Business Ethics Decision-Making: Examining Partial Reflective Awareness

Abstract

The aim of this research was to examine partial reflective awareness in ethical business choices in Croatia. The ethical decision-making is interlinked with sustainable practices, but it is also its prerequisite. Thus, better understanding of business ethics decision-making provides a basis for designing and implementing sustainability in a corporate setting. The research was done on student populations who will soon carry important roles and make important decisions for individuals, organizations, and society. The field research was conducted using Kohlberg’s scenarios. The results reveal that the process of decision-making goes through the lenses of respondents’ own preferred ethics. However, the reflective awareness of respondents’ preferred ethics is skewed and regularities in that deviations point out to the relevance of the context characteristics and arousal factors. In addition, the individuals do not use all available information in the assessment process. The revealed partial reflective awareness contributes to explanation of why people have problems with justifying their choices. As there are many examples of unethical behavior in the environment that remain unpunished, it is necessary to raise awareness of the issue. Improvement in reflective awareness would contribute to more sustainable ethical choices and reveal a possibility of an intervention design within the higher education framework.

1. Introduction

Although business scandals remain present in almost all economies of the world, they are especially widespread in developing countries with transition economies. These countries are in the process of change from one political and/or economic system to another. The transition does not refer only to the system change, but also to the change in people’s general attitudes, values, and mindset, which consequently affect ethical decision-making in business economics. In addition, developing countries tend to be late adopters of novel business practices related to ethical principles, such as corporate responsibility and sustainability. This creates additional motivation to examine ethical business choices in Croatia, as it is a transition post-communist country with moderate corruption rating, high unemployment and emigration rate, and no lack of business scandals.

For a long time, the dominant theoretical concepts in business ethics have been the moral judgment and moral action approaches. Since evidence shows that moral action and moral judgment do not always lead to the same results, a moral judgment-action gap [1] has been identified. This evokes the question of bridging this gap and identifying appropriate concepts for its description. As a step in this direction, Schwartz [2] offered an integrated ethical decision-making model. Theoretical approaches mostly differ in terms of the questions they seek to answer while examining individual and group behavior in confrontation with ethical dilemmas regarding business and policy. An integrated approach would try to answer what, how and why people decide the way they do when facing ethical dilemmas. So, what is an ethical dilemma? Cherre et al. [3] quoted Bagnoli’s interpretation of an ethical dilemma as a “deliberative trap with no way out”. This can certainly be said for the famous trolley problem, but ethical dilemmas have also evolved to a subtler approach, which enables the measurement of the ethical development level and the mental processes that underlie the ethical choice. Some of the more subtle and realistic ethical dilemmas have been developed for research purposes in business ethics. While business ethics tries to understand the underlying principles that govern organizational behavior, novel approaches find that the organizational behavior is aggregated individual behavior within that organization and focus their efforts on understanding decision-making processes while solving ethical dilemmas at the individual level. Such individual behavior is susceptible to judgment according to generally accepted moral norms [4]. If organizational behavior is observed as aggregated individual behavior, then it is a reflection of employees’ ethical and other attitudes. That relevance of the employees’ ethical principles and their alignment with the company’s policies lies in the possibility of their impact on the design and implementation of policies, especially the ones related to sustainability and corporate responsibility. The intertwinement of individual and organizational behavior is a subject of research from the organizational, psychological, and educational perspective [5,6,7]. The focus is commonly placed on the individual’s decision-making competencies and the trade-off between the maximization of personal benefits and practicing virtuous behavior. Behavioral ethics examines actual behavior of people and related influences to reveal patterns and use them as nudges towards ethically desirable behavior [8,9]. Bridging the gap requires the explanation and understanding of both decision-making processes and behavioral patterns. However, the source of the gap might lie in the thought patterns and derive from the inconsistences in persons’ own ability of reflection or metacognition.

The motivation for revealing the mosaic of influences in the ethical decision-making process, as [10] eloquently put it, is that “if we can understand how humans deviate from their own more reflective preferences, we can design interventions that will lead them to their more ethical selves.”

The numerous papers published on the topic of relationship of the education and work experience with the ethical decision-making points to the intuitively plausible relationship and the importance of the subject. However, different research questions, variables and used methods lead to unconclusive results [11,12,13,14] and the need for further investigation. The research interest on the subject has not decreased over time, primarily because of the dominant belief that higher education has a decisive influence on the civil and business citizenship, which is mandatory for a functional and liberal civil state. The aim of the business ethics education is to improve moral behavior and the corporate social responsibility, which lead to improvement of the decision-making and policies, and consequently enable corporate sustainability. That raises the question as to whether students’ moral reasoning skills improve with the implementation of the ethic education or mature later due to the decision-making positions in the enterprise. The comparation of the students’ and employees’ populations deliver mixed results [11,15,16,17], which support ongoing scientific curiosity and the necessity for identification of the factors which play a role in the development of the moral reasoning and ethical decision-making in the business context. Cognitive theory assumes that education is a strong predictor of moral reasoning, as well as cognitive development. Social environments (including culture elements such as basic assumptions, feelings, beliefs, values, behaviors, etc.) may differ in how individuals process moral attributions, or appraise a level of morality based on the assessment of observable behavior. Research [18] confirmed that Croatian universities do not implement adequate educational procedures to encourage moral development, probably because moral development is nurtured in high school education and is not in the focus of higher education. The use of business case studies in higher education has a positive influence on the students’ ethical perception [19], while attending ethics class has a positive influence on their moral judgment [20].

The novel research has taken two main streams depending on whether they focus on the cognitive or behavioral component of the business ethics decision-making. However, the point where the two approaches intersect is the prerequisite of awareness. While the cognitive approach treats the awareness as a stage [21], the behavioral approach observes the awareness as a psychological process [10,22]. The awareness may refer to the perception of the characteristic issues (individual, situational and environmental factors) and to own psychological processes and used information (such as ethical guidelines or contextual factors). The latter is usually referred to as the reflective awareness or ethical metacognition. The awareness as a process may be susceptible to numerous influences (i.e., biases, heuristics, emotions, etc.), leading to the omission of the information, skewed interpretation of the situation and/or skewed decision-making processes. Those deviations are referred to as the bounded or partial awareness, and we adopt the latter term for the purpose of this paper and use it in the sense of an individuals’ ability to identify the information upon which s/he made a decision. While partial reflective awareness caused by the skewed view of one’s decision-making processes and ethical guidelines may lead to skewed ethical decision-making, it can also be observed from the perspective of reflective skills [23,24], meaning that it can be trained and improved, which opens a possibility for intervention design.

1.1. Classical Theories in Business Ethics

In an effort to establish the determinants of ethical decision-making and behavior, Kohlberg [25,26] proposed levels of cognitive moral development: pre-conventional, conventional, post-conventional. Each level consists of two stages/categories, in particular, pre-conventional: obedience and punishment orientation, and individualism and exchange; conventional: good interpersonal relationship and maintaining social order; post-conventional: social contract, individual rights, and universal principles. Even though the model is heavily criticized due to its androcentricity, research design, methodology, and judgment-action gap [27,28], it remained present in the research of business ethics and often used as a base model.

Later development of modeling the cognitive aspect of morality focuses on the decision-making process and one of the most frequently used base models is Rest’s [20], which includes four stages of the decision-making process in ethical dilemmas: awareness, judgment, intention and behavior. Further development of Rest’s model included various individual characteristics and environmental factors. Jones [29] provides an extension of the basic model by including the characteristics of the issue itself. The examined characteristics include: magnitude of consequences (sum of the harms/benefits of the act), social consensus (degree of social consensus about the act—(un)ethical), probability of effect (will the act occur and will there be consequences), temporal immediacy (length of time between the act and consequences), proximity (closeness of the decision-maker to the persons who will suffer/benefit from the consequences), and concentration of effect (magnitude of harm given a number of people— [30] show that their respondents are not sensitive to this guideline). Further improvements refer to the extension to cognitive-affective framework.

