Lost Wallet Study Is Honestly Not So Clear Cut About Honesty

Recent news stories have been touting how honest we all are, as evidenced by a large-scale experiment that involved the use of faked lost wallets.

Many have said that the study conclusively showed a seemingly counterintuitive result, namely that people were “more honest” when a lost wallet had more money in it, apparently showcasing a sense of altruism, along with a personal aversion toward feeling as though they themselves were a thief.

As a headline grabber, the study and the interpretation of the results certainly provides a feel-good sensation and likely brightens our day as a humanity.

That being said, though I don’t want to rain on the happiness about honesty parade, I think it might be helpful to take a closer look at the study and make sure that the popular interpretation is appropriate.

Readers of my column are aware that I’ve previously tackled research studies done in the self-driving and driverless car realm, and offered that we should be cautious in taking at face value any published study and the concomitant interpretation that oftentimes becomes the soundbite conveying what the study means and how to use its results.

Let’s unpack the wallet study, shall we.

Background About the Wallet Study Experiment

Researchers decided that it would be insightful to conduct a kind of social experiment to try and ascertain civic honesty, doing so on a widespread basis not previously undertaken. In this case, they used over a dozen research assistants that fanned out to over 40 countries, 355 major cities, and involved over 17,000 unsuspecting subjects into the experiment, taking place over a roughly three-year time span.

Here’s what they did.

Faked wallets were prepared and the research assistants would go into a place of business, find an employee that seemed to be at a reception desk or equivalent, and would hurriedly hand the employee the wallet, explaining that the wallet had been found outside around the corner, doing so using this script: “Hi, I found this [showing the wallet] on the street just around the corner. Somebody must have lost it. I’m in a hurry and have to go. Can you please take care of it?”

The research assistant would try to push the wallet toward the employee, hoping that the employee would essentially take possession of the wallet, and then the researcher got out of the vicinity as quickly as possible so as to avoid being asked any questions about the wallet or the circumstances thereof. Unwittingly, the employee becomes an experimental subject, and the research assistant is presumably able to pull this off without tipping the employee that anything is untoward.

Of the businesses that the researchers chose, they reported that 23% were banks, 14% were post offices, 22% were hotels, 21% were public offices such as police stations and courts of law, and 20% were cultural establishments. The faked wallets were somewhat unusual in that they were actually transparent oversized business card cases, which was used to enable the subjects to readily see the contents without having to actually open the wallet per se. Within the faked wallets were three identical business cards, each containing the name, title, and email address of presumably the wallet owner (this was faked by the researchers), and had a grocery list, a key, and possibly some cash.

For the experimental treatment, the researchers were aiming to alter the amount of cash in various wallets to discern whether the magnitude of the included cash would impact the likelihood of the subjects opting to try and contact the presumed wallet owner, which the only means to do so would be via the email address shown on the business cards (note that there wasn’t a telephone number on the business cards, nor a mailing address, thus limiting the means by which the wallet owner could be contacted).

There were three categories of the cash magnitude: (i) the NoMoney category in which there was no cash in the faked wallet, (ii) the Money category that had $13.45 in the wallet, and (iii) the BigMoney category that had $94.15 in cash inside the wallet.

According to the researchers, it turns out that in the NoMoney category there were 40% that attempted to contact the wallet owner, and the Money category had 51% that tried to contact the wallet owner, while the BigMoney category had 72% that tried to contact the wallet owner.

From these results, most seem to conclude that since the BigMoney category had the highest percentage of attempts to reach the wallet owner, it is a “surprising” result since you would expect that if people were generally tending toward being dishonest they would seek to pocket the BigMoney and not contact the wallet owner, and presumably would be more likely to contact the wallet owner when there was NoMoney since there wasn’t anything to steal.

That’s how the headlines have come to claiming that we all are apparently more honest than might have been assumed, and that we seem to also be personally motivated via a nagging feeling that if we don’t do the honest act then we must be a thief and our consciences won’t tolerate it.

I suppose you could try to reach that conclusion, but it seems to leave out a lot of other equally fitting explanations.

Unpacking The Wallet Study Aspects

One of the biggest concerns that I see about the study approach was that contrary to what the news media says, it was not a study of overall civic honesty because it purposely narrowed the nature of the subjects being drawn into the experiment.

The subjects were presumably employees of businesses or entities of one kind or another. They were made unwitting participants while on-the-job.

I point this out because it would seem a far cry from say randomly going up to people on the street or in a park and getting them to become subjects in the experiment. That might be a more applicable way to try and aim at the general population.

Instead, the research involved people employed that were at their workplace, and for which I submit the results would potentially differ from doing the same kind of experiment for anyone in the open public (or, let me put it this way, trying to generalize from the study to suggest that all people would do the same as the subjects in this experiment seems, well, overreaching).

Sampling bias is an important matter to any study.

One of the most famous examples of how a sampling bias can misshape a study involves the matter of a telephone survey in 1936 that conducted a poll of whom voters were going to most likely choose for the U.S. presidency, and the poll results got it quite wrong. This was attributed to the aspect that at that time period the only people that could afford phones were not a cross-representation of the American voters.

