically, the DIF results showed that the Indian participants tended to choose more extreme response options when they had the same attitudes as their American counterparts on these three items. However, because our findings were opposite to these response biases (we found that Americans had stronger responses to promise breaches than Indians with lows [in response to promise not kept] and highs [in response to promise kept]), more extreme than those of their Indian counterparts, our findings could not have been an artifact of measurement bias across the two groups. This pattern of response biases was opposite of our predictions, making it more difficult for us to have found support for Hypothesis 2.
We found very similar results of United States–Taiwan and United States–Korea DIFs. Specifically, both the Taiwan and Korean participants tended to choose more extreme response options when they had the same attitudes as their American counterparts on these two items: “George conducts himself by the same values he talks about,” total DIF United States–Taiwan: Pr(χ2, 1) = 0.19, R2 = .0034, and total DIF United States–Korea: Pr(χ2, 1) = 0.64, R2 = 5e-04; and “George shows the same priorities that he describes,” total DIF United States–Taiwan: Pr(χ2, 1) = 0.31, R2 = .0017, and total DIF United States–Korea: Pr(χ2, 1) = 0.07, R2 = .0063, and less extreme response options when they had the same attitudes as their American counterparts on only one item: “George practices what he preaches,” total DIF United States–Taiwan: Pr(χ2, 1) = 2e-04, R2 = .0246; total DIF United States–Korea: Pr(χ2, 1) = 0.57, R2 = 6e-04. The Taiwan participants also tended to choose more extreme response options when they had the same attitudes as their American counterparts on “I would feel good about letting George make decisions that seriously affect my life,” total DIF: Pr(χ2, 1) = 0.16, R2 = .0037, whereas Korean participants tended to choose more extreme response options on “I would not mind putting my well-being in George/Vijay’s hands,” total DIF: Pr(χ2, 1) = 7e-04, R2 = .0223. Again, like we have argued before, a majority of these differences actually were opposite to our predictions and, thus, should have made it more difficult to support Hypothesis 2. Therefore, our findings would not have resulted from measurement biases across American participants and those from the three East Asian nations. In sum, the DIF analyses have indicated that the cross-national patterns we found are not artifact of measurement bias.
Discussion
Study 1 supported our hypotheses that the response to actual word–deed inconsistency is not as strong in Eastern cultural contexts—India, Korea, and Taiwan—as in at least one Western cultural context—the United States. Given the same described behaviors (a promise is made and kept vs. made and not kept), Easterners in our sample showed less difference in their perceived BI evaluation of the speaker across these two conditions than did Americans. Those BI evaluations in turn were associated with trust in the speaker, and the effect of keeping a promise on trust was mediated by BI. Despite these differences, the overall pattern of effects—with word/deed consistency affecting BI and trust in all countries—shows that the constructs of BI and trust operate similarly across cultures. At the same time, the impact of a broken promise on BI perceptions was significantly less potent in Eastern than in Western subject populations.
One factor that was not included in our first study was the impact of hierarchy on reactions to word–deed inconsistency. Although there are cultural differences in the use of language, this difference may be especially pronounced when the speaker is in a position of authority.
Hierarchy as Second Moderator
In addition to general differences in approach to communication, Eastern and Western responses to communications are likely to be affected differently by hierarchical differences between the parties. Those from the East are typically much higher than Westerners in the culture dimension of power distance (Hofstede, 1980). This concept refers to the degree to which hierarchical differences in society are deemed to be valid and appropriate, and to which deference is accorded to authority. For Westerners, individualism leads to a sense of egalitarianism—each person is recognized as being autonomous and worthy, so even those lower in social positions are worthy of respect and recognition. In terms of family life, it is natural and expected that children rebel and break away from their parents. In education, questioning teachers is allowed and even rewarded. In organizations, officials can seldom assume that their dictates will be followed without question or challenge of some form. In all these ways, it is not seen as legitimate to use one’s hierarchical position to control others very much. In Eastern cultures, by contrast, hierarchy is more absolute. Children obey parents, students do what teachers say, subordinates do what their bosses tell them. The power that comes from hierarchy is legitimate and valued.