Gaudine and Thorne [31] examined the role of emotions in ethical decision-making given a certain stage of moral development. They extended Rest’s [32] and Kohlberg’s [25,26] model by introducing the role of emotions. Positive and negative emotions, as well as the level of arousal, influence ethical decision-making differently. For example, “individuals experiencing low arousal and positive affect are less likely to recognize an ethical dilemma”, “arousal is positively associated with an individual’s tendency to formulate a prescriptive judgment consistent with his or her level of moral development” and “increased level of cognitive activity [as a consequence of the arousal] would likely result in an individual accessing more cognitive moral structures, thereby increasing the likelihood that an individual has accessed a cognitive moral structure consistent with their level of moral development”. Arousal is a requirement for awareness and engagement of morality in the ethical decision-making process and the quality of the decision, similar to the high intensity of constructs in Jones’s model. However, the positive/negative aspect of the aroused emotions plays a significant role. According to Kligyte et al. [33], anger inhibits, and fear facilitates, ethical decision-making, while negative effects can be decreased by emotion regulation.

While offering important insights into relevant elements of the ethical decision-making process, the theories of cognitive development and cognitive-affective models focus on the awareness, assessment, and judgment part of the process, frequently ignoring the issue of actual behavior. This creates the so called “judgment-action gap” [1].

1.2. Novel Theories in Business Ethics

Novel theories within business ethics apply interdisciplinary approach and induce concepts that combine ethics with decision theory (and associated frameworks), behavioral economy and social psychology. According to Fort [34], “moral rationality is strongly bounded in terms of business ethics. This is because business life is “artefactual”; the rules of business life can vary widely and are not “natural”. Such a statement leads to the conclusion that the business ethics decision-making is presumably easier to model, describe and to use nudges. However, at the individual level, there are not many differences in the approach to the ethical dilemmas involving different aspects of life, which shows the transferability of the decision-making patterns applied to ethical dilemmas in different aspects of life at the individual level. In addition, the organizational level in business ethics deals with another aggravating factor, that is, aggregation of individual behavior [4,5,6,7,8,9]. The theories within the business ethics framework have been developing in such a direction to become focused on individual behavior and [1] systematize the theoretical development into three areas: Moral Automaticity, Moral Schemas, and Moral Heuristics Research.

While mainstream theories regarding virtue ethics, consequentialism and deontology (also, often referred to as orthodox) seek guidelines as to how individuals should decide and act regarding the ethical norms and primarily examine cognitive processes culminating in moral judgment, the novel approaches differ regarding the inclusion of environmental influences and different psychological processes (not only cognitive) in the examination of decision-making in ethically ambiguous situations. The difference is also noticeable in terms of methodology, as the novel approaches seek to examine actual behavior and are more focused on the reasons of a certain act, rather than the act itself (why, rather than what). According to Bowman [35], the rational choice models explain choices only in situations with the perfect information and possibility to calculate costs and benefits. As people rarely have perfect information at their disposal or the ability of perfect calculation, and are susceptible to feelings, intuition and perception, the rational choice models reach their limits quite quickly when applied to real-world situations. The same author claims that many choices are based on subconscious thoughts and prejudices, and that sometimes people’s acts and values are inconsistent.

According to Gigerenzer [36], bounded rationality studies the cognitive processes, including emotions, which people rely on while making decisions in the world at large, and “it results not from character traits or rational deliberation alone, but from the interplay between mind and environment”. Haidt [37] finds that “intuitions like hunches are the primary source of moral judgments, as rational arguments are commonly used post hoc to justify determinations”. The strain of theory that searches for rational roots of ethical decision-making considering the psychological processes that govern decision-making creates the concepts of applied bounded rationality to ethical decision-making, bounded awareness in ethical decision-making, bounded ethicality, and behavioral ethics.

One of the simplest approaches to bounded rationality says that “individuals act rationally in the sense that they take into account all matters they deem relevant regardless of the amount of information they have at their disposal to assist their deliberations” [38]. Such an approach puts an individual in the rational framework, but also considers the individual’s values, priorities, and intentions, as well as imperfections of the decision-making process. The psychological processes involved in the decision-making process refer to assessment and judgment [39]. The judgment, as in moral judgment, focuses on the choice between the alternatives. The assessment refers to the process that precedes the judgment and involves perception and evaluation of the alternatives (and available information) and assignment of the outcome probabilities and outcome utility (utility in the broadest sense, measured in a currency relevant for the individual).

Some research studies distinguish between awareness and assessment, while others use the assessment as an umbrella term. Ethical decision-making is susceptible to those processes, as is decision-making on any subject. Thus, it makes sense to apply the same framework and methodology to the decision-making in ethical dilemmas. However, such an approach also implies that an individual is aware of the bounded setting and own rules applied to information selection, information processing and criteria for the selection of available ethical alternative. “Bounded awareness leads people to not “see” important, accessible, and perceivable information during the decision-making process” [10], and the same approach could be applied to the internal part of the ethical decision-making process. As there is a gap between the way people act and their values, the extension of this gap may be related to the awareness of their own values. The bounded ethical metacognition or partial reflective awareness would arise if a difference occurs between the reported values and the actually applied values in the decision-making process.

According to Chugh and Bazerman [40], “the bounded rationality research has been limited to exploring how people misuse and mis-integrate the information that is part of their cognitive set”. The examination of bounded ethical metacognition is an extension of such an approach, but the misused or mis-integrated information is the one regarding own set of values used as the criteria in the decision-making process. The motivation for revealing the mosaic of the influences during the ethical decision-making process is to design the interventions or nudges towards ethical behavior [10]. While the bounded rationality approach examines the deviations in the decision-making process in ethical dilemmas, regarding both the process and its outcomes, the bounded awareness approach examines the role of the awareness of alternatives, motivation and consequences, and all relevant decision-making factors, whether within an individual or in his/her environment; and bounded ethicality focuses on the processes that lead to the decision in ethical dilemmas, disregarding the nature of the decision.

Bazerman and Sezer [10] and Chugh et al. [22] defined bounded ethicality as a psychological process that has similar characteristics to (or is a subset of) bounded rationality and bounded awareness, that “lead people to engage in ethically questionable behaviors that are inconsistent with their own preferred ethics”. Rees et al. [41] defined bounded ethicality as the extension of Simon’s bounded rationality in terms of the efforts to understand “the systematic and biased decision-making processes that contribute to a gap between individuals’ preferences and their actual behavior”. For Gigerenzer [36], much of the moral behavior is based on heuristics—mental processes that disregard part of the available information and rely on satisficing, without the optimization computation. The research in the bounded rationality/awareness/ethicality area reveals specific systematic patterns, “mental shortcuts” in decision-making that affect ethical dilemmas (differences in declarative and actual behavior, the role of defaults, the role of fear and anxiety, the role of power, competition and anonymity, system 1 cognitive functioning prone to ethical fallacy, majority behavior, framing effect, blind spots, focusing failure, focusing illusion, implicit attitudes and discrimination, in-group/out-group bias, discounting the future, overclaiming credit, equality heuristic, tit for tat, default heuristic, framing of information, slippery slope, motivated blindness, indirect blindness, ethical fading, status quo bias, preference falsification, overconfidence bias, moral equilibrium scorecard, naïve idealism, availability effect, action bias, forbidden fruit, bias blind spot, representative heuristic, anchoring heuristic, moral cowardice, ethical blindness, denial, prioritization of information, motivated reasoning, [1,3,8,10,26,35,36,40,41,42,43,44,45,46,47,48]). The skewed ethical reasoning sometimes occurs when judging other people’s behavior, where people tend to judge the behavior more harshly if it harms identifiable persons [49,50] or change their ethical judgment if the unethical act is carried out through third parties [51,52].