In the wallet study, the employees were working at their job and so might have perceived a greater need to take action about the lost wallet, not because of any personal sense of altruistic honesty per se, but due to the fact that they encountered the situation while at work. The researchers tried to deal with this facet somewhat by trying to note if there were any cameras or other eyewitnesses, but I contend that even if there wasn’t any such monitoring, the employee still has a sense of duty to their employer and their desire to keep their job.

The researchers suggest that by telling the subjects that the wallet was found outside and around the corner that the subject would feel at ease, as though it would be unlikely that the owner might come looking for the wallet. Thus, presumably, the employee is relieved of any chance of being named as having received the wallet.

Not quite.

This brings me to the next concern, specifically that the experimenters themselves could be considered as shaping the results, without realizing they are doing so.

Though the wallet owner might or might not show-up, the research assistant is indeed a witness to the act of providing the wallet to the employee. At some later point, as far as the employee knows, the research assistant might return and ask whether the employee had ever done anything about the wallet or could attest that they gave the wallet to the employee.

Once again, this ties back to the workplace element. The subject can be found again, and has presumably a want to keep their job, thus they might be more so motivated to take action about the wallets.

Again, this is quite different than if you found a wallet on the street or in a park, by yourself. You’d likely pick-up the wallet, look around to see if someone nearby might be the owner, or might be a witness to your picking up the wallet. Nobody though knows who you are. In the case of the wallet study, the subjects that they work at company X and work at the reception or similar area can be ultimately retraced, and they know it.

In the case of this faked wallet study it has contrived a built-in witness, the research assistant. I realize that the study assumes that the employee assumes that the research assistant was just wandering in and won’t ever return, but I wonder if that’s really what the employees imagined.

As an anecdotal example, the other day I was at a restaurant waiting to be seated, when a person came up to the hostess and explained they had lost their car keys, had been standing around in the parking lot trying to find them, when another person came up and explained they had found the keys and given them to the hostess in the restaurant, including describing what the hostess looked like. Though it was the owner that came to get the keys, my point is that the person that turned in the keys was a witness of having given the keys to the employee.

More Qualms About The Study Interpretation

Another element that seems worthy of attention involves the money amounts.

The media led with the headline that the more money involved, the more honest someone becomes. This needs to be clarified, I believe.

As mentioned, the study used either no money in the wallet, or $13.45, or $94.15 in the wallet. I’m not so sure that we can really call the $94.15 as being “big money” in any overall sense. Yes, the $94.15 is more than the $13.45, but let’s all agree (I believe) that something around $100 is not necessarily the same as if the wallet had $1,000 or had $10,000 in it, or $100,000 in it (just being a bit provocative since I realize that stuffing $100,000 into the wallet would be rather difficult to do).

Does the amount of money tie to the “civic honesty” factor or might it really be more connected to the job repercussions aspects?

If your manager comes to you and asks what did you do about some wallet that was given to you in a lost-and-found manner, and the wallet had a substantive dollar amount, you’d likely look worse if you hadn’t tried to contact the wallet owner. For a wallet that had nothing of value, you could always say that it had no money and therefore you just sat on the matter and took no action. In contrast, if the wallet had any substantive money in it, you’d look like you were negligent in not trying to get it back to the owner.

Furthermore, you could possibly be accused of taking the money from the wallet and also the odds are that perhaps the wallet owner will try to come and find the wallet, or maybe even the witness (the research assistant) might return to try and claim it. All of those possibilities might get weighed in the mind of the employee. Given that they could possibly lose their job for not taking action or having other accusations made, it would tend to behoove them to do something, especially when the money amount gets higher.

Not wanting to seem to be a downer about honesty, but suppose the wallet did contain $100,000, which presumably would be a huge amount and something beyond the job pay of the employee. In that case, there might be quite a mental tussle by the employee as to whether they can pocket the truly big money, though at the risk of losing their job (but it might be worth it for the money involved), versus the risk that the wallet owner or the witness would try to come and find the wallet.

Essentially, even if you buy into the idea that a person will be more honest as the amount of money rises, perhaps there is some threshold at which the “honesty” gives way to the dollars involved, becoming so tempting that honesty no longer holds or might be (shall we say) distorted to fit the situation?

Conclusion

I certainly don’t want to quash your hopes that people are honest, maybe they are, but I think that whenever any research study is undertaken it becomes important to step beyond the headlines.

In the self-driving driverless car realm, there are numerous related stories that I had posted, for example:

1. I had previously pointed out that the driving record of those driving a particular brand of car might differ from the driving practices of the general population because of the sampling selection involved. In the case of Tesla’s, at this juncture the buyers are able to afford a more expensive car than is the average car, and they are early adopters, all of which makes them potentially unlike the general population of drivers.
2. Readers also might recall my in-depth analysis that examined how surveys and polls of Americans concerning their viewpoints of self-driving driverless cars often make a myriad of research choices that essentially bound or direct the outcomes, including whom the survey or poll contacts, the questions and scripts used, etc., and thus in some sense pre-shape the potential results.

I’ve used the wallet study to illuminate ways in which you can look underneath the hood about any research study that pops into the popular press and catches the eye of the world. What is the sampling method used? How was the effort undertaken? What explanations and alternative explanations can be made? And so on.

We need to make sure that we have our eyes wide open and honestly give a close scrubbing to whatever news we’re being fed.