Not Every Negative Judgment Is A Bias

A recent article, "Weight bias may harm obese children," summarizes a July 2007 Psychological Bulletin article:

When the study participants were asked to rank the children in the order of whom they would like to be friends with, they ranked the overweight child last.  … Some studies found that a sizable number of teachers harbor negative views of overweight students, seeing them as "untidy," for example, or less likely to succeed than their thinner peers. Other research found that overweight children often report teasing from family members, including parents.

The article repeatedly uses the word "bias" to describe these negative judgments, but it doesn’t bother to show why these effects are in fact biases.  Unless you want to claim that no one should ever be teased or prefer some as friends over others, or that teachers should never estimate student tidiness or success, the question is: what is the evidence that judgments made about fat kids are in fact too negative on average?  Without such evidence, these negative judgments should not be called "biases." 

GD Star Rating
Tagged as: ,
Trackback URL:
  • Hopefully Anonymous

    Robin, I’m not sure about that. First, articles are a written form of probably under a couple thousand words. It’s simply not possible to provide evidence for every single assertion in an article -I think it’s only the unusual ones or ones where readers couldn’t be expected to connect the dots themselves that require some factual assertions in an article.

    Second, if the article says it was bias that teachers “harbor negative views of overweight students, seeing them as “untidy,” for example, or less likely to succeed than their thinner peers”, the article is making a clear implied insertion that either (1) overweight students are more untidy or less likely to succeed than their thinner peers, or that (2) their teacher perceive them as more untidy or less likely to succeed than the evidence suggests. Those views by the teachers would seem to me to meet the definition of bias. Same with kids who would rank overweight children last as desirable friends, even when the overweight kid on balance shows factors that would make a better friend than their alternative friend choices.

    The article may not prove these are all biases, but these all seem to me to be fairly straightforward disprovable propositions. One could undermine the credibility of this article by either (1) contacting the journalist or editor, and finding out the evidence these assertions of bias were based on, or (2) independently research these bias claims, to see if there’s any validity to them based on past studies or by conducting a study of one’s own.

    But I think it jumps the gun a bit to criticize an articled for not providing evidence for every single asserion. It reminds me of the annoying style of some message board participants, that think it’s a valid critical argument to point out that one doesn’t have 100+ footnotes providing evidence for and explaining nuance of every single assertion in a post.

  • Norman Siebrasse

    I see two things going on here. First, what does the word ‘bias’ mean? ‘Bias’ normally has a pejorative connotation; strictly, a ‘heuristic’ is a simplifying assumption or reasoning method and a ‘bias’ is a systematic departure from the rational actor model that results from the use of some (but not all) heuristics. Given the title of the blog, I take it that Robin is using ‘bias’ in this technically correct sense. However, many people use ‘bias’ more or less interchangeably with ‘heuristic’. I admit that I do this myself, largely because it is often not clear whether a particular heuristic is biased. (Kahneman and Tversky originally focused on biased heuristics because they wanted to establish that human reasoning departs from the rational actor model, but this doesn’t imply that all heuristics are biased. One common take is that many heuristics were unbiased or at least efficient in the ancestral environment, but are no longer.) So, it is possible that the article is just using ‘bias’ in this loose sense.

    But I don’t really think that this is what is going on. Many people consider that it is morally wrong to use a heuristic or ‘stereotype’ in judging any other person, even if that heuristic is accurate on average. This seems to be a form of the ‘inside view’ bias discussed in ‘Beware the inside view.’ The “inside view” bias seems to be raised to the level of a moral principle when it comes to judging other people (though of course, as the article cited by Robin shows, people use such stereotypes all the time).

  • Hopefully Anonymous

    “Many people consider that it is morally wrong to use a heuristic or ‘stereotype’ in judging any other person, even if that heuristic is accurate on average.”

    It seems to me that phenomenon exists independently of whether such a heuristic is optimally accurate. I think one would weigh the efficiency cost of nuance against accuracy gains.

