Ethics in the workplace is an important phenomenon that has a critical impact on organizations and societies. Ethical scandals, such as those in business and politics, have called attention to the need to create policies and other mechanisms that can undermine unethical acts. To illustrate, recent projections demonstrated that loss of revenue caused by customs-related corruption costs World Customs Organization (WCO) members at least USD 2 billion in customs revenue each year (OECD, 2017).

The Corruption Perceptions Index 2019 presented by Transparency International every year reveals that almost 70% of the countries assessed scored below 50 on 2019 CPI, with an average score of 43 on a scale of zero to 100, where zero means highly corrupt (Transparency International, 2020). Moreover, a survey conducted by Ernst and Young (2018) with 2,550 executives from 55 countries showed that 38% of the respondents reported that bribery and corruption practices occurred widely in business in their country, and 11% believe it is common to use bribery to win contracts in their sector. The situation is even worse when we consider only the emerging countries – 52% of the respondents believe that bribery and corrupt practices happen widely in business in their country (Ernst & Young, 2018).

In Brazil, investigators found bribery and a bid-rigging scheme involving state-controlled oil giant Petrobras in 2013. They found out that some of Brazil’s largest construction and engineering companies paid billions of dollars in bribes over the years to assure lucrative contracts from the Brazilian oil state company (Mauro et al., 2019). However, corruption is not bound to developing countries, such as Brazil, and it also occurs in developed nations. For instance, the German company Volkswagen, which employs more than 630,000 people worldwide, admitted in 2015 to manipulating 11 m vehicles worldwide to fool emissions tests (The Guardian, 2019).

Those examples demonstrate how unethical behavior can have enormous costs for business. From this perspective, researchers have investigated how and why unethical behavior occurs in the workplace and how to reduce it (Mitchell et al., 2020). The behavioral ethics field suggests that unethical behavior is not always due to a deliberate choice but can be caused by a non-rational decision-making process (De Cremer et al., 2020). This approach helps to understand why good and moral people can still act unethically. It claims that the environment and the situation can activate cognitive schemas and automatic processes.

Hence, contextual factors such as ethical culture stand out as crucial because they can play a pivotal role in enhancing or diminishing unethical acts (Mayer, 2014). For instance, countries such as the United States and the United Kingdom have focused on corporate culture as a mechanism to reduce corruption (Filabi & Bulgarella, 2018). Ethical culture is a relevant phenomenon because it gives employees guidelines for the appropriate conduct in that environment.

In line with the organizational culture field, the construct of culture strength can be considered an important variable that can affect moral behavior. It is expected that employees who have consistent relationships and a consensus on the unit or organization’s norms and rules will be more likely to behave consistently with those established norms. Thus, the concept of organizational culture strength could be transposed to be conceived as the within-unit agreement members about ethical values and norms of the organization (González-Romá & Peiró, 2014). We propose the ethical culture strength construct and that it plays a critical role in predicting unethical acts.

Regarding individuals, moral identity has been one of the most studied constructs related to moral behavior (Jennings et al., 2015). The social-cognitive theoretical framework from Bandura (1991) can be used to comprehend moral identity as a cognitive self-schema around a set of moral traits (Aquino & Reed, 2002). Even though it has been traditionally studied at the individual level, research has pointed out that individuals who share the same environment and interact tend to have similar thoughts and actions (Chan, 1998). Thus, we can considerer the existence of a collective moral identity, as has been first proposed by Kuenzi et al. (2020), that represents the extent to which employees in their work unit internalize and symbolize moral traits as central for them.

Considering these constructs, we adopt the social cognitive theory (Bandura, 1986, 1991) and the social norm theory (Cialdini et al., 1990; Cialdini & Trost, 1998) as our theoretical lens for understanding how ethical culture, its strength, and collective moral identity can interact to predict unethical behavior. The social cognitive theory claims a cognitive interactionist perspective to comprehend moral behavior: moral behavior would be regulated by the influence between thought and self-sanctions, conduct, and a set of social influences (Bandura, 1991). The social norm theory propose that social norms can guide or constrain social behavior, and that those norms emerge from social interaction (Blay et al., 2018). However, investigators have concluded that the social norms must be salient to be capable to prime norm-congruent behavior (Cialdini & Goldstein, 2004). Therefore, the purpose of this study is to test whether ethical culture, ethical culture strength, and collective moral identity predict and interact to predict unit-level observed unethical behavior (OUB) and unethical pro-organizational behavior (UPB). Our study addresses some of the gaps presented by the literature. We want to provide evidence of the mechanisms related to the interaction effect of ethical culture and moral identity. In addition, we explore if unethical pro-organizational behavior can be affected by the ethical culture. Finally, we propose a study at the unit level, which expands previous literature by considering how the sharedness of the evaluated constructs function and influence each other.