According to Chiu and Hong (2006),
In countries with small power distance, people in all strata of the society have equal rights, and people value and respect individuality. In the workplace, power and authority are decentralized, and the salary range from top to bottom is narrow. (p. 11)
In such contexts, supervisors ideally behave democratically, consulting with subordinates. Similar dynamics can be found in the family and school, where children are treated as equals and teachers are seen as facilitators of student-focused learning, not conveyers of knowledge. These examples stand in contrast to countries that are high in power distance. There, those higher in the hierarchy have enormous privileges, authority is respected, and, according to Chiu and Hong (2006),
In the workplace, power and authority are centralized, subordinates are expected to take orders, and there is a wide salary range from top to bottom. The ideal boss is a benevolent autocratic paternal figure who makes major decisions with the benefits of the subordinates in mind. (p. 11)
Given the differential importance of hierarchy in the East and West, we should expect different sensitivity to word–deed consistency of those in positions of authority, speaking to subordinates. We provide two reasons. First, if there is a tendency among those from the East to understand that spoken words may be chosen to help manage social relations, not convey true meaning, this expectation may be especially strong for those of authority; taking a more paternalistic role than in the West, Eastern leaders are more fully responsible for the welfare of subordinates, and thus, may be especially likely to shape speech in ways that protect subordinate’s feelings and social relations. Second, we expect that in Eastern cultures, people will simply be more deferential to higher status speakers, granting them more leeway for word–deed inconsistency. That is, we expect that higher status individuals will be judged less harshly, not because they are right, but because their status determines that they should be followed regardless. It is not the role of a subordinate to question those in authority.
For these reasons, we expect that when there are word–deed inconsistencies or broken promises, people from the East would evaluate bosses less harshly than they would subordinates. In short, the status of the speaker will affect Eastern and Western perceivers differently, with hierarchical relationship being a major component for evaluating BI of word–deed breaches among people from the East, but not Westerners; thus, we expect that cross-cultural differences in response to word–deed breaches will be stronger when those breaches are done by bosses than subordinates. As a result of this, we also expect that cross-cultural differences in trust evaluation will be stronger when the word–deed breaches are done by bosses than subordinates.
Study 2
Research Participants
To analyze cultural and hierarchical differences in response to BI, we collected data from U.S. and Indian adult subjects (n = 79 from India, and n = 110 from the United States). Data were collected through eLab, as described for Study 1. To ensure that the samples were culturally typical (Brett et al., 1997), we measured particularism (Leung, 2007) and vertical and horizontal collectivism (Triandis & Gelfand, 1998), as in Study1, but replaced locus of control with power distance (“PD”; Hofstede, 1980) due to our theoretical focus on hierarchy in Study 2. PD was also measured using a 1 (low) to 7 (high) Likert-type scale. Results showed that the mean levels of particularism (Cronbach’s α = .85) were higher for Indians (M = 3.11, SE = .15) than Americans (M = 2.00, SE = .09, t = 6.81, p < .001). Mean levels of vertical collectivism (Cronbach’s α = .86) were higher for Indians (M = 6.13, SE = .12) than for Americans (M = 5.41, SE = .10, t = 4.53, p < .001). Mean levels of horizontal collectivism (Cronbach’s α = .80) were higher for Indians (M = 5.67, SE = .13) than for Americans (M = 5.41, SE = .09, t = 1.71, p < .10), and mean levels of power distance (Cronbach’s α = .79) were also higher for Indians (M = 4.04, SE = .15) than for Americans (M = 3.61, SE = .09, t = 2.70, p < .01). Overall, these results replicated previously found cross-cultural differences, suggesting that our sample was typical of Indian and American culture.
