Confirmation Bias Anchoring and Prior Beliefs

The Strength of Anchoring: Finals Day at the Investment Committee

If you've spent time around asset manager searches, you know the drill. Your marketing passes the consultant screen. They ask for diligence materials. They narrow the list. Then, the finalists get invited for a final pitch to the selection committee. The committee's mandate is to score every presentation objectively against the investment rubric. Everyone nods solemnly. It's a process.

At the beginning of my career, I was required to attend monthly investment selection meetings by one of our larger limited partners. This required me to sit through more of these than I care to admit. And what I thought at the time was a painful waste of time became a master class in fundraising and complex decision making. Add to that, the experience of taking fund managers to investment conferences where they are one of 30 presenting over three days. I've seen both sides.

Here's what the selection rubric never captures: the first team to present sets the bar. Their presentation becomes the one everything else gets compared to. Their performance history becomes the story. By the second presentation, committee members are asking questions the first firm planted... "How does your downside capture compare to {insert firm name}?"

At some point, usually around the third manager presenting, a frontrunner quietly forms. Nobody announces it. Nobody needs to. Just watch the questions change. The frontrunner's rough quarter becomes a disciplined process working through a tough market. Another firm's identical rough quarter becomes a flag somebody should probably dig into. Same drawdown, different adjectives.

By the debrief, the committee is sure it weighed four managers on the merits. What it mostly did was allow the first presenter to anchor all the following conversations before the first coffee break.

The minutes describe a rigorous process, rather than the brilliant anchoring of the first presenter.

Everything in this entry is about that finals day, explained by fifty years of research.

The Anchor Sets the Frame

Anchoring, documented by Amos Tversky and Daniel Kahneman in "Judgment under Uncertainty: Heuristics and Biases" (Science, 1974), shows how an early number drags a judgment toward it even when the number carries no real information. In one study, two groups of students had five seconds to estimate the product of a string of numbers. One group saw it written in ascending order, starting 1 × 2 × 3. The other saw it descending, starting 8 × 7 × 6. Same math. The ascending group's median guess was 512; the descending group's was 2,250. The true answer is 40,320. Both groups missed badly, but they missed in the direction their first few numbers pointed.

The finding reaches well past the math. Anchoring holds even when the anchor is obviously arbitrary, even when people are warned about it, and even when they swear it had no effect. The first frame a decision maker meets becomes the reference point for everything after it. That frame might be the opening definition of the problem, the first fee quoted, or the first impression of a firm. It might be whichever manager drew the morning slot on finals day.

And sophistication doesn't buy an exemption. I've watched selection committees full of MBAs behave like those students with the number strings, and I've anchored a few rooms myself when I got to present first. People do adjust away from the anchor, but never far enough. Whoever sets the bar sets the terms of the comparison.

Confirmation Bias: The Mind Defends What It Already Believes

Once a belief is in place, it starts gathering support for itself.

Raymond Nickerson's "Confirmation Bias: A Ubiquitous Phenomenon in Many Guises" (Review of General Psychology, 1998) is the definitive summary of how. Pulling together decades of evidence, Nickerson showed that confirmation bias runs as a whole family of errors. People go looking for information that fits what they already think, read mixed evidence as support, count the confirming cases more heavily than the contradicting ones, and remember the hits while forgetting the misses. A buyer who has quietly decided a firm is the frontrunner will read each later interaction as one more sign they were right. The frontrunner's drawdown is discipline. Yours is a red flag. Same drawdown, different adjectives.

Nickerson's most useful point is that none of this is on purpose. People aren't choosing to cheat the evidence. The bias runs without intent and mostly without awareness, which is why telling someone to be objective accomplishes so little. He found the same pattern in flawed medical diagnoses, in scientific mistakes, and in historical persecution... careful, educated, well-meaning people, all running the same software.

I'd love to tell you experience makes you immune. It doesn't. Mostly it upgrades the quality of your rationalizations.

Identical Information, Opposite Conclusion

The sharpest demonstration of the whole problem belongs to Charles Lord, Lee Ross, and Mark Lepper. In "Biased Assimilation and Attitude Polarization" (Journal of Personality and Social Psychology, 1979), they recruited people with strong opposing views on capital punishment and handed both sides the same packet: two studies of the death penalty's deterrent effect, one supporting it and one undercutting it, complete with the methods behind each.

Balanced evidence should have nudged the two sides toward each other.

It did the opposite. Each group decided the study that agreed with them was the better-designed and more convincing one, then picked apart the other on methodological grounds. After reading the exact same balanced packet, both sides walked away more certain of their original position than when they sat down. The same information drove opinion further apart. That's biased assimilation, and it's the experimental proof that evidence never gets a reading independent of the frame it lands in.