This research seeks to contribute to the ethical culture, behavioral ethics, and moral identity literature. First, we contribute to the ethical culture literature by being among the first researchers to theorize ethical culture strength as a contextual variable in work units and examine how it associates with employees’ perception of unethical behavior. Ethical culture strength expands the current research on ethical culture by considering the sharedness of ethical norms and the agreement between unit members regarding ethics and its power to affect observed behavior. For instance, the research on focused climate strength has been done in other contexts, such as safety climate strength (Zohar & Luria, 2005) and leadership climate strength (Schyns & Veldhoven, 2010). However, none has evaluated ethical climate and ethical culture strength.

Second, we expand the behavioral ethics field by testing if ethical culture and collective moral identity interact to predict observed unethical behavior using the social cognitive framework. Despite the direct association of ethical culture and individual moral identity on observed unethical conduct, we assume that ethical culture and its strength could function as cues to activate the self-concept related to moral identity in the work unit.

Third, we test if ethical culture is related to unethical pro-organizational behavior (UPB) distinctively from observed unethical behavior (OUB). Although ethical culture is traditionally negatively associated with unethical behavior (Kaptein, 2011), it can have a less negative effect on unethical pro-organizational behavior since UPB may have positive consequences for the organization.

Fourth, we conduct this study in a developing country (Brazil), considering much of past research has been done in developed countries. This could deliver important insights into how the phenomenon functions in a nation where the perception of corruption is high (Transparency International, 2020)—considering that the level of corruption varies in underdeveloped, developing, and developed countries. This national perception about corruption and ethics can affect how Brazilian companies’ employees make sense of ethical culture and share concepts related to their moral identity.

Theoretical background

In our study, we consider two types of unethical behavior: (1) observed unethical behavior (OUB) and (2) unethical pro-organizational behavior (UPB). To define ethical behavior, we adopt Russell et al.’s (2017) definition: “Unethical behavior at work is a behavior that violates a prescribed norm that is based on a code of behavior at work that is (a) ascribed to by the relevant organization or professional group, (b) prescribed by relevant regulatory bodies or by statute, or (c) widely endorsed in the society” (p. 254). Based on this definition, we consider observed unethical behavior as the perception of any behavior that violates a prescribed norm (social or organizational).

On the other hand, unethical pro-organizational behavior is related to actions that violate social norms, values, laws, or standards, but that promote the effectiveness of the organization (Umphress & Bingham, 2011). It is also an immoral act, but that is conducted, in part, to benefit the organization (Umphress et al., 2010). It is essential to point out that even though employees engage in UPB to help their organization, in the end, it may have damaging results for the organization, for example, in terms of the company’s reputation.

Collective moral identity

From a social-cognitive perspective, Aquino and Reed (2002) defined moral identity as a self-concept around a set of moral traits (e.g., honest, caring). Thus, when individuals value moral traits, like being honest, they have a strong moral identity as a central aspect of their self-concept. According to Aquino and Reed (2002), moral identity comprises two dimensions: (1) Internalization (represents the “having” side) – the degree to which those moral traits are central for the self; and (2) Symbolization (represents the “doing” side) – the degree to which the person acts and expresses those moral traits.

Moral identity has been considered an essential antecedent of moral behavior. A meta-analysis has examined the relationship between moral identity and moral behavior and found a significantly positive association between them (Hertz & Krettenauer, 2016). In 65.3% of studies included in this meta-analysis, the Self-Importance of Moral Identity Questionnaire was used (SMI-Q; Aquino & Reed, 2002), demonstrating the instrument’s importance.

Usually, moral identity is considered an individual construct. However, we assume that people who work together share experiences and perceptions about morality, which could explain a shared perception of the centrality of moral traits in a team or work unit. Following Kuenzi et al.‘s (2020) proposition, wedefine collective moral identityas the extent to which employees in the work unit internalize and symbolize moral traits as central to their shared unit concept. We also apply a direct consensus composition model, as presented by Chan (1998), to assess collective moral identity using Aquino and Reed’s measure.

Collective moral identity can emerge because individuals strive to achieve or to maintain a positive social identity (Tajfel & Turner, 2004), and this positive social identity is based on favorable comparisons with the in-group. In addition, conformity is an imperative element that define and maintain groups acting together to accomplish goals (Cialdini & Goldstein, 2004). Groups provide a variety of means for defining the individual self-concept and for maintaining and enhancing a sense of self-worth (Forsyth & Burnette, 2010). Hence, unit values and unit norms provide a guide to social identity formation. Consequently, those values and norms could impact in the formation of the individual moral identity. Thus, we understand that contextual cues influence unit members to adopt a collective moral identity (Kuenzi et al., 2020).