Most of the examined effects deal with cognitive failure to recognize and/or act on own or others’ unethical choices/actions. The roots of those choices vary from the lack of mental capabilities and fallacies in reasoning, to motivation, up to the adaptation to the environment. These patterns tend to explain how and why people act the way they do and reveal discrepancies between the preferred and actual behavior. At the same time, most of the findings disregard the quality of the choice made, thus leaving the judgment-action gap open. The specified bounded ethicality framework can be extended to ethical metacognition or partial reflective awareness, where discrepancies in actual and reported values (and their priorities) occur while making ethical decisions, aiming to answer the question, as Chugh and Kern [43] put it, “why are we sometimes blind to our own transgressions?”. The reflective awareness is a part of the reflective thinking [53], which is thought process involving a consideration of belief or knowledge in the light of its prepositions and following conclusions.

Reflective thinking and awareness also involve reflective skills, cognitive aspects of thought and transformational thinking [24]. According to the same authors, reflective thinking plays an important role in sustainable education, as it does not only enable acquisition of competences, but also finding reasons and criteria, improves decision-making and judgment, and facilitates achieving novel solutions due to the reformulation of the knowledge. In case of perceived failure, reflective thinking plays a crucial role in determination of the following decisions [23], as it contributes to a process of becoming a virtuous person by improving the process of learning, achieving self-perfection and tendering harmonious relationships. According to the same authors, the routing of the reflective thinking from one’s ability to the invested effort contributes to the students’ persistence in academic learning and virtue building. Despite its importance, reflective thinking is still in the early stages of application, as it has been scarcely incorporated into learning programs. In this paper, we use the term reflective awareness for the students’ awareness of how they construct their own ethical decisions, do they account for the possible reasons and consequences and whether they are aware of their own belief system (measured from the perspectives of moral cognitive development and moral philosophy). Reflective skills are recognized as important in education and its sustainability, and we propose the extension of that conclusion to business ethics. The relationship between the business ethics curriculum, sustainability and responsibility has previously been suggested as a prerequisite for shaping future decision-makers’ attitudes and competences, as well as for building their integrity and morality [54,55,56,57,58]. Correspondingly, the reflective thinking proved to have an important role in development of quality learning and knowledge application to new and realistic situations [59,60], also, [61] show that reflective (and cooperative) learning environment enable development of competences for sustainability. Stated reveals a firm relationship between the ethics curriculum, reflective skills nurturing and attainment of complex competences (such as competences for ethical decision-making and the competences for sustainability). The examination of students’ reflective awareness in ethical dilemmas contributes to the innovative educational application of the reflective thinking aiming for educational sustainability, but also provides insight into future employees and possibly managers, who will soon face their share of ethically ambiguous situations in the practice. In addition, for Joaquin and Biana [62], sustainability science is ethics, as they both examine “what is good, what is right and what we ought to do”, and the integration of ethics into decision-making processes inevitably leads to the sustainability. While the authors primarily explore the translation of scientific findings into social policies, we build on this thought and propose that integration of business ethics into business should lead to sustainable business. However, the application of the business ethics does not depend only on the set policies, but also on the people who are expected to comply with those policies. Previous research shows that people sometimes do not decide and behave in line with ethical guidelines (including their own). A part of that discrepancy lies in the partial reflective awareness, which reveals the importance of its examination from the perspective of business and policy sustainability.

1.3. Research Gap

The unawareness or partial reflective awareness may cause the discrepancies in ethical decision-making and behavior, as “people are often unaware of the heuristics and environmental structures that guide their moral behavior” [36] and “people typically cannot explain why they feel that something is morally right or wrong, or why they did what they did” [63]. Besides the judgment-action gap (what individuals think vs. do), there are other questions that remain unanswered:

• What individuals are thinking versus report they are thinking (reflective awareness);
• Is there a difference in (1) and are there consistencies in this difference and could relate to context characteristics [29] and arousal [31];
• What is the relationship between moral judgment and reflective awareness, namely, the differences in reported and observed decisions and behavior, and their reasons;
• How revealed regularities can be translated into nudges.

If the focus of the analysis shifts from the outcome to the process that precedes it, the examination might reveal the reasons and patterns in which ethics-related decisions deviate from environmentally given principles and from one’s moral code.

The rest of the paper will focus on the above-stated questions. First, it is assumed that the ethical decision-making is not made randomly and there is a pattern or a structure of decision-making principles. The pattern reveals the underlying rules for ethical choices, based on the individual’s assessment and judgment. The regularities revealed will be examined in relation to Kohlberg’s categories and Jones’s item characteristics. It is also assumed that in the process of assessment and judgment in making an ethical choice, an individual will use satisficing and therefore use only a part of the information. In addition, ethical metacognition or reflective awareness would be confirmed if reported and used information is alike, while the occurrence of the differences would reveal the bounded ethical metacognition or partial reflective awareness.

The rest of the paper is structured as follows. The Materials and Methods section provides sub-chapters: research goal and hypothesis, explanation for methodological choices, data collection and data analysis. The Results section is divided into subchapters according to the main research questions, where the findings are presented and interpreted. The Discussion section emphasizes the findings relevance from the theoretical and practical perspective and their implications. The Conclusion section summarizes the presented research indicating limitations and possibilities for further research.

2. Materials and Methods

This research is a part of the greater research project for which a comprehensive survey on determinants of choices and behavior in business ethics (with an extended questionnaire and multiple samples) was conducted. However, here will be reported only the part relating to the phenomenon examined in this paper. The chronological order of research methodology is presented in Figure 1 and explained in detail in the following sub-chapters.

2.1. Research Goal and Hypotheses

The research goal is to combine cognitive and behavioral streams of research that overlap in the area of the awareness to answer what, how and why students choose when faced with ethical dilemmas. The specific goal is to examine the students’ reflective awareness when answering how and why they make choices in ethical dilemmas.

Based on the theoretical background, we deliver following set of propositions:

• there is an underlying set of ethical rules/guidelines for decision-making;
• individuals are not completely aware of this set of rules and report a partially different set of rules;
• (1) and (2) indicate occurrence of partial reflective awareness.

Building on the propositions, it is expected that the partial reflective awareness will be susceptible to:

• the ambiguity and complexity of the ethical dilemmas;
• the arousal (implied by the characteristics of the ethical dilemmas);
• the level of moral cognitive development;
• the moral philosophy.

Following the propositions and expectations, we form the following set of the hypotheses:

Hypotheses 1 (H1).

Statistically significant statements reveal underlying set of ethical rules/guidelines for decision-making.

Hypotheses 2 (H2).

There will be a difference between the set of statistically significant statements and the set of self-reported statements, indicating an occurrence of partial reflective awareness.

Hypotheses 3 (H3).

The sets of statistically significant and the set of self-reported statements will reveal the difference in the moral cognitive development.

Hypotheses 4 (H4).

The sets of statistically significant and the set of self-reported statements will reveal the difference in the applied moral philosophy.

Hypotheses 5 (H5).

Different sets of statistically significant statements and self-reported statements, as well as related indicators of moral cognitive development and applied moral philosophy will differ given the ambiguity and complexity of the ethical dilemmas, and the arousal (implied by the characteristics of the ethical dilemmas).

