Bryan Caplan  

Bias Against Speculation?

Me on Fox News Channel Tonight... A Fable of Two Countries...
Robin Hanson recently discussed research finding that atheists are widely disliked because people see them as less trustworthy.  He then posed a logical followup question: "So are atheists actually less trustworthy?" and offered a tentative answer: "I'd guess that they are, but that the difference is less than people think."

To my surprise, Robin's answer sparked quite a backlash.  From the comments on Overcoming Bias:
I'd be very, very hesitant to make that claim without some actual proof. (Brett)
"Guessing" that Athiests are less trustworthy than the religious and requiring evidence to prove otherwise (rather than starting from a null hypothesis that there is no difference) is bigotry, plain and simple. Replace "atheist" with "black" in the above post and comments if this isn't readily clear to you.

I see clearly that this blog is run by a bigot, and that many of his followers are as well. I'll be unsubscribing now and hope anyone else who feels that they can be "good without god" does as well. (Locklin)

Time to change the blog title from 'overcomingbias' to 'overwhelmingbias'?

Consider for a moment who your audience is, and consider for a moment how many of them you just shocked with an open display of uninformed prejudice. This blog is supposed to be a worldwide focus for the principle of overcoming bias? I think not. It has strayed so far from it's original goals in recent months that I no longer see a purpose in remaining here. (Anonymous Cow)

You could interpret Robin's critics as simply calling for greater sensitivity.*  But I sense a more fundamental complaint.  Namely: Robin is "biased" because he freely shares his speculation.  The fact that he explicitly distinguishes between (a) claims backed up by research, and (b) his best guess, is no excuse.  A truly unbiased person, apparently, would either base every assertion on research or remain agnostic and silent.  (This position is similar to one held by most academic economists, but they're more likely to distinguish between what they can "say as scientists" and what they really think).

But what's wrong with freely sharing your speculation?  On the Bayesian definition of rationality, beliefs are inevitably a combination of your initial speculation ("prior") and the evidence.  You cannot not speculate.  Guessing that theists and atheists are equally trustworthy is just as speculative as guessing that they're not.  Given this inevitability, it seems better for people to expose their speculation to public criticism instead of pretending that their beliefs are based on "evidence alone."  If you want to "overcome bias," you will reward candor, not feigned agnosticism.

That's my speculation, anyway.

Challenge: As long as speculation is clearly labeled, shouldn't truth-seekers want people to freely speculate?  If not, why not?

*He actually updated his post to note that he is an atheist, appealing to the norm that you can speak more frankly about groups you identify with. 

COMMENTS (14 to date)
John Thacker writes:

Robin then proceeded to post a lot of studies that back up his basic conclusion-- atheists probably are less trustworthy to some degree, but not to the degree that theists seem to assume.

I'd actually be somewhat interested to see studies that distinguish between first and second generation atheists. There are plenty of people who accept the axiomatic moral rules of their parents while eventually rejecting the theism that went along with it.

First generation or not, atheist or not, people who have a fundamentally irrational or axiomatic conception of morality probably are somewhat more honest, particularly in non-repeated games.

Peter writes:

I remember reading that post. As an atheist I went into defense mode and thought that surely atheists can't be any less trustworthy. But after a few seconds it didn't seem so unlikely any more. I never read the comments, but I'm surprised to hear that it generated so much hostility.

Although I've seen quite a lot of bigotry from atheists. I suppose living in Sweden makes me less prone to such hostility since we don't seem to have that many 'crazy' religious people here.

Josh G writes:

I understand the name "overcoming bias" as saying that we all have biases, and that our struggle to overcome them is a constant process. I think that is one of the things that the "overwhelming bias" commenter misses. Even if Robin's trustworthiness hypothesis represents an unfair bias, the first step to getting past that bias, would be to explicitly identify it.

Margaret writes:

First off, I have not read the blog that has been mentioned. Second I am a Christian, or theist if you will.

Now I find that the comments quoted show people who are themselves very biased. It seems that they want everyone to think their way, and if you do not agree with them, then you are biased.

The point is that the question was thrown out there to invite discussion, and not for the purpose of people becoming petulant.

If I was to state a point of view, I would actually state that religious affiliation has nothing to do with the matter. I would base my own argument upon the fact that we are all prone to sinning, and that a lack of trustworthiness tends to point to the condition of being in a state of sin. This applies equally to Christian, atheist or any other religion or designation.

People who proclaim themselves as Christian can be untrustworthy, and I would point to those big tent preachers as being amongst some of the most untrustworthy. I could name a number of names in that direction. I would also point to those Catholic priests who had betrayed the peoples' trust in them as pastors and leaders of a given community. After all, by their actions they breached the trust that was shown in them.

The real bias seems to have come from those who were responding in the comments, because they were the ones who were unable to overcome their own intolerance to be able to think through with an appropriate response as to why such research might in fact have been wrong.

steve writes:

Not hot enough for you?

