Bias check I — assumptions and informal fallacies

The sixth line

R-psp-3 ends with biases and assumptions checked in writing. "In writing" is the operative phrase: a bias check that lives in the analyst's head is not a bias check, and a marker cannot credit it. Three lessons cover the check — this one takes unstated assumptions and the informal fallacies in Van Cleave's chapter 4; the next two take the fallacies of relevance and the decision biases.

Step one: write the assumptions down

Every argument in a case rests on premises the writer did not state because they seemed obvious. Van Cleave's method for finding them: put the argument in standard form and ask what must also be true for the conclusion to follow. Lakeshore's cost comparison — "the lease is cheapest, so lease" — rests on at least three unstated premises: volumes will stay flat (if large-format volume falls by a third, the lease payment is the same and the margin is not); the lease includes maintenance as quoted; the two accounts leave if delivery stays at 84 %. Written down, each becomes something to test or to flag. Unwritten, each is a bias-check failure — none stated reads as none examined.

Five informal fallacies, in our setting

An informal fallacy is an argument whose form is fine but whose content does not support its conclusion. Five appear constantly in case reasoning:

1. False dichotomy — two options presented as the only two. "Either we buy a new press or we lose the accounts." Fix: list the alternatives; leasing and outsourcing also keep the accounts.
2. Begging the question — the conclusion is assumed in a premise. "The new press is reliable because reliable presses are what we need." Fix: evidence for reliability — the vendor's service record, references from other printers.
3. Composition / division — what is true of the parts is assumed true of the whole, or the reverse. "Each of our jobs is profitable, so outsourcing all of them will be profitable too" (composition — the margin falls from 42 % to 18 % when they move together); "Lakeshore is profitable, so the large-format line must be" (division). Fix: check the whole and the part separately.
4. Slippery slope — a chain of consequences asserted without evidence for the links. "If we outsource one line we'll end up outsourcing everything." Fix: show the links or drop the chain; outsourcing large-format says nothing about the rest.
5. Equivocation — one word used in two senses. "The lease is cheaper" — cheaper per year ($58,000against$67,400), or cheaper over five years ($290,000against$337,000 plus a press you then own)? Fix: fix the sense, then re-state the claim with the number.

Audit the clinic's reasoning

"Our patients won't book online, so the software is pointless" — a hidden premise (a system nobody books through has no value) ignores the reminders, which reach every patient; and won't book online is an opinion (L5.5). "If we go digital we'll be hacked and lose every record" — slippery slope: the links from digital to hacked to every record lost are asserted, not shown. "The paper book has never failed us" — begs the question against the two double-bookings last month.

Name five cold

Cover the list. False dichotomy; begging the question; composition and division; slippery slope; equivocation. Then the first step of every bias check: write the assumptions down.