Understanding honesty.

Are internal auditors more honest than the general population?

IN THE JUNE 1993 issue of Internal Auditor, readers were invited to measure their honesty by completing the "Applicant Review," an honesty test. After completing the test, they phoned an analysis center to report their answers and obtain a percentile score that compared their honesty to others.

During the months of June and July, 720 people completed the test and scored their responses. The gender and age of these respondents are shown in Exhibits 1 and 2 respectively. Although some of the respondents may have been friends or spouses of journal subscribers, we believe over 90 percent were internal auditors and they are all described here as internal auditors.

The average percentile score for this group was 52.4, which is about the same as the general population. A percentile score of 50 indicates about half of the people in society will score higher and half will score lower. The distribution of scores for these 720 people is shown in deciles in Exhibit 3. Since they are percentile scores, they should be evenly distributed from one extreme to the other. However, a disproportionately large number of these 720 people obtained a score in the zero to nine range, and more than usual scored above 70.

* The Applicant Review

The Applicant Review was developed in 1983 as a preemployment screening test, primarily for job applicants in retailing. The test was based on five years of research on honesty, white-collar crime, moral development, and the climate of honesty. The test has been revised three times as a result of additional research and recent legislation.

The Applicant Review is one of about a dozen honesty tests used by employers to screen job applicants. We developed this test because of dissatisfaction with other honesty tests that ask too many offensive and ambiguous questions and disqualify too many honest people. We think the Applicant Review is less offensive, and we are pleased with how well it measures honesty. However, our primary motive is to understand character development and to raise the standards of honesty in society.

* How Scores Are Calculated

A person's honesty percentile is calculated by combining several subscale scores, each of which measures a different dimension of honesty. Each subscale score ranges from zero to 100, with 100 being the most honest. Each subscale is briefly discussed, and the average scores for the general population are compared with the average scores for internal auditors in Exhibit 4.

1. Personal honesty. This score comes from combining responses to items that describe your personal honesty, such as the statement "I am an honest person and would never steal or cheat." This score measures what people say about their own behavior and whether they are honest or dishonest (85 for general population versus 75.6 for auditors).

2. Honesty of others. This score measures how honest you think other people are. People who are honest think other people are also honest, while those who are dishonest tend to think others are also dishonest (70 for general population versus 61 for auditors).

3. Blame for dishonesty. This score measures how wrong you think dishonesty is and how much blame should be attributed to those who behave dishonestly (80 for general population versus 85 for auditors).

4. Punishment. This score measures your degree of punitiveness and whether you think dishonest acts should be severely punished or overlooked. Dishonest people tend to overlook and excuse the dishonesty of others (75 for general public versus 67 for auditors).

5. Definition and standards of honesty. This score measures how explicitly you have defined honesty and whether your standards of honesty are strictly or casually defined (80 for general public versus 75 for auditors).

6. Moral reasoning. This score measures your level of moral reasoning based on the ranking items at the end of the questionnaire. These items cannot be easily faked and our research indicates that some types of moral reasoning are much more conducive to honesty than others (75 for both the general public and auditors).

7. Past behavior. This score measures self-reported admissions of previous theft and substance abuse (95 for general public versus 90 for auditors).

The questionnaire includes several faking scales to prevent people from getting fictitiously high scores. A small amount of faking simply reduces the person's percentile score, while excessive faking produces a score of zero. Considerably less faking was reported by this sample of auditors than most job applicants: .8, or less than one question per response, versus 2.0 questions by the general population.

The scores are also adjusted for what is called a social desirability response bias. Some people have a stronger desire than others to report socially desirable responses, and their scores are adjusted to reduce the biasing effect of this personality trait. This group of auditors displayed considerably less social desirability response bias than normal job applicants.

* Interpreting the Scores

The seven subscales plus the faking and social desirability scales are combined into one honesty scale, which is reported as a percentile score. The percentile scores can range from zero to 99, with higher numbers indicating higher levels of honesty. Retail companies using the Applicant Review for hiring purposes typically set a minimum acceptable score between the 25th to 35th percentile.

Many internal auditors seemed disappointed because their scores were lower than they expected. In several cases, callers did not know how to interpret a percentile score. When one caller was told his score was 69, he responded "So I failed it, then?" The analyst tried to explain that his score was better than average, but he still said, "If I ever got less than 70 percent in college, I figured I failed." Even after he was told that his score of 69 meant that in an average group of 100 people, 69 of them would be expected to score lower, he still had difficulty comprehending the difference between a percentile and a percentage.