Study Design
We used the same scenario as used in Study 1, but took a different approach to identifying word–deed inconsistency. Instead of having a promise kept versus promise not kept experimental condition, we told all subjects that the speaker did not do what he said he would do—everyone, then, received the “promise not kept” vignette. However, we also asked whether the subject believed that the speaker had actually made a promise, using a three-item Likert scale (1 = low, 7 = high). The more strongly the subject felt a promise had been made, the stronger the subjectively experienced word–deed inconsistency (a promise not kept) for that subject. Inversely, to the degree that the subject thought a promise had not been made, the less word–deed inconsistency would be seen as a result of knowing that the manager did not deliver the report as he had said he would. The reason there is room for ambiguity in terms of “promise made” is that some people may interpret the actor’s request response of “sure” in the scenario (“Will you do this?” “Sure.”) to be a vague statement and not a clear promise, whereas others may interpret it to be a rock-solid promise. The three-item scale measuring “perceived promise made” included the following items: “George/Vijay promised Fred/Anand that he would deliver his assessment of the negotiations by Friday,” “Based on this report about the meeting, I would expect George/Vijay to have the report by Friday,” and “When George/Vijay said ‘sure’ in the meeting, he meant to send Fred/Anand the assessment.” Cronbach’s alpha for this scale was .67 (.70 for Americans and .64 for Indians), which denotes reasonable reliability. Because in all cases the manager did not deliver the report by Friday, high scores on perceived promise made can be interpreted as high levels of experienced word–deed inconsistency. To be consistent with the language used in Study 1, we use the label “promise not kept” for this variable, but call it “perceived” promise not kept to indicate that it is perceptual variable, not an experimental condition. Perceived promise not kept here is a continuous variable.
At the same time, we added an experimental condition: hierarchical position of the promise maker. In one condition, we identified the manager who said “sure” in response to a request as the boss, whereas in the other condition, we identified him as a subordinate. Thus, we had a 2 (United States/India) × 2 (boss/subordinate) experimental design, with the participants’ score on perceived promise not kept as a continuous predictor as well. As a manipulation check, we asked subjects to identify whether the manager who agreed to deliver the report was the boss or the subordinate of the person to whom he was speaking. All subjects responded correctly based on the experimental condition. A correlation table is provided in Table 4.
Means, Standard Deviations, Scale Reliabilities, and Correlations (Study 2).
Note. N = 189. Scale alphas are shown in parentheses. BI = behavioral integrity.
Correlation is significant at the .05 level (two-tailed). **Correlation is significant at the .01 level (two-tailed).
BI and trust were measured as in Study 1. 9 Cronbach’s alpha for BI was .91 (.89 for Americans and .91 for Indians) and for trust was .90 (.96 for Indians and .85 for Indians). Again, we checked for measurement biases across the two nations by calculating the DIF of the items in the BI and trust scales. Results show that only the last item of the trust scale (“I would feel good about letting George/Vijay make decisions that seriously affect my life.”) showed a significant DIF, total DIF: Pr(χ2, 1) = 0.0072, R2 = .0296. However, because the R2 statistics of the DIF was negligible (<.13) according to Jodoin and Gierl’s (2001) recommendation, there was no need for score adjustments. In sum, the two scales did not show discernible measurement biases across the two nations.
Results
To test Hypothesis 3a, that the impact of not keeping a promise on perceived BI would be higher for Americans than Indians, especially when the speaker is a boss (compared with a subordinate), we conducted a three-way GLM with BI as the DV and Indian/American, boss/subordinate, and perceived promise not kept as predictors. The three-way interaction was significant (F = 5.46, 1, 181; p = .021, ηp2 = .03). 10 To understand the direction of these relationships, we looked at correlations between perceived promise not kept and BI as a function of the hierarchical relationship of the speaker (see Figure 3). When observing a boss speaking, there was a clear negative effect of perceived promise not kept on perceived BI of the speaker for Americans (r = −.31, p < .05) but not for Indians (r = .10, n.s.), and the difference between these correlations was significant (z score = 2.81, p = .005). When observing a subordinate speaking, there was a clear negative effect of perceived promise not kept on perceived BI for the Indians (r = −.34, p < .05) but not for Americans (r = −.20, n.s.), but the difference between these two relationships was not significant (z = 1.01, p = .31). Thus, Hypothesis 3a was supported. Another way to look at these relationships is to say that, for Americans, the correlation between perceived promise not kept and perceived BI was no different when observing a boss versus a subordinate (z = .57, p > .10), but for Indians this difference was significant (z = 1.95, p < .05). It appears that Indians are more critical of BI breaches of subordinates than bosses. However, as shown in Figure 3, Indians have the highest BI ratings for subordinates who keep their word, rather than bosses who keep their word. It seems that for Indians, subordinates are not so much punished (compared with bosses) for not keeping their word, as they are rewarded (more than bosses) for keeping their word.