2.2. The Choice of Scenario Technique and the Sample

One could assume that ethical reasoning automatically discards some of the alternatives, but practice and the theoretical overview show that ethical questions are rarely black and white and require additional analysis. If we assess a choice assuming the underlying function (or algorithm) of an individual’s decision-making, it is expected to observe that individuals ignore certain alternatives or attributes, thus satisficing and not maximizing utility, using only a sub-set of alternatives. The use of the satisficing would mean that there are patterns in the decision-making process, which might not be aligned with a rational one.

We use a common approach in business ethics research, a scenario technique. The scenarios used (Appendix A) represent ethically ambiguous situations without a universally “correct” choice. The scenarios are based on Kohlberg’s scenarios, which are artificial and hypothetical [27]. The offered choices refer to action (should do something), lack of action (should not do something) and inability to decide (cannot decide). The scenarios are not developed to elicit any specific heuristic but consider “large (enough) worlds” (and are considered relatable to student population) to enable examination of thought patterns and the influence of the context.

The methodology combines theoretical approaches, in a way that we use scenarios that are originally assembled from the moral judgment and moral philosophy point of view and examine what one should do, but in quantification and modelling we focus on examining the individuals’ evaluation of the relevant features/statements that led them to the decision. That is, we focus on the decision-making patterns, but also consider the outcome. Such an approach combines moral judgment and bounded ethicality, creating a step towards bridging the two approaches.

There is a main question for each scenario (what the agent should do), twelve accompanying statements describing different aspects of the decision (to which respondents had to assign a degree of importance for making the decision). The statements refer to different ethical aspects of approaching the situation: statements are questions that examine a certain aspect; the respondents do not answer the questions but assign numerical value that denotes its perceived relevance to the decision-making problem. Each scenario is followed by a question to rank the statements according to their importance for the respondent’s decision.

Given that both statement evaluations occur after the decision, even from the aspect of the post hoc justification bias, they should not reveal the difference (as both statement relevance evaluations “justify” the choice made). This means that the possible differences in the statement relevance assessment would not result from the post hoc justification bias, but from the respondents’ ability to identify relevant statements for the decision made.

The statements differ based on Kohlberg’s categories and moral philosophy. The scenarios are developed in a way to disable the anonymity bias and choices are stated for a third person (respondents freely assess the choice without feeling judged and are free of action bias—as they are not the ones who finally take the action).

The questionnaire and the sample are chosen based on the assumption of the transferable decision-making skills, a common assumption in the research of business ethics.

The limitations of the methodology refer to the lack of the presence of peers (so some heuristics cannot be revealed); the scenarios are hypothetical and not conducted in natural environment; and the information in the scenarios is framed in a certain manner. In addition, the influence of the social environment can only be implicitly deduced by comparing the results to previously conducted researches in different socioeconomic environments. Also, the role of emotions and arousal is implicitly induced from the context characteristics.

2.3. Data Collection

The data were collected from the sample of 200 students of second and third year of undergraduate university students of business economics, humanities and music studies in Pula, Croatia. The survey was conducted in 2016. The study was also used as a class exercise. Most of the students (57.5%) studied business economics at the time and now some of them are already a part of the work force (54% completed their undergraduate studies and 21% completed graduate studies), belonging to its youngest (Z) generation. The average age in the sample was 24.02 years, and the majority of the respondents were women (72%, which is in line with the percentage of the female students enrolled in these courses and reflects the percentage of women employed in administration and jobs related to business economics in Croatia).

2.4. Data Analysis

Quantification and modeling examine decision-making patterns and information use, as well as the differences in observed and reported reasons for a decision. That is, we do not assume that specific bias governs the decision-making process in ethical dilemmas but examine whether individuals use all available information (bounded rationality/awareness); and whether they can successfully identify upon which information they made the decision (partial reflective awareness).

It is assumed that omitting statements to a subset indicates that individuals ignore certain attributes and use bounded rationality/awareness (decide upon attributes they find relevant) in providing an ethical choice. We will examine the alternatives which the respondents find relevant for the decision about business ethic cases.

First, we will report what the respondents chose when faced with ethical dilemmas (frequencies of all responses, in Section 3.1), while the rest of the analysis will strive to reveal how they chose an alternative and why (without excluded responses), while the approach to analysis will be the detection of patterns.

Choices for each case are recorded as 1 for an “ethical” decision and 0 for a “non-ethical” decision. The questionnaire offered a “cannot decide” option and all those answers are excluded from the analysis. Incomplete questionnaires are also excluded from further analysis. The analysis is conducted based on the sample of 145 responses for the first scenario, 121 for the second and 149 for the third scenario.

To determine how (based on which statements) the respondents made a choice in ethical dilemmas, we analyze the choices per each scenario using binomial logit regression (reported in Section 3.2). The model outcomes are structured as follows:

$${\mathrm{O}}_{\mathrm{i}} = \left\{\begin{array}{c}1 \mathrm{if} \mathrm{the} \mathrm{i}-\mathrm{t} \mathrm{person} \mathrm{made} \mathrm{a} \mathrm{choice} \mathrm{of} \mathrm{ethical} \mathrm{decision}\\ 0 \mathrm{otherwise} \hfill \end{array}, \mathrm{i}=1, \dots , 8\right..$$

(1)

The hypotheses for each model are set as:

$$\begin{gathered} H_0 \dots p_i={\widehat{p}}_{O_i}\left(O_i=1\right) \\ {H}_1 \dots p_i\ne {\widehat{p}}_{O_i}\left(O_i=0\right). \end{gathered}$$

(2)

Based on a standard probability threshold, the hypothesis interpretation is as follows: if

$$\widehat{p}\ge \tilde{p}, then O_i=1, else O_i=0.$$

(3)

The logit equation can be stated as:

$$O_i^{\prime }=ln\left(\frac{p}{1-p}\right)={\beta }_1{s}_1+{\beta }_2{s}_2+{\beta }_3{s}_3+{\beta }_4{s}_4+{\beta }_5{s}_5+{\beta }_6{s}_6+{\beta }_7{s}_7+{\beta }_8{s}_8+{\beta }_9{s}_9+{\beta }_{10}{s}_{10}+{\beta }_{11}{s}_{11}+{\beta }_{12}{s}_{12}$$

(4)

where s = {HNZ, PRS, NWP} are statements per each scenario;

Or, in terms of probability, it is:

$$\widehat{p}=\frac{e^{{\beta }_1{s}_1+{\beta }_2{s}_2+{\beta }_3{s}_3+{\beta }_4{s}_4+{\beta }_5{s}_5+{\beta }_6{s}_6+{\beta }_7{s}_7+{\beta }_8{s}_8+{\beta }_9{s}_9+{\beta }_{10}{s}_{10}+{\beta }_{11}{s}_{11}+{\beta }_{12}{s}_{12}}}{e^{{\beta }_1{s}_1+{\beta }_2{s}_2+{\beta }_3{s}_3+{\beta }_4{s}_4+{\beta }_5{s}_5+{\beta }_6{s}_6+{\beta }_7{s}_7+{\beta }_8{s}_8+{\beta }_9{s}_9+{\beta }_{10}{s}_{10}+{\beta }_{11}{s}_{11}+{\beta }_{12}{s}_{12}}+1}=\frac{e^{O_i^{\text{'}}}}{e^{O_i^{\text{'}}}+1}$$

(5)

Afterwards, we will analyze the students’ responses about significant statements for a certain decision (last question, Appendix A). The analysis consists of modeling the choice of ethical dilemma in 3 business ethic cases using auxiliary statements to determine a set of relevant influences for the stated choices (to determine whether students use all available information or its subset, based on the numerical values assigned to the perceived relevance). The self-reported most important statements for ethical decision-making are analyzed in such a way that for each scenario those statements were ranked by the frequency within chosen priority (first, second, etc., most important statement). For each priority level, most frequently (mode) self-reported important statement is reported and quantified using the average of assigned numerical value of its relevance. The comparison of the statistically significant statements with the self-reported important statements in decision-making will reveal whether students are aware of their own principles in ethical decision-making (reported in Section 3.3).

