Assessing and Promoting Honesty in a Workforce
A study reported in the December 4, 2014 issue of the journal Nature offers an interesting framework for discussing honesty in an employment context. Although I disagree with the researchers about their methods and inferences, I still think their study reveals something interesting about human behavior; I also agree that further investigation is needed in this area, including an examination of the effectiveness of ethics training. The research reminds us that many employees will behave honestly, even when they can’t be caught, but also highlights the needs for safeguards where honesty may be compromised by self-interest.
The study recruited 128 employees from a major international bank, half working directly in financial services and the other half working in support or administrative roles such as clerical support, human resources, or information technology. All participants had worked in the banking industry for an average of 11.5 years, and were randomly assigned to one of two groups. Employees in the test group were asked a series of questions that required them to think about their professional background and experience; employees in the control group were asked unrelated questions, for example about their leisure time. All subjects were then told that they would be asked to perform ten coin-tossing tasks, and that depending on the outcomes of these coin tosses (for example, if they flipped heads rather than tails), they would receive $20 (rather than $0) per successful coin toss, but only if the number of successful coin tosses was greater than or equal to those of a randomly drawn subject. The coin tosses were performed anonymously, and subjects reported their results online.
By design, it’s impossible to tell whether any particular test subject cheated on his or her coin tosses, but we’d expect accurately reported coin tosses to follow a binomial distribution, centered symmetrically around 50% of the coin flips being successful. That was roughly what the researchers found in the control group. Most of these employees, in other words, reported their coin flips honestly. However, employees who had been surveyed about details of their job before conducting the coin toss reported successful coin flips at an average rate of 58.2%, significantly improving their odds of cashing in on the reward. The probability of such an outcome occurring purely by chance is quite low, so cheating would seem to be a reasonable inference. The researchers argue that members of the test group were predisposed to cheat because they’d been reminded of their work and the culture associated with it.
Because the analysis relies on probability distributions, it isn’t clear exactly how many people cheated in the control and test groups. It appears from graphs published with the study that around 5% in the control group probably cheated, and around 30% of the test group did so. Further, it’s interesting that among those who cheated, only the same 5% of the control group and around 12% of the test group cheated as much as possible within the framework of the experiment and so claimed the maximum reward (10 successful coin flips, or $200). Other people, arguably more if we exclude the 5% who may have cheated anyway, inflated their coin flips just enough to receive a payout, but settled for $120 rather than $200. The scientists note: “This phenomenon [of not taking full advantage of cheating opportunities] is typically attributed to the presence of an honesty norm that causes psychological costs when one cheats, such as the loss of one’s positive self-image”. Hardly anyone in either group, interestingly, claimed to have 9 successful coin flips. Among people who decide to cheat, it seems, there are those who exploit the system for all it’s worth, and those who cheat just enough to get ahead but try not to be greedy (or not to get caught), and no one in between.
In a follow-up study, the researchers repeated the experiment with employees pooled from other industries, such as information technology, pharmaceuticals, manufacturing, and telecommunications, including many members of middle or upper management. They were unable to replicate the same heightened pattern of cheating after reminding test subjects of their work in these fields. Instead, it appears that both the control group and test group behaved in the same manner, with the vast majority reporting their coin flips honestly, and only a tiny handful lying for financial gain. The researchers conclude from this that “the prevailing business culture in the banking industry favours dishonest behavior.”
With respect to the scientists who designed an intriguing series of experiments, I think they overstate their findings. It appears, for example, that their study was limited to employees of a single bank, while their follow-up study combined employees working for a number of different companies in a variety of industries. Generalizing from the conduct of employees in a single company to “the prevailing business culture in the banking industry” seems, without supporting evidence, to be unfair. Further, I’m not sure that the follow-up study, drawing on test subjects from a wide variety of fields, really presents a fair contrast because I’d think that tendencies toward dishonesty in any one company or industry, if they existed, would be overshadowed or even completely diluted by the overwhelmingly honest behavior exhibited by the rest of the cohort. So while the study is fascinating, the researchers have leaped to conclusions I don’t think are justified by their evidence.
Instead, I think it’s important to recognize that most employees participating in this study, regardless of industry, chose to behave honestly even when they couldn’t be caught. Of those who decided to cheat, it appears that some of them will accept an opportunity to cheat without regard to the culture of their employer or industry. If those 4 or 5% are excluded, it appears that only a little over a quarter of the employees in this particular company decided to cheat when reminded about their work. I agree that this presents a problem, but I don’t think we can conclude that the “prevailing business culture” promotes dishonesty if the conduct of most participants contradicts that assumption. Instead, if this study can be replicated in other companies, what we seem to find is three constituencies in the workforce: the majority, who remain honest in the face of significant incentives to cheat; a small minority, who are disposed to cheat regardless of their line of work; and another minority, who were disposed to cheat – some a little, some a lot – when reminded of their work.
Although it wouldn’t be constructive to implement sweeping solutions before we better understand what’s causing this behavior, the research does present some problems that employers cannot afford to ignore. First, the study suggests that at least some employees may be disposed to dishonesty for reasons unrelated to their job, employer, or profession; employers should be concerned about this, but should not overstate the problem, and should not overreact. Although the circumstances under which such behavior arises may be unique in many cases, every industry can produce examples of dishonest individuals who were hired despite rigorous screening, interviews, background checks, and peer reviews; more of the same is unlikely to keep them out of the workforce, or educate them to make more honest choices.
The study also suggests that at least some employees who would ordinarily conduct themselves honestly, even when there was no chance of being caught, will make different choices if prompted to think about their work for the employer or in a particular industry; if this study is accurate, that number may reflect as much as a quarter of the workforce. If we limit the scope of the researchers’ conclusions to the particular company they studied, I’m inclined to agree that this data seems to indicate that some work-related process is responsible. There are undoubtedly other companies where employee behavior will follow similar patterns.
Against this background, there’s no reason why employers shouldn’t adopt reasonable means to protect themselves and their customers from the occasional employee who makes poor ethical choices. Financial controls and codes of conduct are an obvious first step, and have been familiar to many U.S. employers for quite some time. Beyond this, employers should consider mandatory training for employees on ethical conduct, ideally emphasizing the kinds of decisions they may be called upon to make in the course of their work, and either through role-playing or written responses to questions, getting the employees in the habit of thinking about their work in ethical terms. These need not be extraordinary undertakings. Educational psychologists have long recognized, for example, that even relatively simple projects such as asking students to write a one-page essay can have a profound positive effect on academic and disciplinary performance over the course of the school year. Employers can borrow such practices for their own needs.
It should be noted that the most effective trainings are often those where instructors are sensitive to their audience, and to variations between audiences. I’ve found in particular that sophisticated audiences become much more engaged when they aren’t talked down to, when the instructor recognizes their intellect, basic morality, and sense of compassion, and when the audience is offered insights and information it hadn’t anticipated. By contrast, a dull recitation of the text of a policy is unlikely to do the job and may even convey the disastrous impression that such policies apply only to low-ranking personnel. That, of course, would be counterproductive, as it’s those with the most authority who are in the position to do the most damage.
Beyond common sense, however, we need to develop strategies for measuring the effectiveness of ethics training in the workforce. As a commenter on this study pointed out, it would be interesting to see whether study participants performed differently after completing ethics training. I hope to discuss potential metrics for evaluating training in a future essay.