Guardians of Good: What the Research Says Makes an Organisation's Ethics Work.

Jul 23, 2024

45 min read

Updated: Sep 28

It is 4.52pm on the last day of the compliance window, and the annual ethics module is open in one tab while the board pack is open in the other. Ten scenarios, forty minutes if you read them, twelve if you do not. A colleague accepts a hamper from a supplier; a manager asks for a figure to be "rounded up a bit"; somebody overhears something in a lift. The right answer is nearly always the one that says to speak to your line manager, and by 5.04pm the certificate is in your inbox. Upstairs, meanwhile, the quarterly forecast meeting is setting next year's target at a number everyone in the room privately thinks is a stretch too far, and you are the one taking the minutes.

Both of those things are the organisation's ethics. Only one of them has a certificate. I first wrote about this in 2024, from the desk of a biotech Chief of Staff, arguing that ethics has to be built in before the trouble rather than bolted on after it, and forty years of research agree, with a great deal to say about what building it in looks like on an ordinary Tuesday. This post is about the codes that change behaviour, the managers who carry the tone, the systems people trust to raise a concern, and the targets that pull decent people off course. It is also about you, because the person who sees an organisation's decisions up close, before they are tidied into a press release, is very often the person who manages the diary of the one making them.

The certificate says you passed. The compass is for Tuesday.

What you will get out of this

What the research says turns a code of conduct from a document into behaviour, including the study where a code on its own explained very little and five things around it explained a great deal more
The nine parts of an ethics programme, the one that made no difference, the one that made the most, and the order the research suggests building them in
Why the tone in the middle matters as much as the tone at the top, and the study of 2,572 people across three levels that complicated the tidy version
The eight measurable dimensions of an ethical culture, and the two that surprisingly did not predict behaviour
What stops people raising concerns, what the analysis of nearly two million internal reports found, and why a small favour can keep someone quiet
How good people drift, and the targets and goals that speed the drift up, including a real disagreement between two camps of researchers
Whether ethics training works, with numbers, and the kind that works best
What the person who sits beside the decision can do about all of it
The ethics check: a tool that reads your team's ethical infrastructure and tells you where it is strong and where it is decoration
The Ethics Pages: three pages for a decision in front of you, a quick audit of your team's ethical culture, and a review of the targets you set

The code on the wall and the code in the room

Nearly every organisation of any size has a code of conduct, and nearly every code has the same career. It is launched with a video from the chief executive, signed by everyone during onboarding, attached to the annual module, and consulted afterwards mainly by the people drafting the next version. The question the research asks is a fair one: does having it change anything?

The honest answer from the literature is that it depends almost entirely on what happens around it. A critical review of the studies on business codes found their results pointing in different directions, and put much of that down to differing definitions, thin data and a shortage of theory. A later review of 100 empirical papers published between 2005 and 2016 found the content of codes was still predominantly self-defensive, written to protect the organisation rather than to guide the person reading it, and that codes on their own were insufficient to protect workers' rights.

The most useful single study separated the code from the things around it. It looked at five factors beyond the code's existence: what the code actually said, how often it was communicated, how good that communication was, and how far it was embedded by senior management and by local management. The full model explained 32% of the unethical behaviour people observed at work. The explanatory value of having a code on its own was, in the author's word, very modest. The same study found conditions in which a code became counter-effective, which is a polite way of saying that a code nobody lives by can make things worse.

That fits an older finding from a survey of alumni now working in business. Self-reported unethical behaviour was lower among people whose organisations had a code, and it fell further the more strongly the code was implemented and the more embedded it was in daily work. And a field survey that compared organisations with and without codes found that where a code existed, the part of the ethical context most strongly linked to observed misconduct was a culture-based one, while in organisations without a code, climate mattered more. The code sets the words; the culture decides whether anyone believes them.

A code of conduct is the only document in the building that everyone has signed and almost nobody has opened since. What changes behaviour is the conversation that happens around it.

There is a humbling footnote about the little behavioural tricks that often ride along with codes. A widely adopted finding said that asking people to sign an honesty statement at the top of a form, rather than the bottom, made them more honest. Five conceptual replications with 4,559 people and one large preregistered direct replication with 1,235 people, run with the original authors, found no effect at all. The lesson is not that nudges never work; it is that a signature is not the same as a conversation.

The writers who first made the case for organisational ethics reached the same place from the other side: ethics rooted in values rather than bolted to procedures, real questions sitting in shades of grey, and ethics as a standard that heads off misconduct rather than explaining it afterwards. Rules can say where the line is. Only people, talking about real cases, can help each other find it in the fog.

A graphic titled What turns a code into conduct, with the eyebrow From a study of five things around a code of conduct, and what they explained. Standfirst: On its own, having a code explained very little of the misconduct people saw at work. Add the five things around it and the model explained 32%. Here they are, in the order most organisations forget them. Five rows. One, What it actually says, Content: A code written to guide the reader, not to defend the organisation. A review of 100 studies found most codes still mainly self-defensive. Write it for the person with the dilemma. Two, How often it comes up, Frequency of communication: A code mentioned once at onboarding and once a year in a module is a code people forget exists. Bring it into ordinary conversations, several times a year. Three, How well it comes up, Quality of communication: Talking about real cases does more than restating principles. Use one grey-area case from your own recent past. Four, Senior leaders live it, Embedment by senior management: People watch what happens to the rules when the numbers are short. Make the leaders' own use of the code visible. Five, Local managers live it, Embedment by local management: The manager people see every day carries the code further than any launch video. Equip the middle to talk about it, not only to forward it. Foot: Kaptein (2011a); Babri, Davidson and Helin (2021); McCabe, Treviño and Butterfield (1996). — A code of conduct is like a gym membership: owning one is not the part that works.

Try this. Next time the code of conduct comes up, whether in onboarding, in the annual refresh or in a team meeting, bring one real grey-area case from the last few months, with the names taken out, and ask the room what the code says about it and what they would actually do. Ten minutes on one case will do more than the whole module, because it is the conversation the research found doing the work.

The research behind this section

Kaptein and Schwartz (2008) critically examined existing studies of the effectiveness of business codes and found conflicting results, which they attributed to varying definitions of key terms, deficiencies in empirical data and methods, and a lack of theory, and proposed an integrated research model. Babri, Davidson and Helin (2021) reviewed 100 empirical papers on corporate codes of ethics from mid-2005 to mid-2016 and found that code content was still predominantly self-defensive, that codes were insufficient in themselves for protecting workers' rights, that they met tensions when implemented across national and organisational boundaries, and that while perceptions of codes were generally positive, their effects could run both ways. Kaptein (2011a) studied five factors besides the existence of a code, namely its content, the frequency and quality of communication activities around it, and its embedment by senior and by local management, and found the full model explained 32% of observed unethical behaviour while the explanatory value of a code alone was very modest, and identified conditions in which codes became counter-effective. McCabe, Treviño and Butterfield (1996) surveyed alumni of an honour code college and a non-honour code college now working in business and found that self-reported unethical behaviour was lower where there was a corporate code of conduct and was inversely associated with the code's implementation strength and embeddedness. Treviño, Butterfield and McCabe (1998) compared ethical climate and ethical culture in a field survey and found that an ethical culture-based dimension was more strongly associated with observed unethical conduct in organisations with codes, while climate-based dimensions were more strongly associated with it in organisations without codes. Kristal and colleagues (2020) ran five conceptual replications with 4,559 participants and a highly powered, preregistered direct replication with 1,235 participants, conducted with the authors of the original paper, and observed no effect of signing a veracity statement at the beginning rather than the end of a form on honest reporting. Schnebel and Bienert (2004) set out how ethics can be implemented in business organisations. Bruhn (2009) examined the functionality of grey-area ethics in organisations. Bishop (2013) argued for the role of ethics as a standard in twenty-first century organisations. Ciulla (2004) brought together philosophical and organisational perspectives on ethics at the heart of leadership.

Infrastructure, or a very expensive facade

Researchers call everything an organisation builds to support ethical behaviour its ethical infrastructure, and it has three layers: the formal systems (code, training, reporting line, investigations, sanctions), the informal systems (what people say in corridors about what really gets rewarded), and the climate around both. The argument, made in 2003, is that the formal layer only holds if it is built on the other two, hence the lovely title of the paper: building houses on rocks.

What happens without the rock was described by a case study of widespread deceptive sales practices at a large financial services firm. Its compliance programme was, in the researchers' term, decoupled from the core business: it existed on paper and in audits while the sales floor ran on different rules. It did something worse than fail. It created a legitimacy facade, a respectable front that let the misconduct become institutionalised behind it, until the facade fell and took the firm's external legitimacy with it.

Two studies from 1999 explain how organisations end up there. External pressure from regulators encouraged programmes that were easily decoupled, and it was management's own commitment that decided a programme's character: whether it was built to control people through compliance or to support them through values.