Effect of Promise Made × Country × Boss/Subordinate on perceived BI (Study 2).
Next, to test Hypothesis 3b, that the impact of cultural context and hierarchy on BI would carry through to trust (a first-stage moderated mediation model), we conducted a bootstrapping analysis to see whether the three-way interaction effect of American/Indian × Boss/Subordinate × Perceived Promise Not Kept had an indirect effect on trust through BI. We used the “mediation” package in the program “R” to conduct this analysis. As shown in Table 5, the conditional indirect effect of perceived promise not kept on trust was significant for Americans (−.15) but not Indians (.06) when the speaker was a boss, and this difference in conditional indirect effects was significant. The conditional indirect effect of perceived promise not kept on trust was significant for Indians (−.24) but not Americans (−.08) when the speaker was a subordinate, but this difference was not significant. Thus, we can conclude, as suggested in Hypotheses 3a and 3b, that the country differences in the effect of perceived promise not kept are greater when the speaker is a boss than a subordinate. Americans responded significantly more harshly to a boss’ broken promise than did Indians.
Results of Conditional Indirect Relationships (Study 2).
Perceived Promise Not Kept → Perceived BI → Trust.
Note. N = 110 for United States, 79 for Indian. The conditional indirect effects were based on 10,000 bootstrapping resamples. BI = behavioral integrity.
Discussion
This study confirmed that the reaction to perceived word–deed inconsistency is greater for Americans than Indians, but mainly in situations where the speaker is a boss rather than a subordinate. Americans do not differ dramatically in response to the difference in hierarchy, whereas Indians’ responses to word–deed inconsistency differ substantially based on the speaker’s hierarchical level. A boss is not judged or evaluated strongly when he or she displays word–deed inconsistency, whereas a subordinate is deemed to have lower BI when he or she displays word–deed inconsistency. These differences carry forward to judgments of trust in the speaker.
Furthermore, this second study assessed word–deed consistency in a different way than in Study 1. In Study 2, we looked only at cases where the scenario explains that the manager did not do what he said he would do. However, even when our portrayal of the speaker’s actions is the same, there still was variation in subjects’ perceptions of whether a “promise” was really made, which produces variance in subjects’ experience of word–deed inconsistency or of a broken promise. Thus, what matters is not just what was actually done in a situation (in our case, not delivering the data by Friday), but also whether people interpret the affirmative response to a requested action as truly being a promise.
So far, our studies have focused on subjects who live in two different cultural contexts. Our next question is whether these differences carry over to the way judgments are made of people who are outside of a person’s cultural context. That is, do people who are aware of cultural norms, adjust their evaluations of BI based on expectations about local cultures? Is awareness of differential responses to word–deed consistency part of the repertoire of culturally knowledgeable actors?