Sit with that for a minute, because it's the one that should bother you. The instinct, when a buyer has you pegged wrong, is to send better evidence. Lord, Ross, and Lepper showed what happens to better evidence: the frame eats it.

The Intuitive Psychologist

Lee Ross gave the bigger account of why people filter this way. In "The Intuitive Psychologist and His Shortcomings" (Advances in Experimental Social Psychology, 1977), he described the ordinary person as an amateur psychologist, always explaining behavior, assigning causes, and predicting what others will do, but doing it in predictably skewed ways. This is where Ross named the fundamental attribution error: the habit of chalking up someone's behavior to their character while underrating the situation they were in. We watch a person trip and think clumsy, not wet floor.

For a library about how buyers judge firms, that's a working problem, and a daily one. A decision maker watching a firm forms a fast snap theory about who these people are, sharp or sloppy, from a thin slice of behavior, then reads everything afterward through it. One slow reply in the first week becomes proof of a disorganized shop. I've been on both sides of that one. The first impression arrives before the evidence, then becomes the lens the evidence is read through.

Why This Belongs in Decision Architecture

Decision Architecture says it directly: decision makers arrive carrying beliefs, preferences, fears, and assumptions, and they read new information through those filters. This research is the proof that the claim is literally true, and here the science stacks cleanly. Anchoring shows the first frame gets special weight. Confirmation bias shows a held belief defends itself. Biased assimilation shows that even balanced proof gets bent toward the side the receiver already favors.

That's the scientific basis for Decision Architecture's insistence on sequence. If early engagement frames the problem badly, or frames it in a way the prospect doesn't recognize as their problem, later proof never gets a fresh read. It gets absorbed into the frame already sitting there, where the favorable parts get waved through and the unfavorable parts get picked apart. Relevance has to come before proof, because proof dropped into the wrong frame gets swallowed by it. That ordering comes from how minds handle evidence, which makes it a constraint, and constraints don't care about your content calendar.

The machine layer raises the stakes on going first. In AI-assisted research, the first summary a buyer reads, the overview that defines the category and names a few players, becomes the anchor for everything that follows. Then confirmation bias goes to work, and the buyer favors whatever fits that opening frame. A firm that's missing from the first machine-generated impression, or cast badly in it, gets judged, if at all, through a frame built without it. First position has always mattered. When a machine sets the frame, the effect compounds.

The Marketing Read

Nobody gets a fresh read. Not you, not your proposal, not your best case study. By the time your proposal, deck, or case study arrives, a frame already exists, and your material will be interpreted through it. So the real contest is over who builds the frame.

Where you can, go first. The first definition of the problem, the first number on the table, the first firm in the door. Early position isn't about being remembered; it's about becoming the yardstick everything else gets measured against. If showing up first isn't possible, at least find out what the existing anchor is before you present, because you're negotiating against it whether you acknowledge it or not.

The chair decides the play here too. If you're presenting first, you're not just presenting, you're writing the rubric everyone after you gets graded on, so spend your opening minutes defining the problem and planting the questions you want asked of your competitors. If you drew the late slot, walking in with a better version of the first presenter's argument is a losing hand. Change the comparison instead: a different problem definition, a different measuring stick, a reason the morning's yardstick measures the wrong thing.

When the frame already casts you wrong, resist the reflex to pile on proof. Biased assimilation means your best evidence gets read as an exception, or picked apart, or bent into support for the existing label. Change the frame first... reframe the problem, the category, or the comparison set. Then bring the evidence, once there's a frame that can receive it.

And check the machine layer before you spend anywhere else. The first impression of your firm is now written, more often than most firms realize, by an AI summary you've never read. If that summary defines your category without you, every human bias in this entry starts working against you before a single person has formed an opinion of their own.

One last thing, and I'll admit it stings a little coming from someone who sells marketing for a living: most firms respond to being misjudged by producing more material. The research says the frame decides what the material means. Fix the frame, or keep shipping decks to a committee that elected its yardstick before lunch. The minutes will say you got a fair look. The minutes always do.

Primary Sources

• Charles G. Lord, Lee Ross & Mark R. Lepper, "Biased Assimilation and Attitude Polarization: The Effects of Prior Theories on Subsequently Considered Evidence." Journal of Personality and Social Psychology, vol. 37, 1979, pp. 2098–2109.
• Amos Tversky & Daniel Kahneman, "Judgment under Uncertainty: Heuristics and Biases." Science, vol. 185, 1974, pp. 1124–1131. (Source of the anchoring-and-adjustment studies.)
• Raymond S. Nickerson, "Confirmation Bias: A Ubiquitous Phenomenon in Many Guises." Review of General Psychology, vol. 2, 1998, pp. 175–220.
• Lee Ross, "The Intuitive Psychologist and His Shortcomings: Distortions in the Attribution Process." Advances in Experimental Social Psychology, vol. 10, 1977, pp. 173–220.