Individual moral identity has been positively associated with moral behavior because people who consider moral values central to them tend to act more ethically. Thus, we assume that employees with a high collective moral identity will have a shared perception of lower unethical behavior, which lead s us to our first hypothesis.

Hypothesis 1

Collective moral identity will be negatively associated with observed unethical behavior and unethical pro-organizational behavior at the unit level.

Ethical culture

Ethical culture is a subset of the concept of organizational culture and corresponds to the interplay between formal and informal systems that boost ethical behavior or prevent unethical conduct (Treviño & Youngblood, 1990). The Corporate Ethical Virtues (CEV) model from Kaptein (2008) expands previous research on ethical culture by postulating that it is related to the organization’s virtuousness. The CEV model claims that the corporate ethical virtues are the conditions for ethical behavior and can promote employees’ ethical conduct.

The original ethical virtues proposed by Kaptein (2008) are as follows: (1) Clarity: to what extent ethical expectations are clear and understandable to employees and managers; (2) Congruency of management: the extent to which top management and senior management act according to ethical expectations; (3) Congruency of supervisors: to what extent do the immediate supervisors act in accordance with ethical expectations; (4) Feasibility: to what extent does the organization provide sufficient equipment, budgets, and autonomy for managers and employees; (5) Supportability: to what extent does the organization support ethical expectations between management and staff; (6) Transparency: to what extent ethical and unethical conduct is visible to responsible managers and officials; (7) Discussability: to what extent managers and employees have the opportunity to discuss ethical issues; and (8) Sanctionability: the extent to which managers and employees believe there are rewards and punishments regarding (un)ethical behaviors (Kaptein, 2008) developed a measure to capture the virtues of the ethical culture in organizations. In this study, the multidimensional CEV model is used to assess ethical culture.

Past research has found that ethical culture is an antecedent for unethical behavior (Kaptein, 2011), occupational well-being (Huhtala et al., 2011, 2016), absence/absence due to illness (Kangas et al., 2017), intention of rotation (Kangas et al., 2018), organizational citizenship (Ruiz-Palomino & Martínez-Cañas, 2014), work engagement and burnout (Huhtala et al., 2015), among others. Moreover, recent research has demonstrated that various teams within an organization can have different ethical cultures. This construct was relevant to explain outcomes, such as frequency of observed unethical behavior and observed unethical behavior in teams (Cabana & Kaptein, 2019).

Even though ethical culture is a construct that was initially conceived at the organizational level, we can assume the existence of ethical subcultures within an organization, as has been done in the organizational culture literature (Hofstede, 1998). Some past studies have investigated ethical culture at the team level (Kaptein & van Dalen, 2000, Cabana & Kaptein, 2019) or the unit level (Huhtala et al., 2015; Kangas et al., 2017). In this study, we adopt ethical culture as a construct that varies between units, and assume that unit members share similar norms and values about ethics at work.

Members who perceive a more ethical culture in their unit considering aspects such as ethical leaders, fair sanctions, rewards, etc., will tend to see ethical behavior more frequently. On the other hand, units with a shared perception of a weaker ethical culture are more likely to observe other employees’ unethical behavior.

Even though unethical pro-organizational behavior is also considered a violating act, employees behave with the intention to maintain or increase the positive perception of being affiliated with their organization (Umphress et al., 2010). Based on social identity theory, organizational identification can explain how the social context can influence how people behave to conform with the organization (Chen et al., 2016). Thus, we expect that employees with a high organization identification will be positively associated with UPB, as has been showed by past studies (Chen et al., 2016; Umphress et al., 2010).

Considering that the employees who strongly identify with their organization are more likely to engage in UPB, we can assume that people that perceive their unit culture as more ethical will perceive UPB less frequently than the traditional unethical behavior, since UPB is not a clear unethical act because it benefits the organization. Hence, the association of ethical culture and UPB will be lower, since UPB can be viewed as been not so harmful to the organization.

Thus, considering the effect of ethical culture on the perception of unethical behavior, we assert the following hypotheses:

Hypothesis 2a

Ethical culture of the work unit will be negatively associated with observed unethical behavior and unethical pro-organizational behavior at the work unit level.

Hypothesis 2b

Ethical culture of the work unit will have a weaker negative association with observed unethical pro-organizational behavior compared to observed unethical behavior at the work unit level.

Ethical culture strength

Ethical culture strength is derived from the culture strength and climate strength literature. Traditionally, climate strength has been more studied than culture strength and is represented as an extent of agreement within units on climate perceptions that is related to different attitudinal and behavioral unit-level outcomes (Schneider et al., 2017). From this perspective, the study of focused climates emerged, like ethical climate and safety climate research, and, consequently, focused climate strength.

