Moral Overconfidence

A Washington Post article from from last Saturday says:

In the 2006 survey of more than 36,000 high school students, 60 percent said they cheated on a test, 82 percent said they lied to their parents about something significant and 28 percent said they stole something from a store. … 92 percent said they were "satisfied with my own ethics and character." About 74 percent said that "when it comes to doing right, I am better than most people I know." … the percentage of students who lie, cheat or steal could be higher than the survey found.   When asked, 27 percent of the students admitted that they lied on at least one survey question.

It is hard to see how 3/4 of people could be better than most people they know about anything; moral overconfidence is a far more plausible explanation.

On a similar topic, a clever set of experiments, described in "Exploiting Moral Wiggle Room" (and discussed at Marginal Revolution) shows that people act to appear fair, but are less fair when given even feeble excuses for their unfair behavior.

Are you willing to admit that you are about as moral as most people you know?  I am. 

GD Star Rating
Tagged as: ,
Trackback URL:
  • Guy Kahane

    “It is hard to see how 3/4 of people could be better than most people they know about anything; moral overconfidence is a far more plausible explanation.”

    Since most of those surveyed admitted to a variety of wrongdoing of the kind they are not likely to be able to attribute to most others (as most wrongoing is, unsurprisingly, kept hidden from others), the bias here seems even greater than in the many other areas where people rate themselves higher than most others. (We’re assuming of course, and most reasonably, that the survey question/s they may have lied about are not likely to include those where they were asked to rate themselves.)

  • Alejandro Dubrovsky

    (I’m hoping the following comment gets shot down and so save me the time of having to look in dense literature for whether it could be correct)
    Would significantly more than half of the students answer positively to a more specific question like “Are you less likely to steal a cheap book that you desperately want to read than most people you know?” or “Are you more likely to help a blind mid-50s male to cross the road than most people you know?”.
    I ask this because in the examples I read of bias (and I’ve only read popular accounts), the question that triggers the “surprising” answer always seems unnecessarily vague (better driver than average, more likely to succeed). In this instance, to answer “when it comes to doing right, I am better than most people I know” accurately, the person would have to evaluate a huge formula involving weights, a large number of moral principles/dimension and different scaling factors, which of course, they can’t, so will instead look for some important couple of dimensions and evaluate on those, but since people would rank the dimensions differently, getting most people to self-evaluate as above average does not seem surprising or wrong.
    More explicitly, what part of the 50% overshoot in that survey are due to people trying to appear better than most people, what can be attributed to unconcious self-deceit and what can be attributed to lack of computational power?

  • I think most people do not understand who they really are. Or more precisely, that they believe that they understand who they are, but that belief is in error.

    A good recent example of this is actor Michael Richards’ racial-epithet-laced tirade at some patrons in a comedy club. Richard’s later response was “I am not a racist”.

    I suspect Richards probably is not particularly racist, but that in his intoxicated condition and anger at being jeered by a black faces, enabled an anger-driven mental “program” to say thoughts that would normally be rejected as “not me” by Richards’ sense of who he is, instead of being shouted publicly. It is not surprising how this episode was initiated: Richards was being humiliated by being heckled, and the self-concept can react with extraordinary viciousness when it feels the existential threat of humiliation.

    I believe that most people are very much like Richards — unaware of the complexity of the “self” and the responses and reactions that arise in its defence. When they do notice something rather ugly arising in their own behavior, the tendency is to immediately justify it in their own minds, unless (as in Richards’ case) the behavior is so obviously self-destructive and unwise that it is instead simply disowned from the mental model of the self.

    But in reality, the personal sense of self, itself just a collection of related and persistant thought patterns, is simply “claiming ownership” over impersonal mental dynamics which arise based on conditioning and reaction.

    Moral overconfidence is simply another manifestation of this self-concept defending its own existence. Because the basis of the self-concept is separateness, uniqueness, distinction, and privileging its own viewpoints over those of others. This has obvious biological relevance to natural selection, but is less helpful when the objective is to overcome biases.

  • Alejandro Dubrovsky

    That last part in my last comment should be split into the parts attributable to lack of computing power AND the part that means that even given sufficient computation the evaluation is correct (eg i evaluate not stealing cheap books way above helping blind men, you the reverse, and we both agree that i’m more likely not to steal books and you to help blind men, then we are both more moral than the other)

  • Alejandro, I agree, it would be good to see studies that ask people about their relative chances to do very particular things, to see how much differing relative weights can explain this apparent overconfidence.

    Of course, if people believe that there is a common concept of morality, instead of differing private values, then even if they agreed on the relative chance of acts, they would still be biased regarding what is moral.

  • I think Alejandro may be on to something here with his point about asking specific questions. If I recall correctly, various studies have shown that asking people “How bad a driver are you?” garners significantly different answers from “How good a driver are you?” In this case, we might imagine asking the converse general question: “Do you think you are more easily tempted by moral transgressions than most people?” or some such. And asking about holding doors open or donating to charity seems like an even better strategy if we want accurate information, or to encourage accurate self-evaluations.

  • I remember reading recently that religious people tend to behave more “morally” than non-religious. They give more to charity, etc.

    Based on this kind of information, I’m probably considerably less moral than the average. Some of us have to be there to drag the curve down!

  • I don’t mean to be playing the semantic game but I find your closing question pretty impossible to answer. Moral by what standards? The standards of National Socialism (that require I upkeep the race laws)? The moral standards of Bolshevik Communism (that would have me turn my neighbors into the authorities if they don’t sufficiently adhere to communistic tenets)? The morals of the West Coast (that would have me defend the “rights” of people to dress as scantily in public as they wished and to engage in rather serious public displays of affection)? The moral standards of traditional Judeo-Christian values (that consider masturbation a SERIOUS moral wrong)?

    What definition of morality would I be using here?

    Being as I personally can’t see how there could be any such objective thing as “morality”, I can’t see how I could possibly rate myself in comparison to my fellows.

    And if I weren’t overtired I’d find a more satisfying conclusulatory sentence than the current one. 🙂


  • Mnuez, my claim should apply to any concept of morality, such as how you honestly judge it, or as others judge it.

  • Eliezer Yudkowsky wrote:
    “I remember reading recently that religious people tend to behave more “morally” than non-religious. They give more to charity, etc.”

    But the data* of prison population seems to show atheists are just about 1%, when they are a more significative percentage of society. I know there are many possible explanations for that fact, but at least it put some pressure over your affirmation.

  • Pedro, that quote is from Hal Finney, not me.

  • Ups! Sorry.

  • Perhaps, many of the surveyed high school students beleive that most people have low moral values? Perhaps it is not that they are over-confident but are lacking confidence in their fellow members of society.