That character matters. A survey of employees at six large American companies asked what works and what hurts in ethics and compliance management, and the answer was unusually clear-cut. A values-based, cultural approach worked best, with leaders visibly committed to ethics, fair treatment of employees, rewards for ethical conduct, concern for people outside the organisation, and consistency between what the policies said and what actually happened. What hurt most was a programme employees believed existed to protect top management from blame, and a culture built on unquestioning obedience to authority and on self-interest. Where the approach worked, people were more willing to deliver bad news and report violations, and less unethical behaviour followed. A companion study in a large financial services firm found a values orientation linked to seven outcomes and a compliance orientation to four, with the two together linked to people's willingness to report misconduct.

There is a warning from the other side too: programmes built on coercive control can weaken people's own moral judgement, especially in situations the rules did not foresee. The authors called it an atrophy of competence. If every ethical question has been turned into a rule, people stop practising the skill of answering one.

So which parts should an organisation actually have? The most useful study here examined nine components of ethics programmes. Organisations with a programme saw less unethical behaviour than those without, and the more components an organisation had adopted, the less misconduct people observed. One component, pre-employment screening, had no relationship with unethical behaviour at all. The strongest relationship was with accountability policies, the rules that hold people answerable for their conduct. And the study suggested an order: a code first, then training and communication, then accountability, then monitoring and auditing, then investigation and correction, all directly linked to less misconduct, followed by an ethics office, a reporting line and incentive policies, which worked indirectly.

A graphic titled Building an ethics programme, in order, with the eyebrow From a study of nine parts of an ethics programme and the misconduct people observed. Standfirst: The more of these an organisation had, the less misconduct people saw. The first five were linked to it directly and the last three worked through the others. The ninth, screening people before hiring, made no difference at all. Eight numbered steps, each with a tag saying whether its link to misconduct was direct, indirect or the strongest. One, A code of ethics, tagged Direct: The starting point, as long as it is written to guide rather than to defend. Two, Training and communication, tagged Direct: Real cases, discussed, several times a year. Three, Accountability policies, tagged Strongest: The strongest link of all nine: people answerable for how they behave. Four, Monitoring and auditing, tagged Direct: Checking that what happens matches what the policies say. Five, Investigation and correction, tagged Direct: What happens after a concern is raised, and whether people hear about it. Six, An ethics office or officer, tagged Indirect: Somebody whose job it is, and whom people know by name. Seven, A reporting line, tagged Indirect: A route for concerns that people trust enough to use. Eight, Incentive policies, tagged Indirect: Rewards that do not quietly pay for the opposite of what the code asks. Foot: Kaptein (2015); Park and Blenkinsopp (2013). — Eight parts in order, and one that did nothing at all. Sorry, pre-employment screening.

Here good studies disagree. A national survey of public service organisations found the effect of ethics programmes on misconduct was entirely carried by ethical culture: only two of six programme elements mattered before culture was taken into account, and none did after. Studies of 202 teams in a UK business, and of 770 teams across one organisation and its sister organisations abroad, found the same shape, with a programme's effect running through each team's own culture. My reading is that the findings fit together. The components are the scaffolding. The culture is the building, and it can differ floor by floor.

Try this. Pick one component your organisation says it has, such as the reporting line or the investigations process, and find out what happened to the last three things that went through it, in general terms. If nobody can tell you, or the honest answer is "nothing much", you have found the place where the formal layer has come loose from the informal one.

The research behind this section

Tenbrunsel, Smith-Crowe and Umphress (2003) set out the role of the ethical infrastructure in organisations, made up of formal systems, informal systems and the climates surrounding them. MacLean and Behnam (2010) used a case study of widespread deceptive sales practices at a large financial services firm to show how decoupling a compliance programme from core business activities created a legitimacy facade that enabled the institutionalisation of misconduct and precipitated a loss of external legitimacy. Weaver, Treviño and Cochran (1999a) found from survey and archival data that external pressures encouraged easily decoupled ethics programmes while top management commitment encouraged both decoupled and integrated ones. Weaver, Treviño and Cochran (1999b) found that environmental factors were the stronger influence on an ethics programme's scope, while management's ethical commitment was the stronger determinant of its orientation towards compliance-based or values-based control. Treviño and colleagues (1999) surveyed employees at six large American companies and found that a values-based cultural approach worked best, with leaders' commitment to ethics, fair treatment of employees, rewards for ethical conduct, concern for external stakeholders and consistency between policies and actions as critical ingredients, that what hurt most was a programme employees believed existed to protect top management from blame and a culture of unquestioning obedience and self-interest, and that effective management was associated with less unethical and illegal behaviour, more ethical advice seeking and greater willingness to deliver bad news or report violations. Weaver and Treviño (1999) found in a large financial services company that perceptions of a values orientation were associated with seven outcomes and a compliance orientation with four, that the interaction of the two was associated with willingness to report misconduct, and that a values orientation made the greater unique contribution. Stansbury and Barry (2007) analysed ethics programmes as control systems and argued that coercive control may weaken employees' ability and motivation to exercise their own moral judgement, identifying indoctrination, a politicisation of ethics and an atrophy of competence as potential drawbacks. Kaptein (2015) found that unethical behaviour occurred less frequently in organisations with an ethics program, that the number of the nine components adopted was directly related to the frequency of observed unethical behaviour, that pre-employment screening had no relationship with it, that accountability policies had the strongest relationship, and proposed a sequence of adoption beginning with a code, then training and communication, accountability policies, monitoring and auditing, and investigation and correction, followed by an ethics office, a report line and incentive policies. Park and Blenkinsopp (2013) used a national ethics survey of Korean public service organisations and found that the relationship between ethics programmes and misconduct was fully mediated by ethical culture, with only two of six elements significant before culture was controlled for and none after, while the programme appeared to strengthen the culture. Cabana and Kaptein (2025) surveyed 202 teams in a UK business and found that team ethical culture mediated the effect of a well-implemented ethics program on unethical behaviour in teams. Cabana and Kaptein (2026) studied 180 teams in one organisation and 590 teams in its sister organisations across several countries and found that the pathways from specific program components to unethical behaviour ran through specific dimensions of team ethical culture and varied among the sister organisations.

The tone at the top, and the tone in the middle

"Tone at the top" is the phrase every governance report reaches for, and while it is not wrong, it is incomplete in a way that matters a great deal to anyone who works close to leaders.

Interviews with senior executives and ethics officers found that a reputation for ethical leadership rests on two pillars: being seen as a moral person, honest and fair in how decisions get made, and being seen as a moral manager, who role models visibly, uses the reward system and talks about ethics out loud. Leaders with the first but not the second were called ethically neutral: perhaps entirely decent, but nobody below them could tell. A follow-up study added that to be seen as ethical from outside the executive suite, a leader had to do things that were socially salient, that stood out against an otherwise neutral background.

That is the first reason the top is not enough: most people never see the top closely. The second comes from the research on levels. A study of 904 employees and 195 managers across 195 departments found that ethical leadership from both top management and immediate supervisors was linked to less deviance and more helpful behaviour in teams, and that the effect of the top ran through the supervisors. Ethical leadership trickled down, and it trickled down through the middle. A study in a bank found what carried it across: middle supervisors' confidence that they could act ethically, and their expectation that unethical behaviour would be punished, though not their expectation that good behaviour would be rewarded.

Then the tidy version got complicated. A study of 2,572 soldiers across three levels of command on combat deployment found only limited support for a simple trickle-down. What it found instead was broader: leaders at every level shaped the ethical culture of their own units, those cultures cascaded down and across the organisation, and it was the culture that most shaped how people further down thought and behaved. And a study of where leaders' own ethical leadership came from found that having had an ethical role model during their career made a difference, while having ethical role models in top management made no detectable difference at all.

People do not learn ethics from the chief executive's video. They learn it from their own manager's face when the numbers are short.

There is an honest counterweight. A meta-analysis of 101 samples and 29,620 people found ethical leadership linked to followers' attitudes and performance, though sometimes only weakly beyond other leadership styles. A three-wave study of 214 employees found that followers only treated an ethical leader as a role model when they themselves cared about morality and identified with the leader; among those who did neither, ethical leadership went with more unethical behaviour, not less. What a leader can more reliably do, according to a field study of 394 employees and 68 supervisors and an experiment with 121 working professionals, is shape what people believe their role requires of them. When leaders acted ethically and spoke up about ethical issues, their people came to see ethical conduct as part of the job, and that belief was what reduced unethical behaviour.

That is a relief for anyone who is not a chief executive, because the tone that matters most is set in the middle. A Chief of Staff who asks "how will this land with the people it affects?" in a leadership meeting, or an assistant who says "I would like to check that before it goes" about a figure that looks too convenient, is setting tone, and the research says tone set close to people travels further than tone set from a stage.