Even though a strong culture is often characterized as homogeneous, cohesive and where employees’ goals are aligned with management goals, it has been criticized that it oversimplifies the concept (Saffold, 1988). Hence, researchers have come out with different conceptualizations of culture strength; some take only one dimension into account (focusing on alignment or congruence), others consider two (such as agreement and consistency) or three dimensions (like intensity, agreement, and pervasiveness) (González-Romá & Peiró, 2014). Following González-Romá and Peiró’s (2014) suggestion, we conceptualize culture strength as “the degree of within-unit agreement about culture elements (e.g., values and normative beliefs)” (p. 525). This definition clarifies the construct meaning in the literature and removes ambiguity by operationalizing it as a single dimension concept.

Considering this definition, we propose the ethical culture strength concept as a focused culture concept such as has been done in the climate literature (e.g., service climate and safety climate). Thus, ethical culture strength represents the agreement within-unit members about ethical values and norms of the organization. We propose that ethical culture strength has a negative association with the perception of unethical behavior since unitswhich have high levels of agreement regarding ethical norms will have a greater consensus on how to behave when facing moral dilemmas. This indicates the following hypothesis:

Hypothesis 3

The work unit’s ethical culture strength will be negatively associated with observed unethical behavior and unethical pro-organizational behavior at the work unit level.

Interaction of moral identity, ethical culture, and its strength

Bandura’s social cognitive theoretical framework could explain how aspects related to the individual and the context could interact. The social cognitive theory claims that social aspects have behavioral effects through the psychological mechanisms that operate in the self-system (Bandura, 2001). From this perspective, thoughts are not neutral, the self is socially constructed, and people do not only operate reactively, but also proactively (Bandura, 2001).

Moreover, the focus theory of normative conduct from Cialdini et al. (1990) have been recently introduced to the empirical business ethics literature (Blay et al., 2018). This theory advocates two types of social norms: (1) the descriptive norms that are related to the perceptions of what most people do in a given situation, and (2) the injunctive norms that reflect the perceptions of what most people think others should do in a given situation (Blay et al., 2018; Cialdini et al., 1990). From this perspective, individuals that conform to a descriptive norm are more prone to be considered part of a group, because they want to fit-in and to behave as most people in the group do. These norms also emerge from organizations and work units, since they employees interact and create a sense of group identity.

Based on these approaches, by conceptualizing moral identity as a cognitive self-schema (Aquino & Reed, 2002) and considering the effect of social norms on behavior, situational cues such as those brought by the ethical culture could influence behavior by activating knowledge structures and schemas, including moral identity (Shao et al., 2008), and even collective moral identity. Through the lens of social learning theory (Bandura, 1977), we comprehend that employees learn unethical behavior by observing their leaders’ and colleagues’ behavior and by noticing the reward and disciplinary policies for unethical behavior brought by the ethical norms of their units’ ethical culture. Thus, it is expected that an ethical culture would impact the relationship between a moral self-construct, such as moral identity, and observed unethical behavior. The ethical culture has the role of reinforcing the collective moral identity that exists in that group, consequently affecting moral behavior and its perception.

Regarding ethical culture strength, even though past research has considered strength as a “main effects” model, it does consider the possible interaction effect it could have (González-Romá & Peiró, 2014). Therefore, besides its main effect on observed unethical behavior, we propose that ethical culture strength moderates the extent to which ethical culture moderates the relationship between moral identity and the perception of unethical behavior. The ethical culture content will influence this relationship, but this association will be weaker or stronger depending on its within-unit agreement. However, in some cases, there may beless agreement between members concerning aspects from the ethical culture. In that case, it is expected that ethical culture will have a weaker influence on the relationship between moral identity and observed unethical behavior.

Hence, we propose a two-way and three-way interaction with collective moral identity, ethical culture, and ethical culture strength on the perception of unethical behavior at work. Taking this into account, we propose the following hypotheses:

Hypothesis 4a

Ethical culture will moderate the relationship between collective moral identity and unethical behavior such that the negative association of collective moral identity with the perception of unethical behaviorwill be stronger when units have a higher ethical culture.

Hypothesis 4b

The work unit’s ethical culture strength will moderate the effect of ethical culture o n the relationship between collective moral identity and unethical behavior at the work unit, such that a lower ethical culture strength will decrease the association of ethical culture on the relationship of collective moral identity and the perception of unethical behavior.

The research model is summarized in Fig. 1.

Fig. 1
figure 1

Research model

Method

Participants and procedures

The present study sample consisted of 2208 employees from 116 different units working in ten Brazilian organizations. Due to missing information regarding the respondents’ work unit and considering the minimum of three unit members to include in the analysis, the useable dataset was reduced to 1942 employees from 96 units. The average unit size was 16.15 (SD = 10.83). The largest unit included 48 members, and the smallest unit included three members.