The in-depth analysis of the comparison of the model-derived and self-reported statements will shed a light on why respondents chose a certain option. An additional in-depth qualitative comparison will build on the previous analysis and will involve the moral judgment approach, as each statement corresponds to a certain moral philosophy approach, individual/social aspect of moral philosophy and a level of Kohlberg’s categories (reported in the Section 3.4).

3. Results

3.1. What Did the Respondents Choose?

For the three scenarios, the respondents chose what the involved person should or should not do in each situation. The scenarios allow a different approach given the moral philosophy and the individual versus social aspect. The scenarios are complex enough, and each choice could be justified to appear ethical from a certain perspective. Also, each scenario contains a specific combination of the items’ characteristics [29].

While there is a general rule that says we should not steal, in the situation where stealing saves a life, most of the respondents chose stealing (

Table 1. Respondents’ choices.

Heinz Scenario			Prisoner Scenario			Newspaper Scenario
	Frequency	Percent		Frequency	Percent		Frequency	Percent
Missing	6	3.0	Missing	4	2.0	Missing	9	4.5
Should steal	92	46.0	Should report him	67	33.5	Should stop it	36	18.0
Cannot decide	43	21.5	Cannot decide	75	37.5	Cannot decide	41	20.5
Should not steal	59	29.5	Should not report him	54	27.0	Should not stop it	114	57.0
Total	200	100.0	Total	200	100.0	Total	200	100.0

Source: Authors.

). This scenario observes a close interpersonal relationship of the decision-maker to the wife (high intensity proximity) and the drug maker (low intensity proximity) who will suffer the consequences of his choice. High intensity of the temporal immediacy and probability of effect refers to the certain loss of (wife’s) life in the near future in case of inactivity. While there is high personal intensity of concentration of effect in case of saving the wife’s life, there is also low intensity of concentration of effect considering the loss for the drug maker. If the decision is observed in a narrower sense, the general rule of not stealing should be applied (social consensus). Considering the context and the consequences (magnitude of consequences), it might be interpreted that the lack of stealing implicitly leads to the loss of a life and the person involved chooses between the guilt or punishment for stealing and a loss of a life. In such interpretation, stealing is a more ethical choice.

Most respondents could not decide on the choice in the second scenario (Table 1), where an escaped prisoner did not pay his due to society in full, but, later in life, he is a kind and functional member of society—which creates moderate concentration of effect. The magnitude of consequences is moderately intense, so is the probability of effect (due to uncertainty of the act itself), and the temporal immediacy. In this scenario, a friendship/familiarity with the prisoner exists, so there is a moderately strong interpersonal bond (proximity). Social consensus implies the necessity to report an escaped prisoner. However, the decision about returning a prisoner to prison (considering both his escape and later life) is not up to an individual but the court—so in this context—an ethical choice would be to report the prisoner (and perhaps afterwards testify on his behalf). The choice in this situation can be additionally complicated if there is no trust in the legal system (which might be an explanation for such a high number of indecisive respondents (In terms of perceived independence of courts and judges among the general public, judiciary in Croatia scored worst among the EU countries (the 2019 EU justice scoreboard, https://ec.europa.eu/info/sites/info/files/justice_scoreboard_2019_en.pdf (accessed on 15 October 2020)); and the corruption risk of its judicial system is assessed as high, https://www.ganintegrity.com/portal/country-profiles/croatia/ (accessed on 15 October 2020). However, this also represents the possibility of using the information possibly unavailable to the modeler [64], meaning that the respondents use the information other than provided in the scenarios (for example, from previous experience).

Most respondents chose the lack of action in the third scenario (Table 1), where an involved individual could choose to stop or not to stop the newspaper, where the conflict emerges between the rights to free speech and information, and the welfare of the school. In this situation, the decision-maker’s choice will have consequences for the students (school users) and related stakeholders, indicating a combination of business and weaker social relationship to the agents who will deal with consequences (low proximity intensity). While declarative social consensus guarantees the right to free speech and information, the “local”—scenario social consensus represented by two affected groups (the parents and some of the students) calls for stopping the newspaper. Due to the uncertainty of the consequences, there is a low probability of effect. However, temporal immediacy is high, given that the consequences (such as they may be) should occur immediately. The magnitude of consequences and the concentration of effect are low due to the existence of the two opposed groups of (not stated but presumably) similar size, where each would suffer approximately equally significant consequences (from own perspective). The principal is the one who must make a decision, and while it is important not to hurt someone’s feelings and to enable discussion, his main obligation is the welfare of the school. In such interpretation, the principal should stop the newspaper.

The first and the third scenario have an approximately equal number of indecisive respondents, while the second one has the biggest number of such respondents. Also, the percentage of the choices of presumably ethical (or less ethically debatable) choices diminishes over scenarios. While 46% chooses a lesser evil in the first scenario, 33.5% of the respondents choose to report the prisoner and only 18% choose to stop the newspaper. The key might be in the intensity of characteristics (especially proximity, which is clearly distinguished over the scenarios) and the ambiguity (context complexity), as the first diminishes over the scenarios and the second one increases.

3.2. How Did the Respondents Choose?

The resulting binomial logit regression models of ethical decisions are presented in

Table 2. Binomial logit models of ethical decisions for scenarios based on statements.

Heinz Scenario	Coefficient	p-Value	Prisoner Scenario	Coefficient	p-Value	Newspaper Scenario	Coefficient	p-Value
const	0.6177	0.7244	const	4.9664	0.0198 **	const	3.2298	0.001 ***
HNZ1	−0.6178	0.0059 ***	PRS1	−2.306	<0.0001 ***	NWP6	−0.4854	0.0113 **
HNZ2	1.1198	0.0001 ***	PRS2	1.4035	0.0002 ***	NWP9	−0.678	0.0005 ***
HNZ6	−0.6171	0.006 ***	PRS12	−0.5471	0.0665 *
HNZ8	−0.6786	0.0057 ***
HNZ9	−0.3975	0.0337 **
HNZ12	0.7531	0.0003 ***
Mean dependent var	0.6138		0.5537			0.2416
McFadden R-squared	0.3793		0.6267			0.1463
S.D. dependent var	0.4886		0.4992			0.4295
Adjusted R-squared	0.3069		0.5786			0.1099
Log-likelihood	−60.0339		−31.052			−70.336
Schwarz criterion	154.905		81.2874			155.684
Akaike criterion	134.0679		70.1043			146.672
Hannan-Quinn	142.5347		74.6462			150.333
Number of cases ‘correctly predicted’	115 (79.3%)		107 (88.4%)			112 (75.2%)
f(beta’x) at mean of independent vars	0.489		0.499			0.43
LRT: Chi-square (12)	73.3682 [0.0000]		104.238 [0.0000]			24.101 [0.0000]

Notes: models are tested by applying sequential elimination of variables using two-sided p-value, α = 0.1. Statistical significance levels of 10%, 5% and 1% are denoted. *, **, ***, respectively. Source: Authors’ calculation using Gretl.