    “However, many people use ‘bias’ more or less interchangeably with ‘heuristic’. I admit that I do this myself, largely because it is often not clear whether a particular heuristic is biased. (Kahneman and Tversky originally focused on biased heuristics because they wanted to establish that human reasoning departs from the rational actor model, but this doesn’t imply that all heuristics are biased. One common take is that many heuristics were unbiased or at least efficient in the ancestral environment, but are no longer.) So, it is possible that the article is just using ‘bias’ in this loose sense.”

    I’d be surprised if, in reviewing Robin’s blogging record, he hasn’t used “bias” this way himself on a regular basis.

    “But I don’t really think that this is what is going on. Many people consider that it is morally wrong to use a heuristic or ‘stereotype’ in judging any other person, even if that heuristic is accurate on average.”

    I don’t think the article is saying that teachers are biased in thinking overweight kids are more likely to be untidy because it’s morally wrong, even if it is on average correct. I think the article is implying that it’s factually incorrect that overweight kids are more likely to be untidy than other kids, and that that’s why the article called it a “bias” of the teachers.

    However, I agree with you that the article seems to contain coded psychological appeals to the moral wrongness of judging generally, in its overall presentation. I just disagree that the article cedes that the teachers are factually correct in their assessments of what overweight kids are on average.

  • Carl Shulman
    In many cases the term ‘bias’ is used in reference to judgments the speaker explicitly accepts as accurate.

    In this Slate article on the provision of medical services in developing countries, the author says that cost-benefit and Quality-Adjusted-Life-Year analyses are, “inherently biased toward prevention and away from treatment,” since prevention is more effective (the article adds evidence that the effectiveness gap is less enormous than widely thought, but does not show evidence that it is nonexistent).

    ‘Bias’ here seems to be defined socially, i.e. predictably favoring one viewpoint over another in a disputed area, so that it can be seen as on balance a benefit to one coalition over another.

    The attribution of ‘bias’ to the children, without reference to accuracy, also recalls the attribution of bias to the media on the basis of ideological agreement. (

  • Hopefully Anonymous

    Carl, I think the normative assumption in the case of the article referenced in the OP is that the article was attributing ‘bias’ based on inaccuracy. In the case of the children, it was that an obese person would be worse to have than a disabled person (the choice wasn’t just overweight vs. non-overweight).

    We’re all of course tip-toing around the recent study that overweight friends may cause someone to be more likely to become overweight. We don’t know how that information interacts with the relative value of overweight vs. disabled people. Let me make it transparent to c-block someone from the opportunity to have a “gotcha” moment with a future, unwitting comments poster. >:)

  • Hopefully Anonymous

    Meta Post: I wonder what motivated me to invest this much time in a continuing disagreement on what appears to me to be an issue of at best peripheral concern to my current interests.

    I think the dynamics of the post was:
    1. Robin: analyzes article
    2. HA: disagrees with elements of analysis
    3. Norman Agrees with Robin or buttresses his argument
    4. HA: perhaps vested in my initial disagreement, articulates disagreement further, including in reference to Norman (though throws in an element of agreement too)
    5. Carl: pretty much does same thing as Norman from another angle
    6. HA: does the same thing as in #4

    I think there is an element of authority alignment and disalignment (the local authority here being Robin) independent of the search for truth, that probably impacts at least structural elements of the discussion, and at worst can impact the actual search for the best models of apparent reality.

    I think this is perhaps a major contributor to bias (defined as departing from the rational actor model, where the rational actor wants to attain the best models of apparent reality uncompromised by social cost of attempting to conduct a maximally efficient search). Some cultural anthropologies of scientists have discussed this phenomenon.

    I’d like to occasionally depart from formal discussion of a thread topic to do a meta-analysis, and I hope other posters do so too.

  • Norman Siebrasse

    H.A. – I agree that the article does not cede that the teachers are factually correct in their assessments. Nonetheless, as you acknowledge, it does seem to implicitly invoke a moral principle against employing stereotyped judgments, whether right or wrong, and the existence of that principle is what I find interesting.