The majority (55%) of the participants were men and were, on average, 44.8 years old (SD = 12.41). Of the total sample, more than 70% had, at least, a university degree. The respondents had been working in the organization for 13.77 years (SD = 6.98) on average. Of the ten organizations, three were public institutions, and seven were private. Most units belonged to the three public organizations (72%).

The participants had to fill in an online survey. To reduce the common method variance bias, the questionnaire included different response formats (e.g., reversed scored items, different Likert scales), and anonymity was guaranteed to all participants (Podsakoff et al., 2003). The organizations agreed to participate in this study and were responsible for spreading the survey. All employees from the ten organizations received an invitation to answer the electronic survey. The order of the measures was always the same and the items order were random. Informed consent was obtained from all individual participants included in the study. With a return of 2,208 questionnaires, a global response rate of 11.7% was achieved. The international ethical guidelines, consistent with the American Psychological Association (APA) guidelines, were followed in this study.

Measures

Ethical Culture. The Brazilian Portuguese version of the Corporate Ethical Virtues Scale (CEV) (Kaptein, 2008) with 36-item, measuring seven dimensions of ethical culture, was administered. Participants answered to the items (e.g., “My supervisor is honest and reliable”) using a six-point response format (1 = Strongly Disagree, 6 = Strongly Agree). The CFA for a seven-factor model of the scale showed an adequate fit (χ² = 2757.9, df = 587, RMSEA = 0.04, CFI = 0.92, TLI = 0.92, SRMR = 0.04).

Moral Identity. We applied the moral identity scale translated and adapted to Brazilian Portuguese (Resende & Porto, 2017) of the Aquino and Reed (2002) measure. This scale displays a set of moral traits (caring, compassionate, fair, friendly, generous, helpful, hardworking, honest, and kind) that may describe a person. The participants had to visualize the kind of person who has these characteristics and imagine how that person would think, feel, and act. Then, they answered nine items within two dimensions – internalization and symbolization – on a Likert scale that ranges from 1 (strongly disagree) to 5 (strongly agree). The internalization dimension had five items (e.g., I strongly desire to have these characteristics), and the symbolization dimensions had four items (e.g., The types of things I do in my spare time clearly identify me as having these characteristics). The two-factor structure demonstrated a good fit to the data (χ² = 155.9, df = 26, RMSEA = 0.05, CFI = 0.94, TLI = 0.91, SRMR = 0.04).

Unethical Behavior at Work. To assess unethical behavior at work, we administered two scales. We chose to apply proxy-report measures to minimize the social desirability bias since this bias is possibly the most important problem in research on ethical behavior (Wouters et al., 2014).

The first one was the Observed Unethical Behavior in Organizations Scale (MacLean et al., 2015; adapted from Treviño & Weaver, 2001) with seven items (e.g., “Calling in sick just to take a day off”). Respondents were asked how often they observed other employees from their company performing a list of unethical behaviors on a frequency scale of 1 (Never) to 5 (Very frequent). The original scale had eight items, but one item was removed from the scale (“Dragging out work to get overtime”), because most of employees in public organizations in Brazil are not entitled to overtime pay. The one-factor structure demonstrated an adequate fit to the data (χ² = 149.59, df = 13, RMSEA = 0.08, CFI = 0.94, TLI = 0.91, SRMR = 0.03).

The second was the Unethical Pro-Organizational Behavior Scale (Umphress et al., 2010) with six items. Participants had to indicate the degree of agreement with a set of statements about other employees behaving unethically to help the organization in an agreement scale from 1 (totally disagree) to 7 (totally agree). An example item is: “If it would help the organization, other employees would misrepresent the truth to make the organization look good.” The referent was changed from “I” to “Other employees” in order to reduce social desirability bias. The unifactorial structure of the scale showed an adequate fit (χ² = 76.42, df = 9, RMSEA = 0.07, CFI = 0.97, TLI = 0.94, SRMR = 0.02).

Reliability coefficients of the measures used in this study are presented in Table 1 and were adequate.

Table 1 Intraclass Correlation Coefficients, Within-Team Agreement Indices and Reliability Coefficients

Control variables. We controlled for the organization that the unit belonged, unit size, and tenure (measured by the number of years working for the organization). Since data was collected in units from different organizations, it was important to control the impact of belonging to each organization. Past research has shown the impact of group size and tenure, since larger groups are expected to think in a more heterogeneous way than smaller ones (Jehn, 1995), and members with longer time in the organization to be more embedded in the culture (Jehn et al., 1999), which may affect behavior.