. The models show moderate prediction levels (75.2–88.4%) and the likelihood ratio tests point to a good model fit. Statistically significant statements are the best predictors of the respondents’ choices in the observed dilemmas.

The model of the choice in the first scenario (Heinz) points to 79.3% of correctly predicted cases based on 6 statements: HNZ1, HNZ2, HNZ6, HNZ8, HNZ9 and HNZ12 (H1). The statements HNZ2 and HNZ12 have a positive sign and lead to a presumably ethical decision. The other statistically relevant statements lead to an ethically more ambiguous decision.

The model of the choice in the second scenario (Prisoner) results in 88.4% of correctly predicted cases based on the constant and 3 statements: PRS1, PRS2, and PRS12 (H1). The statistically significant constant points to default bias towards the presumably ethical decision. However, more expressed relevance of the statements PRS1 and PRS12 will lead to the ethically debatable choice. This also confirms the ethical ambiguity of the situation and the relevance of the context and circumstances (in this case—whether the prisoner has redeemed himself and whether his confinement benefits the society). More expressed relevance of the statement PRS2 leads to the ethical decision (considering consequences: punishment avoidance as the encouragement for the crime).

The third model of the Newspaper scenario results in 75.2% of correctly predicted cases based on the constant and two statements: NWP6, NWP9 (confirming the H1 hypothesis). The constant in this model points to default bias towards the presumably ethical decision, while statistically significant statements (considering the importance of discussing important problems and students’ education in critical thinking and judgment) point to the influences that lead to ethically more ambiguous decisions.

Given the number of statistically significant variables per scenario (HNZ: 6, PRS: 3, NWP: 2), it seems that the respondents take more statements (aspects of the situation) into consideration when considering a situation with a scenario that involves higher intensity characteristics, which is in line with [29]. Higher intensity of proximity also indicates greater personal involvement and therefore a higher level of arousal [31], which enables recognition of the ethical dilemma, which could explain the observed regularity from a different point of view (H5).

Given that not all of the statements appear statistically significant, it means that not all of the variances in the statements’ relevance appear significant for explaining variances in an ethical choice, and given the prediction levels, there are also unobserved influences that shape the choice. This means that the respondents used a subset of statements to make the decision, but also consider information outside the given scenario (possibly a previous experience, ethical norms, etc.). This indicates a bounded rationality/awareness in ethical decision-making in the observed cases. However, this proves that individuals do not use all available information while making an ethical choice, which has already been proven by previous research [40].

3.3. Are the Respondents Aware of Their Own Ethical Decision-Making Guidelines?

In order to examine whether the respondents can successfully identify which information they based their decision on (reflect on how they made a choice), it is necessary to examine the similarities between the model-derived subset of statements and self-reported most important statements for making the decision in each scenario. If the most important self-reported statements coincide with the statistically significant variables in the models, it means that the respondents are aware of their own decision-making rules/guidelines; otherwise, inconsistencies between the self-reported relevant statements and the statistically significant statements reveal the respondents’ unawareness or partial reflective awareness of the rules applied to the ethical dilemma in the scenarios (metacognition).

Table 3. Self-reported most important statements for ethical decision-making.

	Mean (Average Relevance)	Standard Deviation
Heinz most important: HNZ2	4.39	0.98
Heinz second most important: HNZ3	4.13	1.061
Heinz third most important: HNZ10	3.69	1.09
Heinz fourth most important: HNZ12	3.57	1.376
Prisoner most important: PRS1	1.89	0.998
Prisoner second most important: PRS4	2.78	1.498
Prisoner third most important: PRS8	2.55	1.266
Prisoner fourth most important: PRS10	2.82	1.44
Newspaper most important: NWP1	2.52	0.857
Newspaper second most important: NWP6	3.61	1.131
Newspaper third most important: NWP9	2.77	1.399
Newspaper fourth most important: NWP10	3.85	1.058

Source: Authors’ calculation using Gretl.

shows the list of the four most important statements in decision-making, as reported by the respondents. The statements identified as the four most important ones are the most frequently reported statements (mode) with average assigned relevance and standard deviations associated to the relevance of those statements.

The most frequent self-reported relevant statement for decision-making in the Heinz scenario dilemma is HNZ2, the second one is HNZ3, followed by HNZ10 and HNZ12. The model-derived statements (statistically significant statements based on the model in Table 2, hereinafter model-derived) for the first scenario choice are: HNZ1, HNZ2, HNZ6, HNZ8, HNZ9, HNZ12. We can notice that the self-reported statements HNZ2 and HNZ12 match the model-derived variables (a third of the statements), while others differ (H2). The values of average relevance diminish over the statement importance, along with the increase in the standard deviation. This is expected. Given the comparison, it can be concluded that the respondents demonstrated partial reflective awareness while facing the first ethical dilemma.

The most frequent self-reported relevant statement for decision-making in the Prisoner scenario dilemma is PRS1, followed by PRS4, PRS8, and PRS10. The model-derived statements are: PRS1, PRS2, and PRS12. The match of the self-reported statements and the model-derived variables occurs in PRS1 (a third of the statements, in line with H2), while others differ. The values of average relevance of the statements do not correspond to the reported order of their importance: it is expected that the most important statement has the highest value of average relevance, and that the values diminish as statements become less important. The results suggest that the respondents show a lower level of partial reflective awareness.

The most frequent self-reported relevant statement for decision-making in the Newspaper scenario dilemma is NWP1, followed by NWP6, NWP9, and NWP10. The model-derived statements for the choice in this scenario are NWP6 and NWP9. In this situation, we have complete congruence between the model-derived variables and a half of the self-reported variables, so that both model-derived variables occur as the most frequently self-reported second and the third most important statements for decision-making in the Newspaper ethical dilemma. Other self-reported relevant statements, NWP1 and NWP10, are not a part of the choice prediction model (confirming H2). In terms of the average relevance values, the situation is similar to the Prisoner scenario. The respondents show different levels of partial reflective awareness, which among different subsets of the statements indicate the relevance of the context in the choices made in ethical dilemmas but also their role and the strength of the relationship to the reasoning.

If we observe the regularities in terms of reflective awareness in the scenarios given the intensity of characteristics—especially proximity, it seems that the respondents demonstrate a higher level of partial reflectiveness as the intensity of characteristics diminishes. This observation can implicitly be related to arousal and emotional involvement. While arousal might be a key factor for the awareness in ethical dilemmas, it seems that more intense emotional involvement leads to lower partial reflective awareness in terms of the respondents’ ability to successfully identify the information (statements) based on which the decision has been made (which is an additional confirmation of H5).

3.4. Why Did the Respondents Make a Certain Choice?

The revealed regularities require additional, in-depth analysis that will connect the results regarding the partial reflective awareness (behavioral ethics) to the moral judgment approach. It can be assumed that the respondents made the choices based on their set of ethical guidelines (which they are at least partially aware of), which can be related to the statements that accompany scenarios. Each statement corresponds to a certain moral philosophy approach, individual/social aspect of moral philosophy and a level of Kohlberg’s categories.

Table 4. Comparison of self-reported and model-derived important statements in ethical decision-making.

