Executives are not free from cognitive traps. If anything, the larger the decision, the larger the effect of the bias. The cherry-picking we see on the materials-review floor, and the cases where a product that should be reined in somehow is not — much of this is not the ill will of an individual, but the result of a cognitive distortion shared across the organization. This piece sorts out two traps, confirmation bias and sunk cost, and shows why an independent review becomes the device that corrects for them.
01Cherry-Picking Is an Organizational Cognitive Distortion
Confirmation bias is the tendency to gather only the data that fits your expectations and to discount the data that does not. Trial results favorable to your own product catch the eye; unfavorable subgroup analyses and cautionary notes recede into the background. The cherry-picking of materials can be read as this tendency made visible on paper.
What deserves attention is that this is not the problem of a single person on the floor. The expectation that this drug works is shared from management through development, sales, and the field. The stronger the shared expectation, the more the entire organization tilts in the same direction. So blaming an individual will not make the distortion disappear. The distortion lives in the structure of the organization.
02The Larger the Investment, the Weaker the Brake
The sunk-cost effect is the tendency to cling to a decision the more has already been spent on it. The larger the investment that cannot be recovered once reversed, the more people hesitate to withdraw or restrain. Economically, sunk costs should be left out of the decision, yet the mind does not work that way.
In pharmaceuticals, this produces a pattern in which the products that absorbed enormous development costs are precisely the ones where the compliance brake works least well. The larger the sum that must be recouped, the slower the decision to rein in aggressive claims. The pressure to recover the investment clouds the very risk that should be visible. If, as we saw in Vol. 5, the design of evaluation and reward is skewed toward short-term sales — an agency problem in plain form — this clouding grows thicker still.
Confirmation Bias
Makes you gather only the data that fits your expectations. Results favorable to your own product come to the front, and unfavorable information recedes. It is the source of bias in materials.
Sunk Cost
The larger the spend, the more it delays withdrawal or restraint. The pressure to recoup clouds the risk that should be visible.
Independent Review
An eye placed outside the parties. Precisely because it does not share the expectation, it can name the bias and the clouding. An institutional mechanism for correcting bias.
03Independent Review Is the Institutional Device for Correcting Bias
It is hard for an organization to correct its own distortion by itself. Parties who share the same expectation cannot see what is biased. That is exactly why internal controls and third-party checks are needed. These are mechanisms inserted from outside in order to correct for bias.
The MSA Guidelines (the guidelines on sales information provision activities) require independence and monitoring of the review for precisely this reason. Article 362, Paragraph 4 of the Companies Act requires the board of directors to build an internal control system, and the Daiwa Bank shareholder derivative suit established the liability for neglecting to build one — both rest on the same idea. Rather than relying on the good faith of the parties, the structure corrects the distortion. The independence of the materials review is a design for cutting in from the outside to correct a bias the organization cannot fix on its own. Where the review sits within the three lines of defense is taken up in The Board's View, Vol. 8.
04The Reviewer's Weapons Are Independence and Disconfirming Data
The reviewer's role is to break management's confirmation bias from the outside. Two things are needed for that: the independence to keep distance from the shared expectation, and the power to present inconvenient data.
Being able to lay down a single piece of disconfirming evidence — this claim is not supported by our own trial — opens a hole in the shared expectation. A reviewer who loses independence falls in line with the organization's expectations and stops working as a corrective device. So independence is not the reviewer's authority; it is the weapon itself. How to design the relationship between management and review so that independence is preserved is taken up again in Vol. 9.
- Confirmation bias produces the cherry-picking of materials. The source of the bias is not individual ill will but the expectation shared across the organization.
- Sunk cost delays restraint and withdrawal. The larger the investment, the more the risk that should be visible is clouded.
- Independent review is the institutional device for correcting bias. This is why the MSA Guidelines require independence and monitoring of the review.
- The reviewer's weapons are independence and the presentation of disconfirming data. Lose independence, and the review stops functioning as a corrective device.
- Companies Act, Article 362, Paragraph 4 (establishment of an internal control system). Provides that the board of directors shall decide on the structures to ensure the propriety of operations.
- Daiwa Bank shareholder derivative suit (Osaka District Court judgment, September 20, 2000). Established directors' duty to build internal control and monitoring systems, and the liability for damages arising from its absence.
- COSO. Enterprise Risk Management (COSO ERM). Sets out a framework for the organization to correct judgment bias through risk assessment and control activities.
- Ministry of Health, Labour and Welfare. Report on the Monitoring Project for Sales Information Provision Activities. A primary source that records actual deviation cases with company names anonymized.