Data aggregation

Since the study model was performed at the unit level and the data were collected at the individual level, we investigated whether it was appropriate to aggregate the variables to obtain unit scores. First, we computed the intraclass correlation coefficients (ICC) to determine the proportion of total variance due to the unit level. The ICC values higher than 0.05 indicate that there is enough between-group variance and values lower than 0.05 indicate that there may be little value in conducting multilevel modeling (Bliese, 2000). The aggregation coefficients and indices obtained are shown in Table 1. The ICC values for all the variables were higher than 0.05 in this study; thus, there was sufficient between-group variance.

Next, we estimated within-unit agreement by calculating the rwg statistics (George & James, 1993) and by means of the Average Deviation Index (ADI; Burke et al., 1999). The criterion for AD was computed as c/6 (where c is the number of response categories in the response scale). For variables with five categories on the response scale (moral identity and OUB), the AD must be below 0.83. For variables with six categories (ethical culture), the AD must be below 1, and for the one with seven categories (UPB), the AD has to be below 1.16. For the rwg index, the 0.70 cut point has been a traditional criterion. However, some authors have argued that it dichotomizes agreement, and that this cut-off point may be too high (Lebreton & Senter, 2008). Thus, they suggest that rwg values between 0.51 and 0.70 can be accepted as the existence of a moderate agreement (Lebreton & Senter, 2008).

The variables related to unethical behavior and moral identity had a rwg higher than 0.70, and the AD was below the maximum value. Regarding ethical culture, all the dimensions had a rwg above 0.51(except for the sanctionability dimension), demonstrating a moderate agreement. The AD criterion was met only for the clarity and congruency of supervisor dimensions. However, as we are also interested in the ethical culture’s strength, we suppose it would be important to have more variability in this variable. Thus, we decided to consider all the dimensions that met the rwg cut-off point of 0.51 and eliminated from the subsequent analysis the sanctionability dimension for not achieving any of the within-unit agreement statistics.

As shown in Table 1, different dimensions of ethical culture were shared within the work units: 7–60% of the total variance was explained by unit homogeneity. These results provide evidence of agreement within units for all the variables, except for ethical culture’s sanctionability dimension.

Data analysis

To model the relations between collective moral identity, ethical culture, and ethical culture strength, we employed the AMOS 21.0 structural equation modeling software (Arbuckle, 2012) using maximum likelihood estimation since we had a multivariate normal data and a reasonable sample size. We ran separated models to test our hypotheses of the association of ethical culture, ethical culture strength, and collective moral identity with our dependent variables (OUB and UPB). All scores used in the analysis were at the unit level and we grand-mean centered the variables.

To assess model fit, we chose to use indexes in addition to the chi-square statistic due to the influence of sample size on the chi-square statistic. Thus, we evaluated the root mean square error of approximation (RMSEA), the comparative fit index (CFI), the Tucker-Lewis index (TLI), and the standardized root mean square residual (SRMR).

For RMSEA, values below 0.05 are considered excellent fit, values between 0.05 and 0.08 are considered good fit, and values higher than 0.10 indicate a poor fit (Hu & Bentler, 1999; MacCallum et al., 1996). For CFI, values above 0.95 and 0.90 are considered excellent and adequate fit, respectively (Hu & Bentler, 1999). For TLI, values near 1.0 indicate good fit, and it is conventional to use a threshold value of 0.90 as an indication of good model fit (Hox & Bechger, 1998). For SRMR, a value of zero indicates perfect fit, and a value of < 0.08 is generally considered a good fit (Hu & Bentler, 1999).

Data Availability

The dataset generated and analyzed during the current study is available in the Open Science Framework (OSF) repository at https://osf.io/p49br/?view_only=5d983bb2b9e849da811730334a66c918.

Results

Descriptive statistics and correlation coefficients for all the variables are presented in Table 2. Collective moral identity was not significantly correlated with unethical behavior (OUB and UPB). On the other hand, ethical culture and ethical culture strength were strongly negatively related to unethical behavior (OUB and UPB).

Table 2 Descriptive Statistics and Correlations for Study Variables

H1 hypothesized that collective moral identity would be negatively associated with both measures of unethical behavior. However, the results showed no statistically significant associations between the collective moral identity and both unethical behavior measures. Thus, H1 was not supported.

H2a hypothesized that ethical culture would be negatively associated with observed unethical behavior and H2b that the association would be weaker on UPB. Table 2 shows negative, statistically significant correlations between both measures of unethical behavior and the dimensions of ethical culture (p < .01), except for the feasibility dimension. For the measurement model, the latent factor of shared perceptions of ethical culture was represented by the six ethical dimensions representing the unit’s ethical culture. The standardized factor loadings ranged from 0.31 to 0.88 (see Fig. 2), which indicates a reasonable construct validity of the CEV model. The results of the structural equation model presented in Fig. 1 showed that shared perceptions of ethical culture among members of the work units were significantly related to lower shared perception of observed unethical behavior and unethical pro-organizational behavior. The model had a reasonable fit (χ2(17) = 28.93, p = .04; CFI = 0.97; SRMR = 0.04; RMSEA = 0.08). Thus, H2a was supported. Furthermore, the model showed that ethical culture had a weaker association with UPB compared to OUB, which gives support to H2b.