Try this. Make ethics socially salient once this week, in a small way, in front of other people. Thank someone publicly for raising a concern, or ask out loud in a meeting what the option under discussion would look like from the customer's side. The research on ethically neutral leaders says the good intentions most people already have only count once somebody can see them.

The research behind this section

Brown, Treviño and Harrison (2005) defined ethical leadership from a social learning perspective and developed a measure of it. Brown and Treviño (2006) reviewed the ethical leadership literature and set out future directions. Treviño, Hartman and Brown (2000) interviewed senior executives and corporate ethics officers and found that a reputation for executive ethical leadership rested on visibility as a moral person and as a moral manager, contrasting the ethical leader with the unethical, the hypocritical and the ethically neutral leader. Treviño, Brown and Hartman (2003) found in an inductive interview study that ethical leadership included an overlooked transactional component involving communication and the reward system, and that to be perceived as ethical by those outside the executive suite an executive had to engage in socially salient behaviours. Mayer and colleagues (2009) studied 904 employees and 195 managers in 195 departments and found that top management and supervisory ethical leadership were both negatively related to group-level deviance and positively related to organisational citizenship behaviour, with the effects of top management mediated by supervisory ethical leadership. Ruiz, Ruiz and Martínez (2011) found a partial trickle-down effect of ethical leadership from top manager to immediate supervisor on follower job responses. Wang, Xu and Liu (2018) studied 69 middle-level supervisors and 381 subordinates in 69 sub-branches of a bank and found that supervisors' ethical efficacy expectation and unethical behaviour-punishment expectation accounted for the trickle-down effect while the ethical behaviour-reward expectation did not. Schaubroeck and colleagues (2012) studied 2,572 US Army soldiers across three organisational levels deployed in combat and found limited support for simple trickle-down mechanisms but broader support for a multilevel model in which leaders embedded shared understandings through the ethical culture of units, which cascaded across levels and influenced followers' ethical cognitions and behaviour. Brown and Treviño (2014) found that having had an ethical role model during the leader's career was positively related to subordinate-rated ethical leadership and found no effect for top management ethical role models. Ng and Feldman (2015) meta-analysed 101 samples with 29,620 people and found that ethical leadership showed acceptable validity for followers' job attitudes, performance and evaluations of leaders, with trust in the leader as a mediator, and that it predicted task performance, citizenship and counterproductive behaviour beyond other leadership variables, albeit weakly in some cases. Wang and colleagues (2021) used three-wave data from 214 employees and found that ethical leaders were viewed as role models only among followers higher in moral identity and leader identification, and that ethical leadership evoked unethical behaviour among followers lower in both. Paterson and Huang (2019) studied 394 employees and 68 supervisors in the field and 121 working professionals in an experiment and found that leaders' behaviours and ethical voice influenced role ethicality, which in turn affected unethical behaviour. Lemoine, Hartnell and Leroy (2019) reviewed moral approaches to leadership and distinguished the moral content of ethical, authentic and servant leadership. Harden Fritz, Arnett and Conkel (1999) examined organisational ethical standards and organisational commitment.

The culture you can measure

"Culture" gets used as a word for everything that cannot be put in a spreadsheet, which is a shame, because the ethical kind has been measured carefully for more than thirty years.

The first strand is ethical climate: the shared sense of what counts as right. The foundational survey of 872 employees in four firms found five kinds, in which people look respectively to laws and professional codes, to care for those affected, to self-interest, to their own principles, or to the organisation's rules, and found climate varied within organisations by position, length of service and workgroup. A meta-analysis in 2006 confirmed that climate perceptions relate to how people feel and behave at work, and later reviews have extended the list of outcomes to include bullying, commitment, ethical behaviour, satisfaction and intention to leave.

The second strand is ethical culture, and here the measure worth knowing is the Corporate Ethical Virtues model, built and tested across four studies into a 58-item questionnaire with eight dimensions. In plain English, they are: how clear the standards are; whether supervisors live them; whether senior management lives them; whether people have the time, resources and information to act ethically; whether people are committed to acting ethically; how visible behaviour is, good and bad; how openly ethical issues can be discussed; and whether ethical behaviour is rewarded and unethical behaviour sanctioned.

When the model was tested against behaviour, in 341 sets of a manager and two direct reports, six of the eight dimensions were linked to less observed unethical behaviour: role modelling by management, role modelling by supervisors, capability, commitment, openness to discuss ethical issues, and reinforcement. Two were not. How clear the standards were, and how visible behaviour was, made no significant difference.

That runs against the instinct of most organisations, which is to answer a problem with a clearer rule or more oversight. The difference lay instead in whether people saw the standards lived, had the means to meet them, could talk about them, and saw them reinforced.

A review of 89 works on ethical culture recommended studying it in groups, not only whole organisations, and two studies found ethical climate carrying the link between ethical leadership and less misconduct in units.

Try this. Take the eight dimensions to your own team and score each one quietly from one to five, then ask one trusted colleague to do the same and compare. The gaps between your two sets of scores are often more useful than the scores themselves, because they show you where the culture looks different depending on where you sit.

The research behind this section

Victor and Cullen (1988) surveyed 872 employees of four firms and found that ethical work climates were multidimensional and multidetermined, identified five dimensions (law and code, caring, instrumentalism, independence and rules), and found significant differences in climate both across and within firms, including by position, tenure and workgroup membership. Martin and Cullen (2006) meta-analysed ethical climate theory research and found evidence of relationships between ethical climate perceptions and individual-level work outcomes. Simha and Cullen (2012) reviewed the effects of ethical climates on outcomes including workplace bullying, organisational commitment, ethical behaviour, job satisfaction and turnover intentions. Newman and colleagues (2017) reviewed a decade of research on the antecedents, outcomes and moderators of ethical climates. Kaptein (2008) validated the Corporate Ethical Virtues Model across four studies, finding eight subscales (how clear the standards are, congruency of supervisors, congruency of management, feasibility, supportability, transparency, discussability and sanctionability) in a 58-item questionnaire. Kaptein (2011b) studied 341 triads of a manager and two direct reports and found that six dimensions of ethical culture were negatively related to observed unethical behaviour (ethical role modelling of management and of supervisors, capability, commitment, openness to discuss ethical issues and reinforcement), while how clear the ethical standards were and the visibility of (un)ethical behaviour were not significantly related to it. Roy and colleagues (2024) reviewed eighty-nine works on ethical culture and recommended validated measures, longitudinal designs and group-level data. Kuenzi, Mayer and Greenbaum (2020) developed and validated a measure of ethical organisational climate and examined its role between ethical leadership and unit-level unethical behaviour, with collective moral identity as a boundary condition. Mayer, Kuenzi and Greenbaum (2010) examined ethical climate as the mediator between ethical leadership and employee misconduct. Kish-Gephart, Harrison and Treviño (2010) meta-analysed over thirty years of research on individual, moral issue and organisational environment antecedents of unethical choice and found it to be multidetermined, with several variables predicting unethical behaviour better than intention.

Speaking up, and whether anyone actually does

Every organisation says its door is open. The research on speaking up asks the more useful question, which is whether anyone walks through it.

Mostly, people do not. Researchers who named the pattern organisational silence treated it as something collective rather than a collection of timid individuals, and when they interviewed 40 employees, most had been in a situation where they were worried about something and did not raise it with a supervisor. The reason given most often was fear of being seen or labelled negatively, and of damaging relationships they valued. A study of 3,149 employees and 223 managers in a restaurant chain found that, of the two kinds of leadership it compared, the one more consistently linked to people speaking up was managerial openness, working through people's sense that it was psychologically safe to do so, and that it mattered most for the best performers.

What decides whether someone reports wrongdoing? A survey of nearly 8,600 employees in 15 organisations found people were more likely to blow the whistle when they had convincing evidence, when the wrongdoing was serious, and when it affected them directly, and more likely to go outside the organisation when it seemed to depend on the wrongdoing and threatened retaliation. A meta-analysis of 26 samples and 18,781 people found an important gap: personal, contextual and situational factors predicted people's intentions to report much better than they predicted actual reporting. And a field study of 1,416 employees in China, Germany and Russia found where the gap comes from. When people imagine reporting, they underestimate how much the threat of retaliation, and the organisation's compliance measures, will weigh on the real decision.

The fear is not imaginary: at a large US base, lack of support from others was linked to retaliation, and retaliation damaged the whistleblower's relationship with their supervisor. A set of three studies found that people reported unethical conduct internally when both their supervisor's ethical leadership and their coworkers' ethical behaviour supported it, and that the reason the combination worked was that it lowered fear of retaliation. The researchers titled the paper "It takes a village", and perceptions that reporting would be futile did not carry the effect. People were not mainly silent because they thought nothing would happen. They were silent because they thought something would happen to them.