Percentile scores do not indicate what percent of the time a person is being honest; they simply compare one person with a larger population. These comparisons are often disturbing. We have learned that almost everyone describes himself or herself as an honest person. When people are asked to write anonymous essays describing how honest they really are, virtually everyone begins by saying "Basically, I'm an honest person." However, we have also learned that most people are not as honest as they think. People have a remarkable ability to rationalize and justify dishonest behavior. Consequently, many people are surprised at their low scores when they measure their honesty and compare themselves to others. The following paragraphs discuss some of the most frequent comments of callers.

* I Can't Believe I Got a Low Score!

Low scores are caused by several factors; no single item can produce a low score. A score of zero means there was insufficient data to calculate a reliable score either because too many items were left blank or because of too many faking responses. Faking responses are defined as extreme responses to selected items that are considered highly improbable, such as "When I find money on the sidewalk, I take it to the police even when it is less than one dollar." Police reports show no one ever submits loose change they find on the sidewalk. So if someone claims to always do it, this is considered a faking response.

The most obvious reason for low scores is that respondents' answers endorse dishonesty -- they do not describe themselves as honest people, they think other people are dishonest, they excuse people for dishonesty and are reluctant to punish them, they have poorly defined standards of honesty, they have been dishonest in the past, or their moral reasoning is weak.

Some people have relatively low scores not because they endorse dishonesty, but because virtually all their answers tend to be neutral -- they only disagree rather than strongly disagree. For example, responses to the item "Most sales clerks deserve to steal a little now and then because they are so badly underpaid" ought to evoke a stronger response than just "disagree." An honest person would never think sales clerks deserved to steal from the company regardless of how underpaid they are, and to obtain a high score a person ought to feel strongly about it. For internal auditors, 22 percent answered disagree, and 77 percent strongly disagreed to this question.

* What Is Honesty? Who's to Say What Honesty Is?

Although everyone seems to agree that the Applicant Review is a test of honesty, some people question whether anyone really knows what honesty is. This concern is usually prompted by moral dilemmas, such as whether it is dishonest to tell a child the picture she painted was ugly. Twenty-nine percent from the internal audit group said they would always lie to the child, 35 percent said frequently, 20 percent said occasionally, 10 percent said seldom, and 5 percent said never.

Occasionally, there are moral dilemmas where honest people will disagree about the right choice. Ethics training, which we support, helps people learn how to resolve such moral dilemmas. Most often, however, people face issues of choice rather than issues of dilemma. People know what is right and wrong, what is fair and unfair, what is honest and dishonest; the question is whether they have the moral character to make the right choice. When internal auditors were asked whether they reported their income taxes honestly, only 50 percent said always, 38 percent said frequently, 6 percent said occasionally, 2 percent said seldom, and 3 percent said never.

Honesty is not difficult to define, and most people have a fairly consistent definition of honesty. Honesty means to be free from deceit and fraud, to be open and above board in your transactions, and to be fair and just in how you treat others. People who are honest do not say things they know are not true, they do not take things of value that belong to others, they do not knowingly give false impressions, and they follow the rules they have agreed to accept. When asked how much cash or merchandise they had stolen in the past three years, only 82 percent of the internal auditor group said they had not stolen any cash, and only 52 percent said they had not stolen any merchandise.

* Is Honesty a Personality Trait?

A personality trait refers to an individual's predisposition to act in a consistent way in a variety of situations. Therefore, honesty can be a personality trait, but it is not a single trait and it does not act alone when influencing behavior. Honesty has multiple dimensions, and the Applicant Review measures several different dimensions of honesty.

In our research on honesty, we have focused primarily on three dimensions of honesty: how well standards are defined, how well they have been internalized, and the degree of remorse for violations. People vary on these three dimensions.

For someone to be considered an extremely honest person, he or she must have very high, well-defined general standards of honesty. These standards must be internalized as moral imperatives, and living a moral life must be so important that violating these standards would cause feelings of shame and remorse.

Standards. Whether a person has high, well-defined general standards of honesty depends largely on the person's experience and training. People have to learn right from wrong, and this training occurs at every age as new issues are confronted. In school, for example, students have to learn when it is acceptable to work together on assignments and exams and what is considered cheating.

Learning acceptable standards of honesty is equally important at work. Reporting travel expenses and using company software are business issues. Copying computer programs is a difficult issue for some people because they do not realize that pirating software is stealing. Some employees think violating the copyright protection on a computer software program by taking a copy home is as immaterial as taking a pencil home from the office. When internal auditors were asked whether sales representatives should be fired for padding their expense accounts each month, 2 percent strongly disagreed, 31 percent disagreed, 44 percent agreed, and 22 percent strongly agreed.