Fig. 2
figure 2

Structural equation model for the association of ethical culture with unethical behavior

Notes. Standardized estimates are shown. Model fit: χ2(17) = 28.93, p = .04; CFI = 0.97; SRMR = 0.04; RMSEA = 0.08 [90% CI = 0.02, 0.18; p-close = 0.13].

The third hypothesis was that ethical culture strength would be negatively associated with unethical behavior (OUB and UPB). In Table 2, some of the dimensions of ethical culture strength are significantly negatively associated with unethical behavior. As done in the ethical culture variable, ethical culture strength was represented by the six strength dimensions that represent the unit’s ethical culture strength. Figure 2 shows the standardized factor loadings that ranged from 0.65 to 0.83, which demonstrates an adequate construct validity of the model. The structural equation model results for ethical culture strength (see Fig. 3) showed that ethical culture strength among members of the work units was also significantly related to lower shared perception of observed unethical behavior and lower shared perception of unethical pro-organizational behavior. The model showed a good fit (χ²(17) = 23.89, p = .12; CFI = 0.98; SRMR = 0.04; RMSEA = 0.06). Thus, H3 was supported.

Fig. 3
figure 3

Structural equation model for the association of ethical culture strength with unethical behavior

Notes. Standardized estimates are shown. Model fit: χ2(17) = 23.89, p = .12; CFI = 0.98; SRMR = 0.04; RMSEA = 0.06 [90% CI = 0.00, 0.12; p-close = 0.31].

The final hypotheses (H4a and H4b) predicted the full model, in which there would be a two-way interaction between ethical culture and collective moral identity on predicting unethical behavior and a three-way interaction including ethical culture strength in the previous relation. However, since collective moral identity was not a significant predictor of the dependent variables, the interaction models were also non-significant. Hence, H4a and H4b were rejected.

Discussion

This study’s main purpose was to investigate the contribution of ethical culture, ethical culture strength, and collective moral identity association to predict unit-level OUB and UPB while examining the moderating relationships of ethical culture and its strength. The relevance of this study was established by the exploration of collective variables and the interaction of group constructs.

Our study results suggest that ethical culture content and ethical culture strength have a strong and direct association with OUB and UPB. Thus, employees who work in units that have more ethical and strong cultures perceive more ethical behavior in their units. Moreover, it shows that ethical culture had a weaker association with UPB. This probably happened because employees who behave unethically to benefit their organization are seen as committed to their organization and more engaged, as past studies have shown the relationship of UPB and organizational identification (Chen et al., 2016; Umphress et al., 2010). Thus, this kind of behavior is seen as less harmful by the colleagues, which explains the reasons for a weaker association.

However, we did not find a significant association of collective moral identity with unethical behavior, nor the two-way and three-way interactions were significant in the model. Yet there is empirical evidence of the robust impact of moral identity on moral behavior (Hertz & Krettenauer, 2016), we could not find the relationship of collective moral identity and unethical behavior in our study. The research on moral hypocrisy can explain this, which means appearing moral to others while avoiding the cost of actually being moral (Batson et al., 1999, 2002). The concept of moral hypocrisy could be applied to understand the behavior of a group or work unit. Research in groups and business ethics have shown the powerful influence of peers on unethical behavior (Treviño et al., 2014). For instance, past studies show that when an in-group member cheats, other in-group members are also more likely to cheat (Treviño et al., 2014). Because of this tendency to conform to the descriptive norms of the group, unit members can also try to keep a good image of their group while not actually being moral. Hence, the moral hypocrisy phenomenon could occur in the group to protect the group image and the group concept.

In a series of studies conducted by Batson et al. (1999), they found out that moral hypocrisy motive exists and is powerful and pervasive. Later, Batson et al. (2002) investigated if this powerfulness of moral hypocrisy could have alternative explanations; however, they did not find evidence for it and confirmed the existence of the moral hypocrisy phenomenon. Those studies on moral hypocrisy suggest that moral identity is not sufficient to cause moral motivation and can be even too weak to really influence ethical behavior (Hertz & Krettenauer, 2016). Consequently, collective moral identity may not be sufficient to affect unethical behavior in the group, because moral hypocrisy can probably emerge in work units.