Judging People in Other Cultures
The dynamic constructivist approach to the study of culture (Hong & Chiu, 2001; Hong, Morris, Chiu, & Benet-Martínez, 2000) posits that people do not just operate according to one set of cultural norms consistently, but rather they may be aware of multiple cultural norms and use them based on the context. For example, Hong and associates found that Hong Kong Chinese, who are aware of both Chinese and Western cultural norms, made different attributions of a situation based on whether the Chinese or Western cultural frames are triggered (Hong et al., 2000). Specifically, they made more external (vs. internal) attributions of social events, which was consistent with the Chinese cultural norms, when in a context with Chinese cultural cues (e.g., a photo of the Great Wall) than when in a context with Western cultural cues (e.g., a photo of the Statue of Liberty). Likewise, Taiwan managers with highly integrated bicultural identity showed preference for Chinese managerial practices in a Chinese environment (i.e., an office with Chinese decoration and music), and showed preference for Western managerial practices in a Western environment (i.e., an office with Western decoration and music; Friedman, Liu, Chi, Hong, & Sung, 2012). Similar effects of switching between cultural norms or frames have been shown in other studies with a wide range of dependent measures, including spontaneous self-construal (Ross, Xun, & Wilson, 2002), cooperative behaviors (Wong & Hong, 2005), and neural activities (Chiao et al., 2009), and have also been replicated in studies that used different bicultural samples (Chinese Canadians, Dutch Greek bicultural children) and a variety of cultural primes (e.g., language, experimenter’s cultural identity; Ross et al., 2002; Trafimow, Triandis, & Goto, 1991; Verkuyten & Pouliasi, 2002). In sum, there is ample evidence supporting the idea that bicultural individuals can respond in a culturally typical way as a function of the cultural context at hand. Our question for Study 3 is as follows: Is understanding of cross-cultural differences in response to BI central enough in social dynamics that culturally informed actors would adjust their own judgments of speakers based on cultural context?
We tested this idea by having students in Singapore (which has a population composed of three major ethnic groups: Chinese, Malay, and Indian) evaluate actual word–deed consistency (keeping a promise) in Indian and American cultural contexts. Singapore is not just multi-cultural because of its ethnic diversity but also because of its history; it is a former British colony, which has left a heavy British influence on its values, legal system, and language. Moreover, it is a center of global commerce. As a result, it has a cultural heritage “that reflects values of both the East and the West” (Li, Ngin, & Teo, 2008, p. 950). Given the large representation of Indians within the population, and exposure to Western language, business, and culture, Singaporeans are considered to be culturally aware of both Western and Indian cultural norms, and are often studied as samples of “biculturals” (Chen, Ng, & Rao, 2005). In addition, biculturalism is also reflected in Singaporean students’ fluency in both English and their particular language (Chinese, Malay, or Tamil; Bishop, 1998; Tavassoli & Lee, 2003; Wharton, 2000).
Because Singaporean subjects recognize cultural differences between India and the United States, we expected them to make different judgments in response to a verbal agreement that is not kept. Given the prominence of hierarchy in evaluating actions in India, as shown in Study 2, we focused on perceived power distance as the core cultural point of evaluation. In India, there is greater deference to authority (due to higher power distance), whereas in the West there is much less deference to authority (due to lower power distance). To the degree that Singaporeans perceive differences in power distance between the cultural contexts, we expect to see reactions to word–deed consistency track power distance, producing effects similar to those found in Studies 1 and 2.
Note that the logic here is similar to that of Gelfand and Realo (1999), who sampled subjects in multiple cultures with varying levels of collectivism, but then focused just on individual-level measures of collectivism as the predictor, not country. Here, too, we asked Singaporeans to read about the incident happening in India and the United States—to get greater variance in perceived contextual power distance—but then focused directly on perceived power distance as the driver of the effect.
Study 3
Participants
One hundred fourteen students at a university in Singapore were recruited for course credit or S$4 (the paid and non-paid participants were randomly assigned to the conditions). The mean age of these students was 21 years. Fifty-eight were male and 56 female. One hundred one were Chinese, nine were Malay, three were Indian, three were “Other” (multiple ethnic identities were allowed). Given the multi-ethnic structure of Singaporean society, these students are very familiar with both Indian and Western cultural contexts. Advertisement of the study was sent to the students and interested students participated through eLab, as described for Study 1.