Child-rearing research has demonstrated the importance of teaching children general standards of right and wrong rather than specific rules. For example, taking candy from a store without paying for it is more than just a simple act of not paying for the candy. Children should learn that taking something without paying for it is shoplifting, which is a form of stealing; and stealing is a general form of behavior that is wrong.

Internalization. The degree to which people have internalized standards of honesty is determined by the motives guiding their behavior. Research on moral development explains how people advance through different stages of moral development, and it suggests that only a small percentage of the population reaches the highest level of internalization.

The lowest stages of moral development are called compliance because people comply with the directives of others and do what they are told to do. Their motives for compliance are to avoid punishment or to seek rewards. For example, when internal auditors were asked whether they would steal from the company if two friends asked for help, only 90 percent said absolutely not, 9 percent said they didn't think so, and 1 percent said possibly yes. When asked to rank the most important reason for not stealing, 11 percent said because of the consequences of getting caught, which raises the question of whether they would do it if they were confident of not getting caught.

The next stages of development are called identification because people identify with group standards or conventions of society. Their motives are to follow the rules and laws and do what is necessary for the group or for society to exist.

The highest stages of development are called internalization because standards of right and wrong are a fundamental part of the person's nature. At the highest level of development, people have decided for themselves what is right and proper based on universal principles of truth and justice and they are personally committed to living them. Violating these principles would compromise the self-concepts of these individuals and make him or her less of a person.

Remorse. Honesty should matter; people should not lie or steal with impunity and think nothing of it. An honest person has a genuine concern for doing the right thing and being fair and just. When a mistake is made, an honest person has a genuine feeling of remorse and guilt. This is not to say the person should be overwhelmed with feelings of inadequacy or unworthiness, but that the person feels sufficient discomfort to correct the mistake. Continuing to behave dishonestly is not acceptable; there must be a sincere desire to improve and do better.

* This Test Is Not a Good Measure of My Honesty!

This comment was made by most callers who received low scores. Calling it a bad test probably helped preserve their self-esteem. All paper and pencil honesty tests have certain limitations. Honesty tests can be faked, responses can be influenced by the person's state of mind, and questions can be misinterpreted. Therefore, a person's score may be incorrect. Nevertheless, our research indicates that this test is both highly reliable -- scores are consistent from one time to another -- and valid -- it is a good measure of what we call honesty.

Evidence of the Applicant Review's reliability comes from test-retest reliability studies. Groups of people, mostly students, have been tested twice with a one-week interval between administrations, and their scores on both tests have been highly correlated.

Evidence of the Applicant Review's validity comes from a variety of studies, such as comparing the average scores of known groups who are expected to differ significantly and correlating test scores with admissions of theft. Groups that are expected to have higher honesty, such as housewives and department store supervisors, do indeed have much higher scores than groups that are expected to score low, such as prisoners and youths in correctional centers.

Perhaps the most impressive validity evidence comes from correlating the test scores of people who complete the test and then write an anonymous essay about how honest they really are. When the essays are coded on a 10-point scale and compared with the test scores, the correlation is quite impressive. Although people with low scores begin their essay by saying they are basically honest, they typically admit to a variety of dishonest acts such as stealing from their employer, shoplifting, lying to family members, cheating on income taxes, exceeding speed limits, and falsifying reports.

Some evidence for validity of this test is provided by examining the percentile scores and subscale scores of auditors who admitted to stealing cash and merchandise. Of the 720 respondents, 107 admitted to stealing at least some cash and merchandise within the past three years. The average percentile score for this group was only 27; all of the subscale score means (not just the Past Behavior subscale) were also significantly lower, as shown in Exhibit 5.

* If Someone Gets a Low Score Does That Mean He or She Will Steal?

People who have low scores will not necessarily steal the next time they have a chance; but people with low scores are certainly more likely to steal than people with high scores. A basic principle of employee selection is that past behavior is the best predictor of future behavior. Scores on the Applicant Review are significantly related to past behavior -- people who have low scores voluntarily admit to higher levels of previous theft and illegal drug use.

In our early validity studies we were surprised to discover that the single best predictor of employee theft was a question about marijuana use. We should not have been so surprised at this correlation because marijuana use is both illegal and expensive. People who disregard the law by using illegal drugs can probably disregard standards of honesty, and drug habits are too expensive for most people to maintain on an honest income. Unfortunately, the new Americans with Disabilities Act prohibits asking anything more than whether applicants are current drug users.