Two findings about the social side of reporting are worth keeping close. In scenario studies, people who reported a peer were judged highly ethical and, in the same breath, unlikable. And a set of experiments and a multi-wave survey found that receiving even a small favour from someone made people less likely to report that person's wrongdoing internally, through a feeling of obligation. The effect held even with no prior relationship, and it was stronger when the favour had been asked for.

In a job that runs on favours, it is worth noticing which ones come with strings you did not see being tied.

Now the encouraging part, from the largest dataset in this whole literature. An analysis of nearly two million internal whistleblowing reports submitted to more than 1,000 publicly traded US companies found that firms whose systems were used more actively, with more reports, more detail and more of them reviewed by management, faced fewer government fines and material lawsuits, and smaller ones. A busy reporting line is a sign of health, not of trouble.

Culture moves this too: the ethical culture dimensions were linked to less inaction and more reporting to management and to the ethics hotline. A study of 3,076 US employees found that the more wrongdoing people saw, the less inclined they were to report it internally and the more inclined to go outside, a pattern softened where people believed management handled reports supportively. And values-based communication from the top shaped whether people intended to speak up, while accountability and protection from retaliation decided whether they did.

Try this. Find out, and then tell the people around you, exactly how a concern gets raised in your organisation and what happens next. Not the policy title; the actual route, the name of a person, and what they do with it. The gap between intending to speak and speaking is mostly made of uncertainty about what comes after, and you can shrink it with one well-placed paragraph.

The research behind this section

Morrison and Milliken (2000) described organisational silence as a barrier to change and development. Milliken, Morrison and Hewlin (2003) interviewed 40 employees and found that most had been concerned about an issue and not raised it with a supervisor, most often from fear of being viewed or labelled negatively and damaging valued relationships. Detert and Burris (2007) studied 3,149 employees and 223 managers in a restaurant chain and found that managerial openness was more consistently related to improvement-oriented voice than transformational leadership, mediated by psychological safety, with the strongest effects for the best-performing employees. Miceli and Near (1985) surveyed nearly 8,600 employees of 15 organisations and found observers more likely to blow the whistle with convincing evidence, serious wrongdoing and direct effects on themselves, and more likely to report externally where the organisation appeared dependent on the wrongdoing and threatened retaliation. Mesmer-Magnus and Viswesvaran (2005) meta-analysed 193 correlations from 26 samples (N = 18,781) and found stronger relationships between personal, contextual and wrongdoing characteristics and whistleblowing intentions than actual whistleblowing, with retaliation best predicted by contextual variables. Oelrich (2021) studied 1,416 employees in China, Germany and Russia and found that employees underestimated the effect of retaliation and of compliance measures in hypothetical compared with actual whistleblowing. Rehg and colleagues (2008) found at a large US base that lack of support from others was related to retaliation for both men and women, and that retaliation was negatively related to relationships with the supervisor. Mayer and colleagues (2013) found across three studies that internal whistleblowing depended on complementary ethical influence from supervisors and coworkers, mediated by fear of retaliation in two studies but not by perceptions of futility. Treviño and Victor (1992) found in two scenario studies that people who reported peers' unethical behaviour were evaluated as highly ethical and simultaneously as unlikable. Qiu, Mai and Ellis (2025) found in laboratory experiments and a multi-wave survey that receiving a small favour reduced internal reporting of the helper's wrongdoing through feelings of obligation, even without a prior relationship, and that the effect was stronger with positive reciprocity beliefs and when help had been solicited. Stubben and Welch (2020) analysed nearly two million internal whistleblowing reports to over 1,000 publicly traded US firms and found that report volume was associated with fewer and lower government fines and material lawsuits. Kaptein (2011c) found that several dimensions of ethical culture were negatively related to intended inaction and external whistleblowing and positively related to intended confrontation, reporting to management and calling an ethics hotline. Kaptein (2022) surveyed 3,076 US employees and found a curvilinear relationship between the frequency of observed unethical behaviour and whistleblowing intention, with a diminishing negative relationship for internal and a positive linear relationship for external whistleblowing, moderated by beliefs about management's supportiveness in handling reports. Vadera, Tenbrunsel and Diekmann (2025) found in a field study and an experiment that top management values-based communication affected whistleblowing intentions, while ethical accountability and retaliation systems moderated whether intentions became behaviour. Miceli and colleagues (2012) surveyed more than 3,000 organisational members in the US and found that proactive personality, less coworker invalidation, leverage in the situation and strength of evidence predicted whistleblowing, and that observing wrongdoing was associated with lower perceived organisational support and justice of reporting channels. Dungan, Young and Waytz (2019) used data from 42,020 federal employees and an online sample and found that concerns for fairness beyond the organisation predicted reporting while loyalty predicted not reporting, and that organisational factors such as educating employees about disclosure predicted the channel people used.

How good people drift

Most organisational wrongdoing is done by ordinary, considerate people rather than villains: people who, one reasonable step at a time, end up somewhere none of them would have chosen on day one. The research has names for the steps, and knowing them is useful precisely because they are so quiet.

The best-known framework is moral disengagement: a set of mental moves that switch off the part of us that would normally object. We relabel the act (it is "managing the message"), compare it with something worse (everyone rounds up), move the responsibility elsewhere (the target came from above), spread it thin (it was a committee decision), or play down the consequences for the people affected. The psychologist who set this out observed that many harmful things are carried out through networks of legitimate enterprises run by otherwise considerate people, helped along by what he called the disconnected subdivision of functions and the diffusion of responsibility. Nobody sees the whole; everyone does their part.

Measured as a tendency, moral disengagement predicted a whole range of unethical behaviour at work, including a decision to commit fraud and unethical conduct reported by supervisors and coworkers, over and above other known predictors. A meta-analysis found it linked to misconduct and intention to leave, working through lower moral awareness, weaker moral judgement and less anticipated guilt. It found something oddly hopeful too: people prone to moral disengagement still felt guilt and shame after the misconduct. The switch goes off before the act and comes back on afterwards, which is exactly why the moment before is the one to work on.

Ethical fading describes how self-deception lets the moral side of a choice fade from view, so that a decision about misleading a customer is experienced as a decision about a deadline. The normalisation of corruption describes how such practices settle into routines and get taught to newcomers as the way things are done here.

The people around us speed all this up or slow it down. A study of 187 employees in 35 groups across 20 organisations found that people's antisocial behaviour at work was shaped by the antisocial behaviour of their own coworkers. And good relationships and strong identification with the organisation can lead people to overlook the moral content of an act done on its behalf, precisely because the cause feels worthy.

A meta-analysis drawing on thirty years of research across individual characteristics ("bad apples"), the nature of the moral issue ("bad cases") and the organisational environment ("bad barrels") found that unethical choice is shaped by all three at once. It also found that several factors predicted actual behaviour better than they predicted intentions, which led the authors to suggest that much unethical behaviour is closer to an impulse than a calculation. That is the strongest argument for building ethics into the environment: impulses respond to surroundings far more than to a module completed in the spring.

Try this. Keep a quiet ear out for the language of drift in meetings you attend: the relabelling, the "everyone does it", the "it came from above", the "it was agreed". You do not need to challenge each one. Noticing it, and occasionally asking "if we described this plainly, what would we call it?", puts the moral side of the decision back on the table before it fades.

The research behind this section

Bandura (1999) set out the mechanisms of moral disengagement, including moral justification, sanitising language, advantageous comparison, displacement and diffusion of responsibility, disregarding or minimising harmful effects, and dehumanisation and attribution of blame, and observed that many inhumanities operate through legitimate enterprises run by otherwise considerate people, through the disconnected subdivision of functions and diffusion of responsibility. Moore and colleagues (2012) developed an adult measure of the propensity to morally disengage and found across four further studies that it predicted self-reported unethical behaviour, a decision to commit fraud, a self-serving workplace decision and supervisor- and coworker-reported unethical behaviour beyond other established predictors. Detert, Treviño and Sweitzer (2008) studied 307 students over three waves and found empathy and moral identity negatively related, and trait cynicism and chance locus of control positively related, to moral disengagement, which was positively related to unethical decision making. Ogunfowora and colleagues (2022) meta-analysed the antecedents, correlates and consequences of moral disengagement and found it associated with workplace misconduct and turnover intentions and negatively with citizenship and task performance, with its link to misconduct mediated by lower moral awareness, moral judgement and anticipated guilt, and positively related to guilt and shame after misconduct. Tenbrunsel and Messick (2004) described ethical fading and the role of self-deception in unethical behaviour. Ashforth and Anand (2003) described the normalisation of corruption in organisations. Robinson and O'Leary-Kelly (1998) studied 187 employees from 35 groups in 20 organisations and found that individuals' antisocial behaviour at work was shaped by the antisocial behaviour of their coworkers. Umphress and Bingham (2011) proposed that positive social exchange relationships and organisational identification may lead to unethical pro-organisational behaviour via neutralisation, the process by which the moral content of unethical actions is overlooked. Kish-Gephart, Harrison and Treviño (2010) found in a meta-analysis that unethical choice was multidetermined by individual, moral issue and organisational environment antecedents, and that several variables predicted behaviour better than intention, suggesting an "ethical impulse" perspective. Treviño, den Nieuwenboer and Kish-Gephart (2014) reviewed research on (un)ethical behaviour in organisations, including ethical infrastructures, interpersonal influences, individual differences and cognitive and affective processes.

