Note: this essay is presently published as a “Comment” in the relevant part of the buddy-blog Notes on Liberty.
I am glad Brandon has drawn attention to the confirmation bias in Notes on Liberty. (“Origins of Terrorism in the Middle East.”) The words refer to the universal tendency of human beings to notice and to remember facts that support what they already believe to be true to the detriment of information favoring different and opposing views. Thus someone who believes that human activity has been causing global warming will collect and recall unusually hot days and he will tend to discount unusually cool days.
The confirmation bias is the bane of casual discussion such as are conducted in coffee shops, around the kitchen table and, in immense numbers now, on the web. Unfortunately the confirmation bias also frequently affects adversely empirical research designed to protect against biases in general. Scholarly submissions that present disconfirming evidence regularly have to jump higher hurdles than scholarly papers that extend orthodoxy. Yet, good methods afford partial protection in the social sciences and systematic critique also limits the damage to truth caused by the confirmation bias and by other biases.
But well designed and well conducted social science is expensive and time consuming. In the meantime, we have to live; we must make decisions, we cannot avoid choices. We are not able to wait for everything to become the object of a good study and for the study to be published in a respected journal to do what we have to do. Exaggerated deference to rigorous empirical studies is tantamount to delivering the floor to the most emotional, to the least rational, to the blindest fanatics among us. Like it or not, we must rely on anecdotal evidence most of the time. Yet, anecdotal evidence must, in time, give way to good studies published in a respected scholarly journals.
So, what is to be done about the confirmation bias usually associated with the gathering of anecdotal evidence? First, obviously each commentator of any political fact or perception must exercise extreme self-discipline in this respect, knowing that confirmation bias is not an accident but a normal tendency of the human mind. It helps a great deal if the commentator knows he is addressing an audience, a public, that praises intellectual honesty.
Secondly and most importantly, arguments should be subjected to criticism. I may easily, and in all honesty, be blind to my own confirmation bias but disinterested observers, and especially, adversaries, will ferret it out in no time. It’s also important to have reasonably public venues where biases in general and the confirmation bias in particular can be called out. I believe that Notes on Liberty and my own personal blog, Factsmatter, are two such venues. They should not be taken for granted. They may be the few exceptions among hundreds or thousands of blogs.
The context that motivated Brandon, the founder and co-editor of Notes on Liberty, to denounce my alleged confirmation bias is a comment of mine on his essay: “Origins of Terrorism in the Middle East.” In my comment, I take exception to his dismissal of the idea of “Islam’s violent penchant.”
I believe that, in fact, Islam has inherently violent tendencies. (I recognize at the same time the overwhelming peacefulness of the overwhelming numbers of Muslims.) In support of an assertion to the effect that Islam has a violent penchant, I list a number of violent practices which I argue are especially associated with Islam. Incidentally, I always mean “Islam the culture.” I am not a theologian able to discuss what Islamic scriptures and Islamic doctrine “really mean.” I am only able to observe reality on the ground.
Since my observation is neither exhaustive nor randomly conducted, the risk of confirmation bias is quite real. There is danger that I assign unconsciously to practitioners of Islam objectionable practices that are just as common among followers of other religions. It would be like treating Christians, for example, as especially likely to abuse alcohol as compared to Muslims. (And how silly can one get!)
In the situation at hand, I made the claim, among many others, that the only people who condemn to death by stoning women they judge adulterous do it in the name of Islam, (in the name of Islamic law specifically), and that they have Muslim names. Incidentally, this is a good point to correct myself; I should have said, “ in the last one hundred years.” Going back to what I asserted above, the main corrective to selection bias is criticism. In this case, I expect Brandon – and anyone else who is so moved – to point out to me the group or groups unassociated with Islam in any way who affirm that public stoning to death is an appropriate way to deal with adulterous women.
I will be waiting.
It seems to me that there are three major vices that regularly interfere with intelligent people’s exercise of reason. One is political correctness. The second is the desire to simplify at all costs issues that are inherently complex. The other it a perverse wish to insist that things cannot be as simple as they seem on the surface, that observable reality only masks a deeper, more correct interpretation of real reality.
My qualifications toward discussing these issues are in my vita, linked to this blog.