Scenario	Properties	Statements						Median/Frequency
Heinz scenario	Model-derived	HNZ1	HNZ2	HNZ6	HNZ8	HNZ9	HNZ12
	Kohlberg’s categories	4	3	4	6	5a	5a	4.5
	Moral philosophy	Moral Rights	Moral Rights (Values)	Moral Rights	Moral Rights (Values)	Justice	Utilitarianism	Moral Rights (66.66%), Justice (16.67%), Utilitarianism (16.67%)
	Individual/social aspect of moral philosophy	Social	Individual	Individual	Social	Social	Social	Social (66.67%), Individual (33.33%)
	Self-reported	HNZ2	HNZ3	HNZ10	HNZ12
	Kohlberg’s categories	3	2	5a	5a			4
	Moral philosophy	Moral Rights (Values)	Justice (Values)	Moral Rights	Utilitarianism			Moral Rights (50%), Justice (25%), Utilitarianism (25%)
	Individual/social aspect of moral philosophy	Individual	Individual	Social	Social			Individual (50%), Social (50%)
Prisoner scenario	Model-derived	PRS1	PRS2	PRS12
	Kohlberg’s categories	3	4	5a				4
	Moral philosophy	Justice	Utilitarianism	Utilitarianism				Utilitarianism (66.67%), Justice (33.33%)
	Individual/social aspect of moral philosophy	Individual	Social	Social				Social (66.67%), Individual (33.33%)
	Self-reported	PRS1	PRS4	PRS8	PRS10
	Kohlberg’s categories	3	4	4	4			4
	Moral philosophy	Justice	Justice	Moral Rights	Justice (Value)			Justice (75%), Moral Rights (25%)
	Individual/social aspect of moral philosophy	Individual	Individual	Social	Social			Individual (50%), Social (50%)
Newspaper scenario	Model-derived	NWP6	NWP9
	Kohlberg’s categories	5a	5b					5
	Moral philosophy	Moral Rights	Utilitarianism					Moral Rights (50%), Utilitarianism (50%)
	Individual/social aspect of moral philosophy	Social	Social					Social (100%)
	Self-reported	NWP1	NWP6	NWP9	NWP10
	Kohlberg’s categories	4	5a	5b	5a			5
	Moral philosophy	Utilitarianism	Moral Rights	Utilitarianism	Moral Rights			Moral Rights (50%), Utilitarianism (50%)
	Individual/social aspect of moral philosophy	Social	Social	Social	Social			Social (100%)

Source: Authors.

shows the comparison of the model-derived and self-reported statements for each scenario. A score per model/scenario is shown in the last column using median values and frequencies, which enables comparison of the applied and reported statements from the perspective of moral judgment.

The median of Kohlberg’s categories does not differ by much (meaning that even partial reflective awareness results in approximately equal Kohlberg’s category). Based on this assessment, it cannot be said that the respondents use the reported statements to appear “more ethical” than they are (at least, not successfully). Also, it is expected that the participants in this age group belong to the highest category (sixth). The results can be explained by Rest’s conclusions [21], who finds that individuals do not necessarily progress linearly through the stages—there is a possibility of stagnation or regression to previous stages. When observing the assigned Kohlberg’s categories for the Heinz and Prisoner scenario, it can be noticed that the model-derived variables point to higher level category statements, in comparison to the self-reported statements (confirming H3 only for the first scenario, forcing the rejection of the hypothesis). However, the applied moral philosophy differs between the model-derived and self-reported important statements, which is in line with H4. Moral philosophy and its individual versus social aspect show qualitative similarity but differ in their quantitative expression of relevance. The difference between the model-derived and self-reported decision-making variables regarding moral philosophy is the highest in the Prisoner scenario and the lowest in the Newspaper scenario. The differences in observed and reported relevant statements might result from the observed, but unexplained variances in the variables. However, a more plausible explanation is that individuals use a certain set of rules while deciding on ethical dilemmas but report a slightly different set of rules.

The results in Table 4 also reveal that the respondents apply a different moral philosophy approach as well as the individual versus social aspect of the moral philosophy while exhibiting approximately the same level of cognitive moral development, and they do so based on the context and circumstances (otherwise, we would observe constant frequencies assigned to moral philosophy and its aspects over the scenarios, which is an additional support for H5). The closer relationship ties (as parts of the scenarios’ context) also imply the existence of (more intense) emotions, where “increased level of cognitive activity [as a consequence of arousal] would likely result in an individual accessing more cognitive moral structures, thereby increasing the likelihood that an individual has accessed a cognitive moral structure consistent with their level of moral development” [31]. The context causes the shifts in the moral philosophy approaches and the reflective awareness of the used information diminishes with the scenario complexity and ambiguity (as in the second scenario). This means that:

• there is an underlying set of rules/guidelines for decision-making;
• individuals are not completely aware of this set of rules and report a partially different set of rules;
• (1) and (2) indicate occurrence of bounded rationality and partial reflective awareness (or bounded ethical metacognition).

The partial reflective awareness (or bounded ethical metacognition) examination reveals the following regularities:

• reflective awareness diminishes as ambiguity and complexity of the scenarios increase;
• partial reflective awareness of own level of moral cognitive development decreases with the characteristics and arousal intensity;
• reflective awareness about own moral philosophy diminishes with the characteristics and arousal intensity.

4. Discussion

As expected, the results show that there is an underlying set of ethical guidelines present at ethical dilemma decision-making. In answering the questions of what individuals were thinking versus report they were thinking, why there is a difference, and whether there are consistencies in this difference, we argued that these three scenarios have two properties in common: intensity of characteristics and ambiguity. The intensity of characteristics (especially proximity) is clearly distinguished over the scenarios (husband and wife-high intensity proximity, friend and prisoner-moderately strong interpersonal bond (proximity), principal of the school and students with related stakeholders-low proximity intensity). The ambiguity (context complexity) refers to how many different stakeholders are involved in the decision-making process and what impact they will exert given their interest in a particular decision-making entity: wife-drug maker, prisoner-court, school principal-students, parents, local community, school reputation, etc.

The results show that the respondents consider more aspects when analyzing a situation with a scenario involving higher intensity characteristics. Higher proximity intensity indicates a higher level of involvement, and a higher level of arousal. Furthermore, the respondents used a subset of statements for decision-making, but also considered information outside the given scenario, which indicates bounded rationality/awareness in decision-making. The applied moral philosophy differs between the model-derived and self-reported important statements and vary over scenarios. That means that the respondents apply different moral philosophy approaches as well as the individual versus social aspect of the moral philosophy while exhibiting approximately the same level of cognitive moral development, and they do so based on the context and circumstances—which is in line with the previous research [29].

If we observe those regularities in terms of reflective awareness given the intensity of characteristics over the scenarios—especially proximity—it seems that the respondents demonstrate a higher level of partial reflective awareness as the intensity of characteristics diminishes. This observation can implicitly be related to arousal and emotional involvement. While arousal might be a key factor for the awareness of ethical dilemmas (previously observed by Gaudine and Thorne [31]), it seems that more intense emotional involvement leads to lower partial reflective awareness in terms of the respondents’ ability to successfully identify the information (statements) based on which the decision has been made. The context complexity does not only make it harder to make an ethical decision but also affects reflective awareness in the process.

The visualization of hypothesized and revealed regularities in the partial reflective awareness occurrence and its relationship to examined individual attributes and situational features is presented in Figure 2.

If we observe Figure 2 from the sustainability aspect, it can be deduced that such decision-making situations present a difficult situation, especially in the developing country. First, there is no social consensus on the implementation of the guidelines for sustainable development in the companies, besides that it is a favorable buzzword. Second, sustainability requires consideration of what is the right thing to do, along with examination of the long-term consequences of the decision, both by its magnitudes and probabilities of effect—which creates a high context complexity. This may also be in conflict with the temporal immediacy that sometimes forces fast decision-making. Lastly, the decision-maker will pay much more attention to the decision if it involves close people, environments or objects (proximity). The ethics and sustainability are inseparable [62], so it is expected that the difficulties in decision-making are similar. Thus, it is relevant to develop incentives to encourage higher sensitivity to the corporate sustainability topics with the aim of achieving higher level of partial reflective awareness and to improve decision-making.