When asked whether they plan to use marijuana in the future, only 81 percent of internal auditors said absolutely never, 14 percent said they didn't think so, and 5 percent said possibly yes. Three percent of the respondents said they had used marijuana within the previous four weeks. The average percentile score for these 20 people was only 12.6, and their subscale scores were also significantly lower, as shown in Exhibit 5.

While personal honesty is an important predictor of whether someone will lie or steal, our research indicates it is only one of three important variables to be considered. The other two variables are situational pressures and opportunities available for stealing.

Situational pressures refer to the problems a person faces at a particular time, especially a need for money. Although everyone would probably appreciate having more money, some people have intense financial needs to cover short-term emergencies. Occasionally these financial pressures become overwhelming, such as supporting a drug habit, covering gambling losses, or maintaining the social image of an extravagant lifestyle. Situational pressures are particularly intense when people have a "nonsharable need" -- they are too embarrassed to share their financial problems with anyone.

Some situations are more conducive to theft than others. The opportunity to commit fraud is more convenient when people have access to financial accounts, when no one audits them, and when they think it is unlikely they will be caught, prosecuted, or punished. Fraud is easier when companies have poorly defined or weakly enforced internal accounting controls. Occasionally, fraud is so convenient that even honest people are seduced even though they had no intention of being dishonest.

Whether someone will steal in a particular situation depends on more than just the person's level of honesty. Even honest people can succumb to theft when the opportunities are convenient and they face intense situational pressures. All three variables need to be considered to understand and predict employee theft.

* I Got a Low Score Because I Was Honest.

Some of the callers were indignant because their scores were low and they thought only dishonest people would get higher scores. When one caller was told part of the reason her score was low was because she disagreed with such statements as "I am an honest person and would never steal or cheat," she said she was only being honest and should not be penalized for her honesty. She claimed everyone would have to disagree with these statements unless they were lying. When internal auditors were asked whether they were honest and would never steal or cheat, 2 percent strongly disagreed, 30 percent disagreed, 51 percent agreed, and 17 percent strongly agreed.

People who have low scores often do not believe there are people who are really committed to honesty. Conversely, people with high scores are usually surprised to hear the skepticism and rationalizations of some other people because they genuinely believe most people are honest.

People who are extremely honest or dishonest seem to seriously underestimate both the number and character of people who are opposite to themselves. This misperception has been conspicuous in group discussions about such items as "Nearly every worker has stolen something from his or her company at one time or another." In this study, only 7 percent of internal auditors strongly disagreed, 34 percent disagreed, 43 percent agreed, and 14 percent strongly agreed. When this item is discussed in workshops, widely opposing views are usually expressed. Those with low scores strongly agree with this statement and think anyone who disagrees with them is either naive or lying. However, people with high scores claim they have never stolen anything from their employers, and they assume most people are just like themselves. Although they have heard stories of significant employee theft, they assume these are exceptional cases that attract undue attention rather than customary practices.

* My Score Is Low Because I Work with Dishonest People.

Since a person's perception of the honesty of others is one of the subscale scores, some people think their honesty scores are low because of their exposure to dishonesty. People who work in law enforcement, tax auditing, loss prevention, and youth corrections claim their scores on the Applicant Review are reduced because they are overexposed to dishonest behavior. This fear is legitimate but not very material, as the data presented earlier indicates.

Previous research has found that the average honesty scores of loss prevention officers in retail companies are higher than almost any other occupation we have measured. As expected, their subscale scores on the honesty of others tend to be lower than normal. But their other scores are noticeably higher than normal because they are continually exposed to issues of honesty and theft, and they witness the consequences of dishonesty. Therefore, the effects of one's job or prior experiences with dishonesty should not necessarily cause one to have a low score. As with all past experience, the impact on one's character will depend on how the person responds to it and thinks about it.

* Summary

The responses of the internal auditors to the Applicant Review indicated their levels of honesty were about the same as the general population. They demonstrated less evidence of faking, less social desirability response bias, and stronger feelings that dishonesty is wrong. However, auditors tend to describe themselves as less honest people, they are inclined to overlook the dishonest acts of others, and their definitions and standards of honesty are not as clearly defined.

David J. Cherrington DBA, FPHR, is Professor of Organizational Behavior at Brigham Young University.

J. Owen Cherrington, PhD, CPA, is Professor of Accounting and Information Systems at Brigham Young University.