    The article provides a possible explanation of that moral principle, as it very plausibly suggests that there are negative health consequences to being teased regardless of whether the stereotype on which the teasing is based is correct. But that’s really an objection to the response to the stereotype, rather than an objection to stereotyping itself, so I doubt it really underpins the moral principle.

    I should add that I don’t want to be taken to be saying that I think the moral principle against stereotyping is a bad principle. I believe there is good evidence that out-group stereotypes are systematically negative, which implies they are ‘biased’ in the strict sense. Given the remarkable human tendency to split into groups, I wouldn’t be surprised if stereotypes were generally a consequence of group formation, rather than accurate generalizations based on available information. I seem to recall that when I was in grade school we made fun of one classmade because her name was ‘Olga’ — it’s hard to see how that could be rational, and my guess is that making fun of overweight kids is likely based on the same phenomenon.

    Also, there is an obvious disadvantage even to accurate stereotypes in that they are harmful to individuals who don’t fit the (negative) stereotype, and may well justify a moral principle against them. It certainly makes the situation morally different from the use of the outside view in business decisions. But even if a moral principle against stereotypes is fully justifiable, it is still interesting to ask why, descriptively, such a principle has arisen, particularly given that it is so often honored in the breach.

  • Hopefully Anonymous

    I agree that you’re identifying and discussing the existence of some interesting principles. I’d like people to move in the direction of giving primacy towards actions that maximize my personal odds of persistence, rather than prioritizing other “moral principles”. More realistically, I’d like to coordinate with the subset of humanity similarly vested and capable to solve the collective action challenges we face towards accomplishing this goal. To the degree that various currently existing moral principles get in the way of that, I’d like to see them minimized, although there’s probably an energy cost vs. benefit in achieving that that has to be weighed against various other ways that we can maximize our persistence odds.

  • the question is: what is the evidence that judgments made about fat kids are in fact too negative on average? Without such evidence, these negative judgments should not be called “biases.”

    ah, but then you’d need more studies, longitudinal ones. Unfortunately, “bias” internalized by fat kids does more to undermine their self-image, arguably affecting their future success, you know, self fulfilling prophecies? In which case, there’d be no “bias”, lower expectations having been artificially reinforced through social mechanisms. If one were to examine the pejorative nature of these assessments in relation to the fat kids’ *potential*, you’d find bias certainly enough. I don’t have any statistics to prove it of course, only one case study. Weighing -no pun intended- the assessments made of me in childhood as compared to what I’ve become (after losing 150 lbs), these could certainly be described as bias. Except of course, I’m still not tidy so maybe it isn’t bias after all :).

  • michael vassar

    Norman: Rationality != niceness, and evolution is downright nasty. Making fun of Olga could have been rational for those who did it because they were evolved to find it fun to identify an out-group and attack it as a way of building affiliation.

    Stereotypes surely have some systematic bias towards negativity, but they also seem to influence which traits are valued, e.g. to influence what vector is defined as negative. For instance, US and Japanese propaganda during WWII largely agreed on how the US and Japan differed, it’s just that each defined how they were as “good”. “Market dominant minorities” are generally recognized by outsiders as possessing traits that make them more suited to financial success in modern societies, but the outsiders emphasize the negative aspects of those traits.

  • michael vassar

    Another truly outlandish instance of attribution of bias to children (or rather, to teens) without reference to accuracy comes from the work of Frank Farley and Valerie Reyna, who have asserted that teens are “too rational” about risk and that they need to be taught to make risky decisions using the less rational “system 1” cognition.

  • Norman Siebrasse

    Re meta-analysis: Maybe authority alignment is at work, but there is a simpler explanation that also fits your points 1-6, namely that your initial critique was a bit off the mark (though not wrong), which is why Carl and I both responded along the same lines. I try to employ the principle of charity (see the wiki entry) and put the best spin on what the other person is saying in order to prevent digressions over misunderstandings. I think this is particularly useful in responding to Robin’s posts, which I take as being intended to get the ball rolling on an interesting topic rather than providing a definitive analysis. (But maybe that’s just another example of authority alignment.) And I do think that this particular discussion has been very useful in the search for truth, in that it has clarified an ambiguity in the term ‘bias’ that I’ve noticed in much of the discussion.