Try working through the same discussion using eugenics. For example, while intelligence may be hard to select for certainly hieght and some non-objective measures of beauty could be bred.

Lars P writes:

It often helps to turn these questions around, like this:

Q1: Are atheists are less trustworthy than theists?
Q2: Are trustworthy people more likely to be religious?

Q1 and Q2 are the same question, but can cause quite different ways of thinking about the issue, as well as different emotional responses.

John T. Kennedy writes:
As long as speculation is clearly labeled, shouldn't truth-seekers want people to freely speculate?

Yes, of course. As an atheist I wasn't the least bit offended by Hanson's speculation. It's true or it's not, and from a standing start I have no idea which.

If it turns out that atheists tend to be less honest than theists that doesn't make me any less honest.

Thomas Boyle writes:

It seems to me that all atheists are trustworthy when they state their non-belief - and, indeed, they are incurring a social cost to state it.

On the other hand, it also seems to me that many, if not most, theists display a disconnect between their stated beliefs and their actions, leading me to believe that they are not trustworthy in their statement of belief.

Just sayin'.

jva writes:


Q1 and Q2 are not the same.

It can be true that atheists are more trustworthy than religious people but at the same time trustworthy people are more likely to be religious than atheists. Each of those answers tells nothing about the other. How exactly is that possible is left as an exercise.

RonB writes:

The basis of honestly is not necessarily based on a belief in god or not, but rather the ability of the individual to rationalize their action. An ability in which neither group is deficit. One would also have to differentiate between interactions within the group or with those outside the group. To make a generalization, some of the most trust worthy people I know have been atheist and some of the most dishonest religious, but I would not use a belief in god or atheism as a criteria for trusting any given individual.

Dan Carroll writes:

Robert speculated/guessed that atheists are “less trustworthy” than religious people, citing in a later post that he himself is an atheist to justify his speculation. As a Christian, I have a hypothesis that he is wrong. Here is my reasoning:

Theoretical points favoring his speculation: The religious believe they are held accountable to a set of moral standards. Sometimes(certainly not always) a combination of guilt and fear is used to keep the masses in line. In most cases where guilt and fear are not used or not effective, the religious are issued commands by a higher authority which they respect, which provides motivation to act accordingly. The atheist, however, has none of these to guide and motivate behavior. Without the proper incentives, as economic theory goes, the atheist is more likely to act in an untrustworthy manner (whatever that means).

Theoretical counterpoints: In the US, atheists are a small counter-cultural minority, probably not heterogeneous, and are mostly highly educated. The religious, which includes the nominally and barely religious, are a large and highly diverse group. Therefore, the “morality” of religious folk is likely to approximate the average person, while the “morality” of atheists is not likely to approximate the average. Since education is generally correlated with self control, the atheist is more likely to exercise self control. Furthermore, individuals who have damaged their lives through untrustworthy behavior are more likely to be attracted to structure and discipline of religion, thus skewing the religious average downward. If one excludes the nominally religious and only studies the devout, then the devout are also a minority. In that case, one could hypothesize that both the trustworthy behavior and the average education of the devout group rises (the latter has been established statistically for the U.S.). Econometric studies could isolate both years of education (easy to idenfity) and rate of church attendance (extremely difficult to identify), but without evidence, establishing a hypothesis is tricky.

Generally, I think the complexity of the issue and the theoretical counterpoints weigh heavily in favor of refuting Robin’s speculation. However, perhaps this hypothesis just reflects my own bias.

caveat bettor writes:

This reminds me of the Bible story of Daniel, who speculated that a kosher (paleo?) diet would provide healthier outcomes than an ancient Middle Eastern think tank institutional cafeteria, and proposed an experiment (which proved Daniel's speculation).

I speculate that some Christians are afraid. I speculate atheists get that way, too. Too bad.

Jon writes:

I want people to freely speculate, but if they do and it reveals overt racism I will probably voice my objection.

Don't people speculate about blacks? Don't they engage in Bayesian inference when they consider blacks? I suppose they do. And when they do and they expose that their priors involve the assumption that blacks are drug users (in fact they aren't a lot different than whites, though they are incarcerated more) or that they are genetically predisposed to stupidity (also bogus, though it is true that their performance in academics lags whites and I would suggest it is due to external factors) I would be inclined to condemn the statement even though it truthfully reflects the speakers opinion.

Some people genuinely believe in racist conclusions.

José A. Azpúrua writes:


We would have to determine, first, what does trustworthiness mean to us in the sentence we may have used.

Also... in which sense do we want to determine the trustworthiness.

A hired killer IS a trustworthy person if he/she performs the requested, and paid for, murder.

Human communication is, thus, extremely difficult to achieve with precision and clarity.

Quite often, discussions drift off course; and people achieve no agreement or consensus because of the differences in meanings, or pursued objectives.

Comments for this entry have been closed
Return to top