Targets, incentives and goals gone wild

Almost fifty years ago a management scholar wrote a paper with a title that deserves to be on every planning template: On the folly of rewarding A, while hoping for B. His point was that organisations routinely reward one behaviour while sincerely hoping for another, and then act surprised when they get what they paid for. It remains the most practical sentence in business ethics.

Goals are the sharpest version of it. In a laboratory experiment, people with specific goals they had not met were more likely to overstate their performance than people told to do their best, with or without money attached, and the effect was strongest when people fell only a little short. Anyone who has watched a team two per cent below a quarterly number will recognise that finding.

Later work filled in why. When people worked towards high goals over several consecutive rounds, the effort depleted their self-control, and the depletion carried the rise in unethical behaviour; the more consecutive goals, the stronger the effect. In three field and laboratory studies, high performance goals led people to morally disengage, more so the more committed they were to the goal. A study of working adults found specific goals made people more willing to advocate unethical methods in a dilemma, though only those already inclined to justify their behaviour actually overstated their results. Performance pressure, the felt sense that you must do better or face consequences, raised people's need to protect themselves, stirred up anger and self-serving thinking, and those drove cheating. And a focus on the bottom line to the exclusion of everything else passed from supervisors to employees and was linked to employees undermining their colleagues. Seen in top management, that focus was a mixed blessing: better customer service, and more unethical conduct towards customers.

Here is a lively disagreement. In 2009 four researchers argued that goal setting's benefits had been overstated and its side effects, unethical behaviour among them, largely ignored, and that goals should carry a warning label like a prescription medicine. The two researchers most associated with goal-setting theory replied in the same journal that the critique relied on anecdote and unrepresentative citations, and that most of the pitfalls it named they had already published themselves. Both sides agree that specific, challenging goals raise performance. Where they differ is on how often the side effects arise and how prominent the warning should be.

The most useful study for anyone who sets or tracks goals sits in the middle of that argument. Across field and laboratory studies, outcome goals, the kind that specify a number to hit, put people into a defensive, prevention-focused state that raised unethical behaviour, more so the harder the goal. Learning goals, the kind that specify what to master or try, reduced both the defensive state and the unethical behaviour, without reducing performance, even when the goals were difficult. The goal did not need to be easier. It needed to be about something other than the number alone.

A graphic titled Setting targets without the temptation, with the eyebrow From the research on goal setting and unethical behaviour. Standfirst: Specific, stretching goals raise performance; the argument is about their side effects. On the left is the common version, and on the right what the studies suggest instead. Five pairs, each with the common version on the left and the alternative on the right. Instead of A number on its own: Pair it with a learning goal. Learning goals cut unethical behaviour without cutting performance, even when they were hard. Instead of Hit it or miss it: Avoid the cliff edge. Cheating was most likely among people who had fallen only a little short. Instead of Stretch, every quarter: Watch the run of consecutive stretch goals. Each round wore down self-control, and unethical behaviour rose with it. Instead of Whatever it takes: Say what it must not take. Pressure to perform raised anger and self-serving thinking, and both drove cheating. Instead of Only the bottom line: Name the other things that count. A bottom-line-only focus passed from managers to their teams. Foot: Welsh and colleagues (2019); Schweitzer, Ordóñez and Douma (2004); Welsh and Ordóñez (2014); Mitchell and colleagues (2018); Greenbaum, Mawritz and Eissa (2012). — Every target should come with a second question: and what are we willing not to do for it?

Try this. For any target you help set or track, ask three questions before it goes out. What would someone be tempted to cut to hit this? Is there a cliff edge where missing by a little feels the same as missing by a lot? And is there a learning goal sitting alongside the number? If you work on the leadership team's planning, those three questions are some of the most ethically useful minutes of the year.

The research behind this section

Kerr (1975) presented examples from society and organisations of reward systems that pay off for one behaviour while the rewarder hopes for another. Schweitzer, Ordóñez and Douma (2004) found in a laboratory experiment that people with unmet goals were more likely to engage in unethical behaviour than people attempting to do their best, both with and without economic incentives, and that the relationship was particularly strong when people fell only a little short of their goals. Welsh and Ordóñez (2014) found in a laboratory experiment with high, low, increasing, decreasing and do-your-best goals across multiple rounds that depletion mediated the relationship between goal structures and unethical behaviour, moderated by the number of consecutive goals. Welsh and colleagues (2020) found across three field and laboratory studies an indirect effect of high performance goals on unethical behaviour through state moral disengagement, with goal commitment as a boundary condition. Niven and Healy (2016) randomly assigned working participants to specific or vague goals and found that those with specific goals were more likely to advocate unethical methods in an ethical dilemma task, but that in an anagram task only those high in moral justification overstated their performance more. Barsky (2008) proposed a model linking goal attributes to unethical behaviour through ethical recognition and moral disengagement. Mitchell and colleagues (2018) found that performance pressure heightened self-protection needs, anger and self-serving cognitions, which motivated workplace cheating. Greenbaum, Mawritz and Eissa (2012) developed a four-item measure of bottom-line mentality and found in 113 focal-supervisor-coworker triads that supervisor bottom-line mentality was related to employee bottom-line mentality and, through it, to social undermining. Babalola and colleagues (2020) found in two field studies that perceptions of top management's bottom-line mentality led to both mental preoccupation with work, which reduced customer incivility and improved service performance, and self-interest cognitions, which instigated customer-directed unethical conduct. Ordóñez and colleagues (2009) argued that the benefits of goal setting had been overstated and its systematic harms, including narrow focus, distorted risk preferences, unethical behaviour, inhibited learning, corroded culture and reduced intrinsic motivation, largely ignored. Locke and Latham (2009) replied that the critique breached principles of good scholarship, drew on anecdotal evidence and unrepresentative citations, and that most of the pitfalls it cited had already been published by them. Welsh and colleagues (2019) found across field and laboratory studies that outcome goals stimulated a prevention focus that increased unethical behaviour, exacerbated by goal difficulty, while learning goals reduced prevention focus and unethical behaviour without decreasing performance, even when difficult.

Does ethics training work?

The annual module from the opening deserves a fair hearing, because the research on ethics training is more mixed, and more encouraging, than either its fans or its critics tend to admit.

The early meta-analyses were sobering. One pulled together 25 business ethics programmes and found a minimal overall effect on ethical perceptions, behaviour or awareness. Another, of 26 ethics programmes for scientists, found the overall effect modest. Both found that the design mattered a great deal: the programmes that worked better were run as stand-alone seminars rather than folded into other courses, used cases, were interactive, and let people practise making real decisions.

Then the picture improved. A later meta-analysis of 66 studies, 106 courses, 150 effect sizes and 10,069 participants found that ethics instruction produced sizable benefits and had improved considerably over the previous decade. A review of business ethics instruction found professional, focused, workshop-based programmes especially effective, while noting serious problems with how many programmes measured success. And a study that sorted 330 training programmes into types put numbers on the differences. Programmes that taught people a process for making professional decisions had an effect size of 0.50, and field-specific compliance training 0.46. General discussion of ethics managed 0.31, and training in following norms 0.37.

A graphic titled Which ethics training works best, with the eyebrow From 330 ethics training programmes sorted by type. Standfirst: The bars show how much each type of training changed people, as an effect size, where around 0.2 is small and 0.5 is medium. Teaching people how to make a decision beat talking about ethics in general. A bar chart labelled Effect of training, as an effect size (higher is better), on a scale of 0 to 0.6. How to make the decision, Professional decision processes: 0.50. The rules for your field, Field-specific compliance: 0.46. Following the norms, Norm adherence: 0.37. Talking about ethics, General discussion: 0.31. All four helped; the practical ones helped most. A separate meta-analysis found training where people were emotionally involved in real problems beat intellectual debate, and a bank's single session was still showing results two years later. Foot: Mulhearn and colleagues (2017); Katsarov and colleagues (2022); Warren, Gaspar and Laufer (2014). — Good news for the annual module: it works. Better news: something else works more.

A meta-analysis of 30 studies of integrity training for researchers went further: approaches where learners were emotionally involved in working out real problems were the most effective, and intellectual deliberation, long the gold standard, was significantly less so. That is very close to the experiential learning that business ethics teachers were arguing for twenty years ago: letting people meet dilemmas in a safe setting before they meet them for real.