    With all that said, I doubt that this kind of meta-analysis is worthwhile. Given that this blog is about bias, meta-analysis is likely to involve accusing each other of various kinds of bias. I doubt this will be more interesting than identifying similar types of bias in the news etc, and it is certainly likely to be more divisive. For example, I think that this comment of mine is the least interesting and most divisive thing that I’ve posted on this blog, and yet defensive ego-centrism bias has made me feel compelled to post it nonetheless.

  • Hopefully Anonymous

    Michael, it would be interesting to see how Chinese were stereotyped (positively or negatively) in Japan vs. the pacific island nations in the 20th century, because in one sphere they were not seen as a market dominant minority and in another sphere they were. Perhaps the same applies to South Asians in East Africa vs. the Caribbean (although recently south asian status in the Caribbean seems to have dramatically improved). I’d be surprised if this type comparative work hasn’t already been done.

    As for rational reasons to pick a group to stigmatize to construct insider/outsider status (that favors the stigmatizers), I think it’s an interesting question as to how people get picked. I can think of various theories, and probably there are multiple causal factors. Generally, I think the stigmatized groups are less able to counter the stigmatization (due to lower intelligence, fewer cultural resources to counter the stigma, unique collective action barriers, or because the principle on which they’re stigmatized has the added value of being true -such as if overweight people really were, on average the worst friends, and it really was cost-prohibitive to assess a given overweight person further before choosing not to befriend them), but it’s also possible in some cases they are compensated in other ways for the stigmatization. For example, I could see market dominant minorities actually favoring a level of group stigmatization if it will help preserve endogamy and reduce competitors from participating in advantageous behavior.

    I recall reading a paper about a minority ethnicity within a region of Kenya, that wouldn’t participate in redistributive feasts that other ethnicities in the region participated in, even though there was a social cost in not hosting redistributive feasts. It could in theory be rational for all parties, for the majority to consider the insular minority group an outgroup, while the minority group maintains its endogamy and wealth accumulation advantage, so that the system as a whole is more productive and competitive than competing systems. But in practice it seemed to me that the wealthier minority group was getting the benefit of the bargain. I’ll try to dig up the article/study if time permits.

    It just occurs to me that overweight persons may be compensated in at least one other way: they don’t have to practice abnegation or engage in rigorous excercise regimes. So an element of their stigma may be similar to stigma put on hedonists and people who chose not to go to college- there may be an informal assessment that the greater advantage they’re taking in one area of life than the majority makes it fair for them to shoulder stigma burden rather than the rest of us. This is in contrast to stigma placed on people with factors perceived to be more immutable, such as having dark skin, or being short.

  • Norman, I agree that many people think it wrong to use certain clues to evaluate people, but I want to resist accepting the word “bias” to describe such a situation; we need the word “bias” to refer to avoidable errors in estimation.

    Kathleen, there are surely self-fulfilling prophecy elements of making negative estimates about people. But unless you want to argue that no one should ever have a negative estimate of anyone, this doesn’t imply that fat should not be the basis of a negative estimate. The question is why the fat-based self-fulfilling prophecy effects are more problematic than any other such effects.

    Hopefully, check your email.

  • Hopefully Anonymous

    I think your post did make an important an obvious (but should have been stated) point: that any agreement or disagreement could have been done simply because each poster believed that someone was correct or incorrect on a particular point. I think that’s the starting point claim of every point that’s not transparently performative.

    However, I think it also illustrates the value of meta-analysis even of our own, real time posts, because we’re probably regularly wronger than we need to be not just our disagreements and agreements, but our reasons for doing so and our reasons for the structure in which we do so.

    Meta-analysis might be better done through (putative) anonymity, because I acknowledge our human limitations at engaging in mutual critiques, given the various social costs associated with it.