The relevance of the revealed (in)consistencies has implications for the theoretical research and modeling of repetitive decision-making and decision-making with variations in the context, as well as methodological aspects of research in ethics. The repetitive ethical decision-making based on the skewed set of guidelines (in comparison to the ones people thought they used) may be connected to—or even the cause of—various cognitive and behavioral biases. While a part of the discrepancies in the patterns may be contextually incentivized, the other part arises from the individual’s mental processes. Most of the contextually induced differences in the decisions seem almost self-explanatory, but what is puzzling is the intrinsic part of the decision-making process. It makes sense that individuals consider different sets of guidelines given a different context. However, the analysis reveals a parallel set of guidelines within a decision-making process: one set that has been used and the other that has been reported (the one they thought they used). Briefly, it might seem as a post hoc justification, but given that both evaluations occurred after the decision and the respondents answered the questions individually and anonymously, there was no need for justification. The observed and reported results indicate partial unawareness of own ethical decision-making guidelines, namely, partial reflective awareness or bounded ethical metacognition.

However, at this point, we cannot pinpoint the source of the revealed fallacy and additional neuroscience approach in future research might help with that. While partial reflective awareness can be a result of unawareness and information perception and utilization process in a relation to the context characteristics and related level of emotional involvement (and we assume as much based on the previous research by [29,31]), it might also indicate a possibility of the existence of biases: extension of post hoc justification bias (the respondents justify the decision to themselves, self-justification) or a form of motivated reflective blindness (respondents miss the information contradictory to their self-image [46]).

The partial reflective awareness has implications for research in ethics based on the self-reported data, as it demonstrates the discrepancy between the information that was used and the information that was reported for making a decision. In a way, this may be an explanation for the source of action-judgment gap. While people make judgments (or actions, for that matter) they are not fully aware of their own guiding principles. We see that students have a problem in accurately identifying information upon which the choice has been made. It is possible that the same thing happens with the actions, and we can assume that it will transpire in different amounts, but at this point, we cannot suggest the degree of the discrepancy. It is worth to further examine whether the actions are better indicators of the used set of ethical guidelines (vs. reported ones, but keeping in mind the other imperfections in the human decision-making process). This calls for additional questions and instruments, with the intention of revealing and generalizing discrepancies in underlying decision-making patterns and/or guidelines versus self-declared ones. This also indicates that different methodologies necessarily lead to different results, in this case—confirming both consequentialist and behaviorist findings as the two sides of the same coin.

Finally, the revealed partial reflective awareness in the observed sample might be related to the transition of the mentality according to the transitional stage of the country and generally set ethical guidelines, indicating that individuals might apply certain ethical guidelines before they become aware of them. However, this research reveals that there are inconsistencies that indicate partial reflective awareness, while possible reasons for these inconsistencies can be investigated in future research.

The practical implications comprehend the relation to reflective skills and their properties: they can be improved and developed through education [24]. That means that individuals can, with the practice, become more aware of their own decision-making guidelines, but also be aware of the deviations from those guidelines. If individuals become aware of their own deviations, they may put additional effort into shaping their future decision-making in a desired direction. In this context, improvement in reflective awareness would contribute to the better, more sustainable [54,55,56,57,58,59,60,61,62] ethical choices and reveal a possibility of an intervention design within the higher education framework. In addressing the question of how revealed regularities can be translated into nudges, we conclude that it is necessary to use additional questions and instruments in order to reveal underlying patterns and/or guidelines for decision-making in relation to those who self-identify themselves. However, the implementation of the narratives [59], writing about the experiences of ethical dilemmas, raising awareness of own process of ethical decision-making, interpreting and understanding ethical decision-making and the exchange and receipt of the feedback may be implemented in the higher education system as an intervention design. Such design is in line with sustainable education [24], and in this case, sustainable business ethics education. The stated approach benefits the reflective skills improvement [24,59], which impacts not only ethical decision-making, but also plays a role in becoming a virtuous person [23]. Implementation of such an approach could help the Croatian economy to move away from bad practices that rank Croatia low in terms of competitiveness: judicial independence, efficiency of legal framework in challenging regulations, transparency, critical thinking, etc. The fact that some Croatian undergraduate university students think about ethical business choices is encouraging for more intensive ethical deliberation in higher education in the future. That will hopefully raise the bar of moral reasoning and potentially resolve the problems in Croatian society, such as corruption, fraud, etc. Social environment gives people meaningful context and may play an important role in the decision-making process; therefore, it should be changed and morally upgraded through the educational system.

5. Conclusions

The complexity and ambiguity of the context not only makes it difficult to make ethical decisions, but also affects the reflective consciousness in the process. It also indicates that different methodologies necessarily lead to different results, in this case, confirming both consequentialist findings and behavioral findings as the two sides of the same coin.

The limitations of this research build on the already declared limitations of using Kohlberg’s scenarios, which basically means that it does not examine the moral action, just underlying principles of moral judgment. In addition, the sample is a convenience sample of students, which limits the generalization of the findings.

Our methodology combines theoretical approaches, in a way that we use scenarios that were originally used from the moral judgment point of view and examine what one should do, but in quantification and modelling we focus on examining the individuals’ evaluation of the relevant statements that led them to the decision. In terms of the implications, it is assumed that ethical decision-making principles are transferable and improvable. The link between the ethics education and business ethicality is both intuitively sound and often empirically supported. However, the challenges confronting business schools in producing businesspeople trained to make ethical decisions are time and again subject to further research given the many influences affecting success. While students come with a myriad of dependent and independent influences on business ethics education topics, we consider them to be a very suitable sample for presenting new evidence on this research topic, even in different countries. Our research results prove that a bias in decision-making exists and indicates the occurrence of bounded rationality/awareness, namely partial reflective awareness.

In a broader sense, our results support the research thesis that organizations should not rely on the decision-making rules of employees to encourage ethical decision-making, but build a system of organizational values. Namely, the organizations that do not have clearly defined organizational values are exposed to the risk that decisions are made by the individuals who are not fully aware of the set of rules under which they make them, which may further result in negative outcomes. Additionally, introducing reflective awareness as a skill that can be improved in both classrooms and boardrooms may serve as an intervention towards better decision-making aiming for sustainability at the personal and organizational levels.

The possibilities for different courses of research development in the future are various, whether by using different type scenarios or by observing behavior in experimental setting. Also, it is advised to examine the existence of partial reflective awareness in other decision-making contexts. It would also be interesting to extend the research to other countries to compare what ethical decisions, how and why the graduates make. Besides the use of different methods and examination of different contexts, a research extension should encompass a probability sample of general population. Such researches would create a possibility of findings comparison and conceivably offer a general explanation or description of the partial reflective awareness’ role in the decision-making. Additional research extension should also comprise the “action approach” in the business ethics, with the aim to contribute to action-judgment gap closure. Finally, as previously suggested in the Discussion section, we believe that the research topic would benefit from multidisciplinary approach by inclusion of neuroscience insights.

The contribution of this paper is in the empirical example of examining the assessment and moral judgment in the decision-making process in ethical dilemmas. The obtained research results confirm the approach of bounded rationality. The results also indicate the existence of partial awareness of internal decision-making processes, which further calls into question the assumption of reflectivity, which is the basis for some research methodologies. The authors find a possible explanation for the deviation in the motivated self-blindness, i.e., the difference in declarative and actual guidelines for decision-making. In this sense, we have made a “deflection” from previous research, which mainly deals with the influence of behavior on the decision-making process in ethical dilemmas.