And the study I would send to anyone who thinks training is merely cosmetic followed bank employees before and after the introduction of formal ethics training. Two years after a single session, the indicators of an ethical culture had improved and stayed improved: less observed unethical behaviour, stronger intentions to act ethically, more confidence in the organisation's handling of ethics. The rise in how much people valued the organisation's stated values faded after the second year, which the authors took to mean that it was not the values poster doing the work. A review of 92 studies sounded the right caution: what we know about workplace training is still fragile.

So the answer is yes, with conditions. The ten-scenario module in the last hour is close to the design the research likes least; a focused workshop built around real decisions from your own sector, where people make the call and feel its weight, is close to the one it likes most.

Try this. If you have any say over the next ethics session, whether you are organising the leadership offsite or the team away day, swap one slide of principles for one real dilemma from your own organisation's recent past, anonymised, and ask small groups to decide what they would do and why. It costs nothing and moves the session from the least effective type of training to the most.

The research behind this section

Waples and colleagues (2009) meta-analysed 25 business ethics instructional programmes and found minimal overall impact on ethical perceptions, behaviour or awareness, with criteria, content and method moderators. Antes and colleagues (2009) meta-analysed 26 ethics program evaluations in the sciences and found modest overall effectiveness, with more successful programmes run as separate seminars and case-based, interactive and allowing practice of real-world ethical decision-making. Watts and colleagues (2017) reviewed 66 empirical studies, 106 ethics courses, 150 effect sizes and 10,069 participants and found that ethics instruction resulted in sizable benefits and had improved considerably within the last decade. Medeiros and colleagues (2017) meta-analysed business ethics instruction and found professional, focused and workshop-based programmes especially effective, while identifying considerable problems with evaluation criteria. Mulhearn and colleagues (2017) clustered 330 ethics training programmes into eight approaches and found Professional Decision Processes Training (d = 0.50) and Field-Specific Compliance Training (d = 0.46) more effective than General Discussion Training (d = 0.31) and Norm Adherence Training (d = 0.37). Katsarov and colleagues (2022) meta-analysed 75 effect sizes from 30 studies of research integrity training and found experiential approaches in which learners were emotionally involved most effective, and primarily intellectual deliberation significantly less effective. Sims (2002) argued for experiential learning in business ethics teaching. Warren, Gaspar and Laufer (2014) studied bank employees before and after formal ethics training and found sustained positive effects on indicators of an ethical organisational culture two years after a single session, while a boost in espoused organisational values dissipated after the second year. Kreismann and Talaulicar (2021) reviewed 92 studies of stand-alone business ethics training programmes and found knowledge about their efficacy still fragile.

The view from the seat beside the decision

Remember how harm happens in legitimate organisations: through the disconnected subdivision of functions, where nobody sees the whole. An executive assistant or Chief of Staff is one of the very few people who does. You see the email that went to legal and the one that went to the board. You see the target in the forecast meeting and the conversation about how the team will hit it. You see what gets moved when the diary is squeezed, and what never gets moved. That view is exactly what disengagement relies on nobody having.

You also sit where the research says ethics is most often set: the middle. Much of what reaches people from a senior leader passes through you, in the framing of a message, the agenda of a meeting, the follow-up that says what was decided and why. Decent executives often fail to be seen as ethical because they never make it visible, and you are often the person who can.

The research also names the traps particular to your seat. Your working life runs on favours, given and received, and the study on favours found even a small one made people less willing to report the person who gave it. You are close to leaders you are loyal to, and among more than 42,000 federal employees, loyalty to the organisation went with not reporting. You see a great deal, and among 3,076 employees, the more wrongdoing people saw, the less inclined they were to report it internally. None of this makes anybody a bad person. It makes the seat beside the decision a place where good intentions need a little structure to survive.

What that structure can look like, in practice:

Put the question on the agenda, not only in your head. A standing line on the leadership agenda for significant decisions, "who is affected who is not in this room?", makes the ethical side socially salient without anyone having to raise a flag.

Minute the reasoning, not only the decision. "Agreed to proceed" and "agreed to proceed because the supplier confirmed the safety data on the 12th" are different records, and the second is the one that stops a decision drifting.

Know the route before you need it. Find out exactly how concerns are raised, who reads them, and what happens next, so that if you ever need it you are not also working that out under pressure.

Notice the favours. Not to refuse them, but to know which ones you would find awkward to set aside if something went wrong.

Keep your own record. A short, private note of what you saw and when, kept at the time, is the kind of convincing evidence the research found makes people more willing and more able to act.

The person who manages the diary sees the decision before it has been tidied into a press release. That is a view worth using well.

Try this. Choose one of the five structures above and put it in place this month. The agenda question is usually the easiest to start with, because it helps everyone in the room and asks nothing brave of you.

The research behind this section

Bandura (1999) observed that many harmful activities are carried out through legitimate enterprises by otherwise considerate people through the disconnected subdivision of functions and diffusion of responsibility. Mayer and colleagues (2009) found the effects of top management ethical leadership on group outcomes mediated by supervisory ethical leadership. Paterson and Huang (2019) found that leaders' behaviours and ethical voice shaped followers' perceptions of role ethicality, which affected unethical behaviour. Treviño, Hartman and Brown (2000) described the ethically neutral leader, who may be an ethical person but has not made ethics and values an explicit part of the leadership agenda. Treviño, Brown and Hartman (2003) found that to be perceived as ethical by those outside the executive suite, executives had to engage in socially salient behaviours. Qiu, Mai and Ellis (2025) found that receiving a small favour reduced internal reporting of the helper's wrongdoing through felt obligation. Dungan, Young and Waytz (2019) found among 42,020 federal employees that loyalty to one's organisation was associated with not reporting unethical behaviour. Kaptein (2022) found among 3,076 US employees that the frequency of observed unethical behaviour had a diminishing negative relationship with the intention to whistleblow internally. Miceli and Near (1985) found that convincing evidence of wrongdoing made observers more likely to blow the whistle.

The ethics check

Every finding in this post comes back to one practical question: in the part of the organisation you can see, is the ethics built in, or is it on the wall? Here is a tool for answering it.

Name your team or area, then answer ten questions about how the code is used, how leaders and managers talk about ethics, how concerns are raised and what happened to the last one, how targets are set, and what the training looks like. It will tell you where your ethical infrastructure is strong and where it may be decoration, what to do first, what to check afterwards, and it gives you a note to copy.

Where to start

If you take one thing from all of this, make it the conversation. Codes on their own explained very little; codes discussed, lived by managers and reinforced explained a great deal. Programmes worked when they strengthened the culture and failed when they became a facade. Training worked best when people wrestled with real cases. Across every part of this research, ethics lives in what people say to each other about real decisions, and it withers in documents nobody discusses.

After that, there are three moves worth making.

Bring one real grey-area case into a team conversation this quarter, and ask what the code says and what people would actually do. Look at one target you help set and give it a learning goal alongside the number. And find out, and share, exactly what happens when someone raises a concern, so that the route is known before anyone needs it.

I have put the three pages I would use into a download: a decision check for a significant decision in front of you, a quick audit of your team's ethical culture across the eight dimensions the research measures, and a target review for the goals you set or track.

A graphic titled Building the ethics in, on a page, with the eyebrow Five moves, the sentence each one needs, and the research behind it. Standfirst: Each of these fits inside work you are already doing. The right-hand column is what the evidence says the move buys you. A table of 5 moves, each with the sentence to say and the evidence behind it. The real case, Code, in conversation. Say: Here is one from last spring. What does the code say, and what would we do? Evidence: A code alone explained very little; the conversation and management around it explained far more. The empty chair, On the agenda. Say: Who is affected by this who is not in the room? Evidence: Leaders are seen as ethical when they make it socially salient, not when they keep it private. The second goal, Beside every target. Say: Alongside the number, what are we trying to learn or get better at? Evidence: Learning goals cut unethical behaviour without cutting performance. The known route, Before anyone needs it. Say: If you are worried about something, this is who to tell and what happens next. Evidence: People underestimate how much fear of retaliation weighs on a real decision to speak. The plain word, When the language drifts. Say: If we described this plainly, what would we call it? Evidence: Relabelling is one of the first moves of moral disengagement, and one of the easiest to spot. Foot: Kaptein (2011a); Treviño, Brown and Hartman (2003); Welsh and colleagues (2019); Oelrich (2021); Bandura (1999). — Five sentences that do more for an organisation's ethics than a laminated values poster.

The original case for proactive ethics was that it is far better to build the ark before the storm than to bail afterwards. The research has turned that into something more specific and more encouraging: the ark is built out of ordinary conversations, sensible targets, a route for concerns that people trust, and managers in the middle who make the good intentions visible. Those are things an organisation can do this year, and they are things you are already well placed to help with.