    This blog is about overcoming bias, and I think it’s not a departure from the mission to engage in this meta-analysis in the commentary. Plus I think the regular participants and all the named contributors are mature enough to handle it.

  • Norman Siebrasse

    I agree that moral attitudes are evolved, and that there is no particular reason for them to be nice. I believe there is fairly good support for the view that intergroup competition leads to intergroup hostility: see Rabbie, Ch. 7 in Harcourt & de Waal ed.s ‘Coalitions and Alliances in Humans and Other Animals” (1992). That is, when your are already in a group which is actively competing with another group you will tend to develop negative views of the out-group. But my ‘Olga’ example was intended to illustrate a different phenomenon — spontaneous arbitrary group formation. The in-group cohesion argument is less persuasive in this context, since there is an obvious cost in the form of foregone opportunities to interact / trade. Put another way, one could say that group cohesion was diminished, not strengthened, because even if the in-group became more cohesive, the larger society (the class as a whole) became less cohesive. It’s far from obvious that this would bring a net benefit to the in-group, particularly when the person was excluded on a purely arbitrary basis. In the ancestral environment, if our group were attacked I would suspect it would be less likely to survive because whatever increase in cohesion we gained among members of the in-group would be offset by the fact that our group itself was smaller. Of course, it’s possible that the cohesion effect would outweigh the division effect, but it seems to me that if group cohesion were really at stake we would have evolved a way to achieve group cohesion that didn’t require sacrificing group strength. Why wouldn’t we have evolved to increase cohesion through inclusive group ceremonies, rather than by creating internal divisions?

    Now, I’m not really sure that spontaneous arbitrary group formation actually exists as a phenomenon. Maybe that particular experience was just aberrant and really isn’t an example of any general phenomenon. It is also possible that I’m misremembering the incident and that we had some good reason for disliking Olga apart from her ‘funny’ name. Or it could be that non-adaptive arbitrary group formation in non-competitive situations sometimes occurs as a result of over-generalization of adaptive out-group hostility in competitive situations.

    Anyone have thoughts as to how common spontaneous arbitrary group formation might be? The best historical example I can think of is the Great Schism between the Roman and Eastern churches over the filioque clause of the Nicene creed — is the Holy Spirit from the Father or from the Father and the Son? This was no doubt in large part a power struggle between bishops, but why did the followers find this kind of division motivating?

  • Hopefully Anonymous

    My earlier comment (also in response to Michael?) addressed a lot of your concerns about the mechanism of ingroup/outgroup constructions and how they arise directly, I think it was removed for being too long. I may create a quick anonymous blog and then repost a link to it.

  • michael vassar

    “Why wouldn’t we have evolved to increase cohesion through inclusive group ceremonies, rather than by creating internal divisions?”

    I think because in our adaptive environment we never lacked for enemies, real or imagined, so why bother, as it were. Also, assuming no strong group selection, there may be a prisoner’s dillemma at work. Prior to selecting an out-group member, the risk of becoming such an out-group is spread among the group, with significant expected cost. When someone is chosen, this reduces risk for all group members by an amount greater than the loss from trade, even if increased cohesion doesn’t make it an absolute benefit.

  • Norman Siebrasse

    But precisely because we never lacked for enemies, it would have been all the more important not to weaken the group by creating internal division.

    The group weakening effect of picking on an arbitrary group member is clear and direct. Why would picking on a group member evolve as an indirect way of creating cohesion when the immediate direct effect is to reduce cohesion? Maybe there is some weird feedback mechanism a la runaway selection, but those kinds of arguments are not easy to make out.

  • Doug S.

    “Spontaneous arbitrary group formation” could be a result of within-group competition for status. Attacking a group member with low status signals that you have more status than the member that was attacked. (The high status kids in school aren’t the ones that pick on the outcasts. They don’t need to. It’s the medium status ones, who are afraid of becoming outcasts themselves, that do the worst bullying.) Therefore, to raise your own status in a group with ill-defined status, find someone to attack, attack them, and get others to support you in the attack.