Procedures
Students were provided with a similar scenario to those used in Studies 1 and 2, with three conditions: keep/not keep the promise, takes place in India/United States, and the speaker is a boss/subordinate, using a between-subject design. Students read a scenario, and were asked to evaluate the BI of the speaker as they expected would happen in India/United States, and identify the level of trust in the speaker they expected would occur in India/United States. They were also asked to evaluate the degree to which the social norms in the cultural context where the incident happened were higher or lower in power distance. This perception of power distance was used as the cultural dimension of relevance, because it indicated the degree to which the subject sees Indian/American culture as high or low in deference to those in positions of authority. The study then included three predictors: promise kept/not kept, boss/subordinate, and perceived power distance. Manipulation checks for “promise kept/not kept” and “boss/subordinate” showed that all respondents were able to accurately report the experimental condition they received.
Measures
The measures used for trust and BI were the same as those used in Studies 1 and 2. Cronbach’s alpha for BI was .95, whereas it was .92 for trust. 11 The Power Distance scale was taken from Earley and Erez (1997), and included eight items, such as “In most situations managers should make decisions without consulting their subordinates” and “employees should not express disagreements with their managers.” The only change we made for our study was that the subjects were asked to respond in the way that they believe people in the local culture (India or the United States) would respond. Thus, it represented their perception of the cultural norms of power distance operating in the scenario. Cronbach’s alpha for this scale was .89. Correlations among the key variables are shown in Table 6. In all analyses, we added a control for whether or not the subject was of Indian ancestry. We reasoned that those Singaporeans who are Indian might react differently to the Indian scenario than non-Indian Singaporeans.
Means, Standard Deviations, Scale Reliabilities, and Correlations (Study 3).
Note. N = 114. Scale alphas are shown in parentheses. BI = behavioral integrity.
Correlation is significant at the 0.05 level (two-tailed). **Correlation is significant at the 0.01 level (two-tailed).
Results
To test Hypothesis 4a, that perceptions of BI in response to keeping or not keeping a promise would be influenced by both the perceived power distance of the cultural context and whether the speaker was a boss or subordinate, we conducted a GLM with BI as the Dependent Variable and included a three-way interaction between keep/not keep promise, boss/subordinate, and perceived power distance. The three-way interaction was significant (F = 3.98, 1, 106; p = .048, ηp2 = .04). The results are graphed in Figure 4, which shows means for ±1 standard deviation for perceived power distance, for promise kept and promise not kept, for boss and subordinate speakers. Figure 4 shows that when Singaporean students observed a subordinate speaking, the reaction to promise kept versus not kept did not vary based on the perceived power distance of the context. However, if they observed a boss speaking, the reaction varied based on the perceived power distance of the context. To confirm the depicted relationship when the subordinate was the speaker, we conducted a GLM with perceived BI as the dependent variable, and Promise Kept/Not Kept × Perceived Power Distance as the predictor, including only subjects in the subordinate condition. That interaction effect was not significant (F = 0.36, 1, 51; p = .55, ηp2 = .007). To confirm the relationship depicted in Figure 4 when the boss was the speaker, we conducted the same GLM, including only subjects in the boss condition. That interaction effect was significant (F = 5.31, 1, 55, ηp2 = .09). Thus, like we found in Study 2, the strongest impact of culture occurs when looking at word–deed consistency of bosses, although the most salient target in Study 2 was Indian bosses, who appeared to be left unjudged by subjects for their word–deed inconsistency, whereas the most salient target in Study 3 was the American boss, who appeared to be judged much more critically than others for word–deed inconsistency. Given that Study 3 results were based on how Singaporean students judged reactions to the same scenario in two foreign cultures, it is noteworthy to see a similar three-way interaction effect that occurred when we studied actual American and Indian adults in Study 2.
Effect of Promise Kept × Power Distance × Boss/Subordinate on perceived BI (Study 3).