The part of this that shows up in how you are seen at work sits inside what the Connected Leader Assessment calls Judgement and Decision-Making: how you weigh options when the right answer is not obvious, how you spot the consequences others have not considered, and whether people trust the calls you help shape. It is the dimension that decides whether you are the person who processes decisions or the person who makes them better.

The assessment is free, takes about twelve minutes and gives you a read across all five dimensions. If you want to go further on this one, the Skill Sprints are short, focused programmes on a single dimension, and the Irreplaceable Audit is the place to see how much of your work already sits at the point where decisions are shaped, which is exactly where an organisation's ethics are won or lost.

Stay savvy.

Meg ✌️

Free tools

The ethics check, the interactive tool in this post, free
The Ethics Pages, the download in this post, free
The Connected Leader Assessment, twelve minutes, free, with a read on Judgement and Decision-Making, the dimension this post sits in

P.S. Want to Go Deeper on Ethics in Organisations?

If this resonated and you want to go further, here is the evidence the piece is built on. These are the papers that shaped my thinking on what makes an organisation's ethics hold in practice, from codes and culture to targets and speaking up, and they are well worth your time.

Every DOI below has been checked. Where I could not confirm a figure from the paper itself, I have left the figure out rather than guess.

📃 Ciulla (2004) Philosophers and leadership scholars on why ethics sits at the heart of leadership. https://doi.org/10.5040/9798216967286

📃 Schnebel and Bienert (2004) How ethics can be implemented in business organisations, rooted in their values. https://doi.org/10.1023/b:busi.0000039409.58757.a8

📃 Bruhn (2009) Why most ethical questions at work sit in shades of grey, and what that means for organisations. https://doi.org/10.1007/s10551-008-9994-7

📃 Bishop (2013) The case for ethics as a standard that forestalls misconduct in modern organisations. https://doi.org/10.1007/s10551-013-1618-1

📃 Harden Fritz, Arnett and Conkel (1999) Organisational ethical standards and how they relate to people's commitment. https://doi.org/10.1023/a:1005939325707

📃 Sims (2002) The case for experiential learning in teaching business ethics. https://doi.org/10.1023/a:1021107728568

📃 Tenbrunsel, Smith-Crowe and Umphress (2003) Ethical infrastructure: formal systems, informal systems and the climate around them. https://doi.org/10.1023/a:1025992813613

📃 Weaver, Treviño and Cochran (1999a) External pressure encouraged easily decoupled ethics programmes; top management commitment encouraged both kinds. https://doi.org/10.2307/256975

📃 Weaver, Treviño and Cochran (1999b) The environment set a programme's scope; management commitment set whether it leaned on compliance or values. https://doi.org/10.2307/256873

📃 Treviño and colleagues (1999) Six large companies. Values-based approaches worked; programmes seen as protecting top management hurt most. https://doi.org/10.2307/41165990

📃 Weaver and Treviño (1999) A values orientation was linked to seven outcomes, compliance to four; together they raised willingness to report. https://doi.org/10.2307/3857477

📃 MacLean and Behnam (2010) A compliance programme decoupled from the business became a legitimacy facade for misconduct. https://doi.org/10.5465/amj.2010.57319198

📃 Stansbury and Barry (2007) Ethics programmes built on coercive control may weaken people's own moral judgement. https://doi.org/10.5840/beq200717229

📃 Kaptein (2015) Nine programme components. More components, less misconduct; screening did nothing; accountability did most. https://doi.org/10.1007/s10551-014-2296-3

📃 Cabana and Kaptein (2026) 180 teams and 590 more in sister organisations. Programmes worked through each team's ethical culture. https://doi.org/10.1007/s10551-025-06115-5

📃 Kaptein and Schwartz (2008) Why studies of business codes disagree, and a model for studying them properly. https://doi.org/10.1007/s10551-006-9305-0

📃 Babri, Davidson and Helin (2021) Review of 100 papers. Code content was still mostly self-defensive, and codes alone were not enough. https://doi.org/10.1007/s10551-019-04192-x

📃 Kaptein (2011a) A code alone explained very little; the code plus five things around it explained 32% of observed misconduct. https://doi.org/10.1007/s10551-010-0652-5

📃 McCabe, Treviño and Butterfield (1996) Codes went with less unethical behaviour, and more so the more strongly they were implemented and embedded. https://doi.org/10.2307/3857499

📃 Victor and Cullen (1988) 872 employees in four firms. Five kinds of ethical climate, differing within firms as well as between them. https://doi.org/10.2307/2392857

📃 Martin and Cullen (2006) Meta-analysis confirming ethical climate perceptions relate to how people feel and behave at work. https://doi.org/10.1007/s10551-006-9084-7

📃 Treviño, Butterfield and McCabe (1998) Where a code existed, ethical culture was more strongly linked to observed misconduct than climate. https://doi.org/10.2307/3857431

📃 Kaptein (2008) The Corporate Ethical Virtues model: eight dimensions of ethical culture in a 58-item questionnaire. https://doi.org/10.1002/job.520

📃 Kaptein (2011b) 341 triads. Six of eight culture dimensions predicted less misconduct; clear standards and visibility did not. https://doi.org/10.1177/0018726710390536

📃 Kish-Gephart, Harrison and Treviño (2010) Bad apples, bad cases and bad barrels: unethical choice is shaped by all three at once. https://doi.org/10.1037/a0017103

📃 Newman and colleagues (2017) A decade of research on ethical climates reviewed. https://doi.org/10.1017/beq.2017.23

📃 Roy and colleagues (2024) Eighty-nine works on ethical culture reviewed, with a call to study it in teams. https://doi.org/10.1017/beq.2022.44

📃 Simha and Cullen (2012) What ethical climates do to bullying, commitment, behaviour, satisfaction and turnover. https://doi.org/10.5465/amp.2011.0156

📃 Kuenzi, Mayer and Greenbaum (2020) A better measure of ethical organisational climate, linking ethical leadership to less unit misconduct. https://doi.org/10.1111/peps.12356

📃 Mayer, Kuenzi and Greenbaum (2010) Ethical climate carried the link between ethical leadership and employee misconduct. https://doi.org/10.1007/s10551-011-0794-0

📃 Brown, Treviño and Harrison (2005) The standard definition and measure of ethical leadership. https://doi.org/10.1016/j.obhdp.2005.03.002

📃 Treviño, Hartman and Brown (2000) A reputation for ethical leadership needs a moral person and a visible moral manager. https://doi.org/10.2307/41166057

📃 Treviño, Brown and Hartman (2003) To be seen as ethical outside the executive suite, leaders had to make it socially salient. https://doi.org/10.1177/0018726703056001448

📃 Brown and Treviño (2006) Review of the ethical leadership research and where it should go next. https://doi.org/10.1016/j.leaqua.2006.10.004

📃 Schaubroeck and colleagues (2012) 2,572 soldiers across three levels. Culture cascading through units mattered more than a simple trickle-down. https://doi.org/10.5465/amj.2011.0064

📃 Ruiz, Ruiz and Martínez (2011) Ethical leadership trickled partly from top managers through immediate supervisors. https://doi.org/10.1007/s10551-010-0670-3

📃 Ng and Feldman (2015) 101 samples, 29,620 people. Ethical leadership predicted attitudes and performance, sometimes only weakly beyond other styles. https://doi.org/10.1037/a0038246

📃 Mayer and colleagues (2009) 904 employees, 195 managers. Top management's ethical leadership worked through supervisors. https://doi.org/10.1016/j.obhdp.2008.04.002

📃 Babalola and colleagues (2020) Top management's bottom-line focus drove better service and more unethical conduct towards customers. https://doi.org/10.1111/peps.12355

📃 Lemoine, Hartnell and Leroy (2019) How ethical, authentic and servant leadership differ in their moral content. https://doi.org/10.5465/annals.2016.0121

📃 Kaptein (2011c) Ethical culture meant less inaction and more reporting to management and to the hotline. https://doi.org/10.1007/s10551-010-0591-1

📃 Mayer and colleagues (2013) It takes a village: supervisors and coworkers together lowered fear of retaliation and raised internal reporting. https://doi.org/10.1016/j.obhdp.2013.01.002

📃 Detert and Burris (2007) 3,149 employees. Managerial openness predicted speaking up, through psychological safety. https://doi.org/10.5465/amj.2007.26279183

📃 Morrison and Milliken (2000) Organisational silence as a collective barrier to change. https://doi.org/10.2307/259200

📃 Milliken, Morrison and Hewlin (2003) 40 interviews. Most had stayed silent about a concern, mainly for fear of being labelled. https://doi.org/10.1111/1467-6486.00387

📃 Mesmer-Magnus and Viswesvaran (2005) 26 samples, 18,781 people. Factors predicted intentions to report far better than actual reporting. https://doi.org/10.1007/s10551-005-0849-1