We also recognize that our moral identity measure had self-referent items, which may have impacted the effect of the concept on group variables. Even though we have found evidence for the emergence of the collective phenomenon in the units investigated, the self-referent items may have decreased its influence on unit-level variables. Thus, we encourage that future research should replicate the investigation on collective moral identity by means of a referent-shift model.

Additionally, we must consider the cultural context where this data was collected. Brazil has a phenomenon known as the Brazilian jeitinho, which refers to a social mechanism used by Brazilian citizens to deal with difficult situations that arise in daily life and for troubleshooting (Duarte, 2006a, b), and was historically built-in society since the colonization of Brazil by the Portuguese (Barbosa, 1992). This concept has been more strongly associated with corruption and inappropriate behavior at work (Smith, 2008). The Brazilian jeitinho is associated with the breakdown of social norms, yet it is still perceived as a valid strategy to solve problems and deal with bureaucracy (Pilati et al., 2011). Moreover, Brazil is a country where the perception of corruption is very high, occupying the 106th position in the global ranking of 180 countries (Transparency International, 2020). Therefore, considering the Brazilian context, moral hypocrisy could easily emerge in individuals and in groups. This finding indicates that groups claim to have a colletive moral identity, but it does not truly predict their perception of their peers’ moral behavior. This phenomenon may explain corruption’s pervasiveness in different countries and should be replicated in other countries with a high perception of corruption.

Our research makes several notable contributions. Overall, we demonstrate the emergence of collective phenomena related to unethical behavior in organizations. First, we contribute to the ethical culture literature by introducing the novel concept of ethical culture strength and showing that it has a significant negative association with unethical behavior. Second, we demonstrate that ethical culture has a significant association with observed unethical pro-organizational behavior, even though it has a weaker association.

We also contribute to the literature by demonstrating that moral identity may not always be a good predictor for ethical behavior - this effect may vary depending on the context. Particularly when we consider work units, collective moral identity may not affect the perception of unethical behavior by unit members. Past research has found out that collective moral identity was able to moderate the relationship between ethical organizational climate and unethical behavior in organizations (Kuenzi et al., 2020). However, we were not able to find this result in our study. Since collective moral identity is not a significant predictor, it is evident that ethical culture and its strength will not show a significant interaction to predict unethical behavior. This means that future research should seek other individual difference concepts that could interact and explain ethical behavior at the workplace. Moreover, future research should focus on investigating the emergence of moral hypocrisy in groups and work units.

Our findings also indicate that groups within units develop a common understanding of shared norms and traits, such as ethical culture and moral identity, affecting the group’s behavior. We also contribute by bringing comprehension of the phenomenon in a non-WEIRD (Western, Educated, Industrialized, Rich, and Democratic) sample, and where the perception of corruption is large and spread among the society.

Regarding practical implications, our research highlights the potential relationship of ethical culture and unethical behavior. Frequently, attention is given to the so called “bad apples” disregarding the organizational context’s powerful effects. Hence, our research points out that organizations that invest in ethical norms and values can positively affect the company. The CEV model proposed by Kaptein (2008) helps managers’ intervention by clearly specifying the aspects that can enhance ethical behavior and prevent unethical acts. Moreover, we assume that, besides the organizational ethical culture, there are ethical subcultures in different teams or work units that expand the comprehension of how the environment affects the workplace.

Despite our research contributions, we point out some limitations that would provide valuable opportunities for future research. The first limitation is that the data is cross-sectional, which undermine causal inferences of the model. Future research could benefit by measuring the effect of ethical culture using a longitudinal design. Second, all the measures used in this study were self-reported surveys, which have issues related to common method bias (Podsakoff et al., 2003). Despite the effort to reduce this bias, future research could improve by collecting data with different sources or collecting raw data such as the number of hotline reports. Third, we measured only moral identity using the classic self-report scale from Aquino and Reed (2002). It would be fruitful to assess other moral self-constructs (e.g., moral emotions, moral judgment disposition, etc.) that might have a stronger impact on unethical behavior, possibly interacting with contextual factors. Fourth, we have a potential social desirability bias in subjects’ responses, even though we guaranteed anonymity and confidentiality.

Regardless of these limitations, this study provides a clear contribution to our understanding of ethical culture and its association with unethical behavior (OUB and UPB). It highlights that contextual factors, such as ethical culture and its strength, are more relevant to comprehend observed ethical behavior than collective moral identity. It advances the field by demonstrating the association of ethical culture and observed unethical pro-organizational behavior. Finally, considering that only three studies from 132 on organizational ethics research were conducted in South America (McLeod et al., 2016), this study advances the literature by assessing the phenomenon in a Latin American country, where the perception of corruption is high. We hope that this research provides useful insights and measures for researchers and professionals interested in examining the important role of an organization’s ethical culture and its strength.