To assess Hypothesis 4b, that this three-way interaction effect would carry forward to trust judgments of the speaker (a first-stage moderated mediation effect), we used the “mediation” package in R and the bootstrapping option to test whether there was an indirect effect of the three-way interaction on trust through BI, as we did in Studies 1 and 2. As shown in Table 7, the conditional indirect effect of promise not kept on trust was more negative in the boss condition when Singaporean subjects thought the context was lower in power distance (i.e., more Western; −2.04) than when they thought the context was higher power distance (i.e., more Eastern; −1.37), and the difference between these two effects was significant. However, there were no significant differences in reactions based on perceived power distance in the subordinate condition. Thus, we can conclude, as suggested in Hypotheses 4a and 4b, that bicultural observers expect greater negative reaction to promises not kept when they perceive that the scenario context is lower in power distance, but only if the speaker is a boss.
Results of Conditional Indirect Relationships (Study 3).
Promise Not Kept → Perceived BI → Trust.
Note. N = 114. The conditional indirect effects were based on 10,000 bootstrapping resamples. BI = behavioral integrity; PD = Power distance.
Conclusion
Study 3 shows that cultural differences in response to BI are central enough to social interactions that culturally informed people—in this case, students from Singapore—responded differently to word–deed inconsistency based on their perceptions of the cultural context. They recognized that responses to word–deed inconsistency, for bosses, are quite different in cultures that are lower in power distance than those that are higher in power distance.
Overall Conclusion
Our results suggest that, although BI does matter for people from the East and has a pattern of effects on trust that are similar to those from the West, the impact of word–deed consistency is less pronounced for those from the East than the West, it is more strongly affected by the hierarchical level of the promise maker, and these differences carry forward to effect trust responses. Cultural differences in the way people make BI perceptions matter, because they profoundly affect responses to leaders. Awareness of this dynamic is critical for cross-cultural or expatriate managers. Managers who operate in multiple countries and cultural contexts need to be aware of differences in the foundations of employee trust in these different countries, so they can manage more effectively. This knowledge should enhance both “cognitive” cultural intelligence (or “CQ”)—because it provides information about norms and practices in different cultures—and “behavioral” CQ—because it can guide how managers act—which enhances cultural judgments, cultural adaptation, and employee performance (Ang et al., 2007).
These results provide further support that words have a somewhat different purpose in Eastern managerial contexts than in American ones (Nisbett et al., 2001; Sanchez-Burks, 2002). It is likely that, to some extent, Eastern speakers make statements that are not intended as an indicator of future action, but are instead intended for other purposes such as maintaining social ties or helping the other party save face in front of others (Ting-Toomey, 1988; Yum, 1988). If that is the case, then an apparent breach from a verbally stated promise does not present as much of a problem to employees who view the leader through that lens. Indeed, for American managers speaking in India, Taiwan, or Korea, they can sometimes expect that the listener will make appropriate “discounts” of stated intentions, so the problem created by a word–deed inconsistency may be relatively small. Conversely, Indian, Taiwanese, or Korean managers operating in Western contexts need to be especially vigilant about these issues, as their perceived integrity (and hence their credibility) will likely diminish if they fail to follow through on an utterance. Our data suggest that BI matters in both cultural contexts, but that American employees are especially vigilant.
Also, the difference we found in response to bosses and subordinates is consistent with Eastern culture being more particularistic (Trompenaars & Hampden-Turner, 1998); judgments are not applied in a similar way across all individuals, but are adjusted based on who the person is and what that person’s relationship is to you. Americans are more universalistic in ethical judgments, we found, applying the same criteria to bosses and subordinates. Indians, in contrast to Americans, judged their subordinates’ word–deed inconsistencies more carefully than those of the superiors, and that these differences carried through to affect trust judgments. Furthermore, we were able to replicate the perceived power distance underpinning of this difference (Chiu & Hong, 2006; Hofstede, 1980), in two independent samples and using two different methodologies—hierarchy affected Indian and American respondents differently when judging home–country interactions, and perceptions of the power distance of the cultural context also impacted bicultural Singaporean respondents observing interactions in other countries.
