Confirmation Bias

Confirmation bias is the tendency to seek out, interpret and remember information in a way that supports an existing belief or hypothesis. Raymond Nickerson's review, "Confirmation Bias: A Ubiquitous Phenomenon in Many Guises" in Review of General Psychology 2(2), remains the standard survey and its title captures the important point: it is not one behavior but a family of them, running from which sources a person consults to how they weigh what those sources say. In financial decisions it is most damaging not because it makes people reason badly, but because it quietly determines what they are reasoning about.

The foundational demonstration is older than most of behavioral finance and is about search, not stubbornness. Peter Wason's 1960 study, "On the Failure to Eliminate Hypotheses in a Conceptual Task" in the Quarterly Journal of Experimental Psychology 12(3), gave participants the number triple 2-4-6, told them it followed a rule, and invited them to propose further triples to test. The rule was simply "any three ascending numbers." Most participants formed a narrower hypothesis, such as "even numbers increasing by two," and then proposed only triples consistent with it, so the tests they ran could confirm their guess and could never disconfirm it. Nobody in that experiment reasoned badly. They gathered the wrong evidence.

A second, better-known finding needs a more careful statement. Lord, Ross and Lepper, in Journal of Personality and Social Psychology 37(11) in 1979, reported two things from a study of attitudes toward capital punishment. Biased assimilation, the finding that people rate evidence agreeing with their view as better-conducted and more convincing than identical-quality evidence disagreeing with it, has held up well. Attitude polarization, the more dramatic claim that being shown mixed evidence pushes people further apart, has not replicated reliably when later researchers measured actual attitudes before and after rather than asking participants how much they thought they had changed.

That distinction is genuinely useful and it is good news. The popular version of confirmation bias implies that presenting someone with contrary data will entrench them, which counsels against ever trying. The evidence supports the weaker and more workable position: people evaluate disagreeable evidence ungenerously, but showing it to them does not usually make things worse. What it does not do reliably is change their mind on its own, which is why the useful intervention happens before the position forms rather than after.

Confirmation bias operates at three points in a decision, and they are worth separating because they need different fixes.

1. Search. Which questions get typed, which sources get opened, which people get asked. This is where most of the damage happens and it is almost entirely invisible from the inside, because the evidence that was never gathered leaves no trace.
2. Evaluation. Supporting evidence is accepted at face value; contrary evidence is scrutinized for methodological flaws until one is found. Both feel like careful thinking.
3. Recall. Confirming episodes are remembered more readily than disconfirming ones, so the accumulated impression drifts even when each individual judgment was fair.

Four financial versions of the same shape:

• Reading only bullish commentary on a holding you already own, and treating the resulting consensus as independent.
• Checking a fund's three-year return, which looks good, and not its ten-year return, which does not.
• Asking a second professional for an opinion, and then a third, until one agrees with the plan you already had.
• Reading a market forecast that matches your positioning as analysis and one that contradicts it as noise.

A hypothetical, resolving a process question rather than an amount. Two investors are deciding whether to hold a concentrated position in their employer's stock. The first researches the company, finds the case for holding persuasive, and holds. The second writes down, before researching anything, the three things that would make them sell: a specific deterioration in the business, the position exceeding a stated share of net worth, or a change in their own job security. Then they research. The second investor can still decide to hold, and often will. The difference is that they have built a test they could fail, which is the whole content of the countermeasure. Deciding in advance what would change your mind is a structural fix; resolving to be open-minded is not.

Pros

• Identifying it points at a specific and fixable step, the search, rather than at a general failure of judgment.
• The countermeasure is cheap and concrete: state your disconfirming evidence before you look for anything.
• Knowing that biased assimilation is well supported while polarization is not means contrary evidence is worth presenting, which is a more useful conclusion than the popular version allows.

Cons

• It is invisible from the inside. The evidence you did not gather does not announce itself, and the search feels thorough.
• The label is often used to dismiss a disagreement rather than to examine one's own process, which is both unpersuasive and a misuse of the concept.
• Not every consistent view is confirmation bias. Sometimes the evidence really does point one way, and treating consistency as automatic evidence of bias is its own error.
• Awareness does not remove it. Only changing the procedure does.