Book review: The (Honest) Truth About Dishonesty

Authored by Dan Ariely in 2012 (ISBN 978-0-06-218359-0).

The enjoyable short book could have been shorter if presented as a wiki page.  I believe I can summarize it as follows.

The author, a professor with Duke University, asks how we can predict cheating:  what encourages and discourages it.  He and colleagues from various schools set up experiments to find out.

He starts with SMORC, the Simple Model of Rational Crime, which posits that all of us cheat when it is in our rational interest to do so.  The theory is most likely incorrect because it predicts much more cheating and crime than actually exists.  All of us encounter situations where stealing would be easy and nearly risk-free, and yet we do not.  Ariely suggests we do not because we like to think of ourselves as good people and cheating makes that difficult.  He then extends his approach:  we are likely to cheat if we can do so and still think well of ourselves.

His team developed a test, called the matrix test, where participants are asked to perform a mathematical operation upon twenty matrices within a certain amount of time.  The control group completes the test and submits the results to a proctor who tallies the results and pays based on the number of completed matrices.  The average participant completes seven matrices (scores 7) in the given time.

Next, another group takes the same test but tallies the results themselves, puts the test sheets into a shredder, walks to another room, states the number of completed matrices, and receives payment.  If SMORC were an accurate view of the universe then all the subjects in this test should state "20" as the completed number, because that is the number that maximizes return and still has zero risk.  Yet, under these circumstances, these subjects on average answer 9 correctly (score 9):  more than the control group, and so cheating is for sure occurring, but less than predicted by SMORC.

Ariely and his colleagues repeat this test under many conditions, described below.

1) Vary the amount of money paid for each correct answer, ranging from 25 cents to 10 dollars.  One might assume that cheating would increase as the amount to be gained increased, yet it had no effect.  If anything slightly less cheating occurred with the higher reward amounts.

2) Tire the subjects.  Allow both sets of subjects to shred their tests before collecting their pay.  Before the matrix test give both groups a writing test, with the control group not allowed to use the letters X and Z (not so difficult) and the other group not allowed to use the letters E and N (pretty difficult).  Mentally fatigued subjects cheated more.  (Related, though not Ariely's research:  parole boards are more likely to grant parole during early-morning cases and cases presented immediately after lunch, presumably when the board members possess more energy.)

3a) Reduce the role of the proctor.  The proctor gives instructions and answers questions but the money is dispensed by a person in another room.  Under these conditions the non-control group states, on average, 10 solved matrices (score of 10), 3 more than the control group and 1 more than the original non-control group.

3b) Give examples of dishonesty.  For the non-control group have an accomplice stand up at the beginning of the test and ask the proctor, in front of the other subjects, "can we not cheat by mis-stating our results?"  The proctor answers, "yes".  Under these conditions the non-control group scores 12 on average.

3c) Give examples of dishonesty.  For the non-control group have an accomplice stand up one minute after the test has started and say, out loud, "finished with all 20".  Plainly, this is not possible, but suggests to the others that cheating will not be caught AND that other people are cheating.  Under these conditions the average score was 15.  This situation Ariely labels the Madoff condition because the participants are taking social cues from an insider.

3d) Give examples of dishonesty.  These tests took place on the Carnegie Mellon University campus located in Pittsburgh, PA.  Also in the city is University of Pittsburgh, a rival school.  Now, repeat 3c, except the accomplice is wearing a "UPitt" T-shirt.  Under these conditions the average score was 9, even better (less cheating) than under the nominal 3a conditions.

The above results are consistent with Ariely's thesis, which is that all of us cheat to the extent we can justify it to ourselves.  The shenanigans on Wall Street, in Enron, and with doctors accepting perks from pharma representatives, are examples of normal people justifying immoral behavior by deceiving themselves.  The deceit is made easier if those hurt are unknown to the actor and the actor can create reasonable justifications.

"Create" from the previous sentence led Ariely to another set of experiments.  Dots were briefly presented on two sides of monitor and the participant is asked to choose the side with more dots.  More money was paid for answering one side or the other.  Of interest were the circumstances where the number of dots was very close.

All participants, before engaging in this test, took a personality test that measured creativity.  Again, two groups were created, one where cheating was not possible and one where it was.  Again, the group where cheating was possible scored higher than the control group.  Among the non-control, further, those with the highest creativity scored highest (cheated most).

These result are very discouraging.  Cheating is everywhere.  The most creative among us cheat more.  Observing examples of cheating lead to more cheating, a vicious circle.  What can be done?  Four things, according to Ariely, which I join into one:  immediate negative feedback.

Ariely identifies four countermeasures:  pledges, signatures, moral reminders and supervision.  Pledges and signatures should occur or be made before the activity commences.  For example, Ariely recommended to the IRS that the attestation be put at the top of the forms.  By the time the taxpayer has reached the bottom he or she will have already cheated (if at all) and will have justified it to himself or herself (if at all).  Moral reminders administered directly before taking tests seem to lessen cheating.  And, contemporaneous supervision leads to less cheating:  in one version of the matrix test the non-control group was still permitted to shred their results and self-state their payment, but while taking the test they were observed by another participant (who then took the test while the first participant watched).  In these non-control groups no cheating was observed.

Book review: To Kill A Mockingbird

Authored by Harper Lee in 1960 and winner of the Pulitzer prize.

I read this book thirty years ago and this past month.  My eighth-grade daughter received it as an assignment, much as I did.  I did not remember the plot at all, which is surprising, because this is a great book!

Set in fictional Southern Maycomb in the 1930s, Ms. Lee narrates from the point of view of an adolescent girl, Scout.  I recall reading an essay by Richard Rhodes, author of The Making of the Atomic Bomb, who said the writer's first step is to determine the voice with which to tell the story:  first-person, third-person omniscient, etc.  Rhodes authored The Donner Party from the point of view of various journal entries from the doomed travelers.

The author is playful with her language and one can hear the quaint, "backwards" Southern mannerisms and speech patterns, and some of the people of Maycomb county were low-life bigots, but the author plainly respects her characters.  I finished my read with a really warm feeling for the main characters:  so full of goodness.