📃 Rehg and colleagues (2008) Retaliation against whistleblowers followed lack of support, and damaged relationships with supervisors. https://doi.org/10.1287/orsc.1070.0310

📃 Treviño and Victor (1992) People who reported a peer were seen as highly ethical and unlikable at the same time. https://doi.org/10.2307/256472

📃 Stubben and Welch (2020) Nearly two million internal reports. More active reporting systems went with fewer and smaller fines and lawsuits. https://doi.org/10.1111/1475-679x.12303

📃 Miceli and Near (1985) 8,600 employees. People reported with convincing evidence and serious wrongdoing; retaliation pushed reports outside. https://doi.org/10.1111/j.1744-6570.1985.tb00558.x

📃 Moore and colleagues (2012) The tendency to morally disengage predicted fraud and other unethical behaviour at work. https://doi.org/10.1111/j.1744-6570.2011.01237.x

📃 Detert, Treviño and Sweitzer (2008) Empathy and moral identity lowered moral disengagement; cynicism raised it. https://doi.org/10.1037/0021-9010.93.2.374

📃 Bandura (1999) The mechanisms of moral disengagement, and how harm spreads through divided functions. https://doi.org/10.1207/s15327957pspr0303_3

📃 Ogunfowora and colleagues (2022) Meta-analysis. Moral disengagement predicted misconduct, yet guilt and shame still followed it. https://doi.org/10.1037/apl0000912

📃 Tenbrunsel and Messick (2004) Ethical fading: how self-deception lets the moral side of a decision disappear. https://doi.org/10.1023/b:sore.0000027411.35832.53

📃 Ashforth and Anand (2003) How corruption becomes normal: routines, rationalisations and socialising newcomers. https://doi.org/10.1016/s0191-3085(03)25001-2

📃 Robinson and O'Leary-Kelly (1998) 187 employees in 35 groups. People's antisocial behaviour followed their coworkers'. https://doi.org/10.2307/256963

📃 Umphress and Bingham (2011) Doing wrong for the organisation, and how the moral content gets overlooked. https://doi.org/10.1287/orsc.1100.0559

📃 Ordóñez and colleagues (2009) Goals gone wild: the side effects of goal setting, and the case for a warning label. https://doi.org/10.5465/amp.2009.37007999

📃 Locke and Latham (2009) The reply: goal-setting's founders said the critique breached good scholarship. https://doi.org/10.5465/amp.2009.37008000

📃 Schweitzer, Ordóñez and Douma (2004) People with unmet goals cheated more, especially when they fell only a little short. https://doi.org/10.2307/20159591

📃 Welsh and Ordóñez (2014) Consecutive high goals depleted self-control, which carried a rise in unethical behaviour. https://doi.org/10.1016/j.obhdp.2013.07.006

📃 Kerr (1975) On the folly of rewarding one behaviour while hoping for another. https://doi.org/10.2307/255378

📃 Barsky (2008) How goal setting can cost ethics through recognition and moral disengagement. https://doi.org/10.1007/s10551-007-9481-6

📃 Niven and Healy (2016) Specific goals made people advocate unethical methods; only some then overstated results. https://doi.org/10.1007/s10551-015-2545-0

📃 Mitchell and colleagues (2018) Performance pressure raised self-protection, anger and self-serving thinking, which drove cheating. https://doi.org/10.1037/apl0000254

📃 Greenbaum, Mawritz and Eissa (2012) Supervisors' bottom-line mentality passed to employees and went with undermining colleagues. https://doi.org/10.1037/a0025217

📃 Waples and colleagues (2009) Meta-analysis of 25 business ethics programmes: minimal overall effect. https://doi.org/10.1007/s10551-008-9875-0

📃 Watts and colleagues (2017) 66 studies, 10,069 people. Ethics training produced sizable benefits and had improved. https://doi.org/10.1080/10508422.2016.1182025

📃 Medeiros and colleagues (2017) Professional, focused, workshop-based ethics training worked best. https://doi.org/10.1007/s10805-017-9281-2

📃 Warren, Gaspar and Laufer (2014) One training session at a bank. Better ethical culture two years later. https://doi.org/10.5840/beq2014233

📃 Kristal and colleagues (2020) Signing an honesty statement first, rather than last, made no difference in six studies. https://doi.org/10.1073/pnas.1911695117

📃 Treviño, den Nieuwenboer and Kish-Gephart (2014) Review of (un)ethical behaviour in organisations, from infrastructure to individual thinking. https://doi.org/10.1146/annurev-psych-113011-143745

📃 Park and Blenkinsopp (2013) Ethics programmes cut misconduct only through the culture they built. https://doi.org/10.1108/ijpsm-01-2012-0004

📃 Antes and colleagues (2009) Meta-analysis of 26 programmes: modest effects, best when case-based and interactive. https://doi.org/10.1080/10508420903035380

📃 Mulhearn and colleagues (2017) 330 programmes. Decision-process training d = 0.50; general discussion d = 0.31. https://doi.org/10.1007/s11948-016-9803-0

📃 Katsarov and colleagues (2022) 30 studies. Experiential training with emotional involvement beat intellectual deliberation. https://doi.org/10.1007/s10648-021-09630-9

📃 Miceli and colleagues (2012) More than 3,000 people. Evidence, leverage and less coworker invalidation predicted reporting. https://doi.org/10.1177/0018726712447004

📃 Kaptein (2022) 3,076 employees. The more wrongdoing people saw, the less they meant to report it inside. https://doi.org/10.1007/s10551-020-04663-6

📃 Oelrich (2021) 1,416 employees. People underestimated how much retaliation weighs on a real decision to report. https://doi.org/10.1111/beer.12337

📃 Dungan, Young and Waytz (2019) 42,020 federal employees. Fairness predicted reporting; loyalty predicted silence. https://doi.org/10.1016/j.jesp.2019.103848

📃 Qiu, Mai and Ellis (2025) A small favour made people less likely to report the person who gave it. https://doi.org/10.1002/job.2872

📃 Vadera, Tenbrunsel and Diekmann (2025) Values talk built intentions to report; accountability and protection turned them into action. https://doi.org/10.1287/orsc.2021.15292

📃 Brown and Treviño (2014) Career role models shaped leaders' ethical leadership; top management role models did not. https://doi.org/10.1007/s10551-013-1769-0

📃 Welsh and colleagues (2019) Learning goals cut unethical behaviour without cutting performance, even when hard. https://doi.org/10.1016/j.obhdp.2018.11.001

📃 Welsh and colleagues (2020) High goals led to moral disengagement, more so for the most committed. https://doi.org/10.1002/job.2467

📃 Paterson and Huang (2019) What leaders did and said about ethics shaped whether people saw ethics as part of the job. https://doi.org/10.1177/0149206318771166

📃 Wang and colleagues (2021) 214 employees. Ethical leaders were role models only for some followers. https://doi.org/10.1007/s10551-019-04353-y

📃 Wang, Xu and Liu (2018) Middle supervisors carried ethical leadership down through expecting misconduct to be punished. https://doi.org/10.1007/s10551-016-3361-x

📃 Cabana and Kaptein (2025) 202 UK teams. A well-run ethics programme worked through each team's culture. https://doi.org/10.1111/beer.12661

📃 Kreismann and Talaulicar (2021) Review of 92 studies. What we know about workplace ethics training is still fragile. https://doi.org/10.1177/1534484320983533

P.P.S. Why Do I Even Have the Nerve to Write This?

Fair question.

I am someone who has spent years working alongside remarkable executives, navigating chaos, translating vision into structure and figuring things out, sometimes beautifully and sometimes the very hard way. I have had the privilege of being the steady right hand to leaders who move fast, take risks and expect a lot. That environment has shaped me more than any classroom ever could.

I also happen to have an MBA, and I am studying psychology because people fascinate me. How we work. Why we disconnect. What actually holds us together when the pressure rises.

But really, none of that is the point.

I am here because I have lived the erosion and rebuilt from it. I have seen the cost of carrying too much and the relief that arrives when you finally stop disappearing inside your own competence.

So I share what I have learned in case it helps someone else. Take what is useful. Leave the rest. And remember this one truth that professionals like you often forget:

You are probably doing far better than you give yourself credit for.

Guardians of Good: What the Research Says Makes an Organisation's Ethics Work.

What you will get out of this

The code on the wall and the code in the room

Infrastructure, or a very expensive facade

The tone at the top, and the tone in the middle

The culture you can measure

Speaking up, and whether anyone actually does

How good people drift

Targets, incentives and goals gone wild

Does ethics training work?

The view from the seat beside the decision

The ethics check

Where to start

Free tools

P.S. Want to Go Deeper on Ethics in Organisations?

P.P.S. Why Do I Even Have the Nerve to Write This?

Recent Posts

Comments