These findings have two implications for leaders. One is that, whereas bosses in Eastern cultures can act with relative impunity, they cannot do so in American cultural settings. Thus, it is likely that Eastern managers trying to manage Westerners will be surprised at being judged severely for actions that are often ignored in Eastern cultures, and Westerners are likely to be surprised at being allowed seemingly unfettered control when they expect to be carefully judged for their words and actions. Leaders who know these cultural differences can develop higher CQ by being “aware of the expectations they hold for themselves . . . verify the accuracy of their cultural assumptions . . . [and] adapt their leadership behaviors appropriately” (Rockstuhl, Seiler, Ang, Van Dyne, & Annen, 2011, p. 828). A second implication is whether leader trust is developed differently in East and West: If Eastern cultures are not as responsive to word–deed consistency when looking at words and actions of bosses, might there be some other flavor of “integrity” that is a more important factor for judging leadership and generating leader trust in Eastern cultures? It could be that, in the East, judgment of leaders’ trustworthiness is tilted more toward their actions alone, with somewhat lower weight put on the words themselves. This implication should inform theories of leadership; although the importance of integrity is known to be universal (Dorfman, Hanges, & Brodbeck, 2004), we may still find cultural differences in factors that determine perceptions of integrity (Resick, Hanges, Dickson, & Mitchelson, 2006).
Facing the challenges of cross-border management, being well versed in both Eastern and Western cultures should help (Friedman et al., 2012). As shown among our bicultural Singaporean participants in Study 3, to the extent that they understand the power relations within Indian and American cultures, they are able to discern cultural nuances—judging correctly how people in the respective cultures evaluate the BI and trustworthiness of the target, including understanding the implications if the person being judged is a boss or subordinate.
This research contributes not just understanding that enhances CQ, but also to the growing literature on BI (Simons, 1999, 2002; Simons et al., 2015). Although Simons (2002) pointed toward the importance of learning more about how BI judgments are formed, little attention has been paid to BI formation (other than Simons et al., 2007). As Simons et al. (2012) put it, “perceptual moderators . . . represent a relatively untapped area for research” (p. 338). In this study, we theorize, and found, an effect of culture on BI formation: The same factual level of words (promises) and actions (promise-keeping) produce different BI perceptions in Eastern than Western cultural contexts. Research on BI needs to not just look, as it mostly does now, at the effects of BI, but also how BI perceptions are formed. Indeed, the foundational concept of BI—that people are judged by their word—needs to be adjusted to recognize that language meaning is highly contingent on cultural context.
As with any research, this set of studies has limitations. Our scenario methodology risks drawing on respondents’ lay theories more than actual emotional responses, and also did not focus on actual employees. Although the vignette methodology did allow us to present a very standardized stimulus across cultures, and to see how the exact same broken promise is understood and responded to differently, our confidence in generalizing and applying these lessons would be heightened by examination of, for example, actual employee surveys and performance information for a multinational company. Another limitation is that we focus on just one implication of BI, which is its effect on trust. We know from prior work that BI evaluations of a boss also have effects on employee job satisfaction, organizational commitment, interpersonal justice, and turnover intentions (Simons et al., 2007). We limited our focus her to trust, because it is the most proximal effect of BI (Simons et al., 2014), and because our focus was on the way that cultural context moderates BI judgments in the first place. Still, future work can look at whether cultural differences in BI perceptions also carry forward to affect not just trust but also these other outcomes of BI. Second, although we theorize broadly about BI and expect to see cultural differences in response to both promise breaches and value breaches (the second component of BI), our empirical focus has been on promise breaches as a trigger of BI perceptions. Further research is needed to confirm that there will be the same cultural differences in response to espoused versus enacted value congruence. Third, our studies include paid subjects from an online subject pool of working adults, and a student subject pool. We do not see any biases created by this sample, but further studies might be done with different samples. The current set of studies makes a definite contribution but does not offer the “final word” on the subject of BI across cultural contexts.
Footnotes
Appendix
Associate Editor: William Gardner
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by funding from each of the author’s universities. In addition, preparation of this article was partially supported by a grant to Professor Hong (General Research Fund #14655416 from the Research Grant Council of Hong Kong SAR) and a grant to Professor Chi (MOST-105-2410-H-002-158-MY3 from Taiwan’s Ministry of Science and Technology).