Learn

Evidence quality is a property of the claim

The same document can be strong evidence for one statement and almost worthless for the next. A supplier's audited accounts settle what it earned last year and say little about whether it will deliver on time this spring. Evidence quality lives in the match between a source and a claim, never in the source alone.

Direct and contextual evidence

Direct evidence speaks to the exact claim: a signed quote for the price you will pay, a delivery log for the lead time you care about. Contextual evidence describes the surroundings: an industry report on average prices, an article about congestion at the region's ports.

Context is useful for setting expectations and for spotting a figure that looks out of place. Trouble starts when it stands in for the direct kind, and an average across a whole market is quietly read as the quote from one particular firm.

Four questions for any source

First, does it bear on this claim or on a nearby one? Second, was the author in a position to know, through measurement or first hand access rather than repetition? Third, is it independent of the person who benefits from the claim? Fourth, could it have come out the other way?

That last question is the sharpest. A vendor's own case study was always going to be favorable, so it tells you little. A reference customer you found yourself, who was free to complain, tells you much more.

Common ways evidence goes wrong

Length is mistaken for weight, so a long report with one loose paragraph on the question carries the meeting. Age is ignored, so a figure from four years ago is treated as current. Repetition is mistaken for confirmation, when five articles quoting one press release are still one source.

Each of these errors has the same remedy: trace the claim back to where it was first measured, and judge that origin, not the echoes.

How the room grades a source

In a Decision Room, the Scout looks for public sources that bear on each material claim, and the Inspector decides what each source actually proves, marking the claim supported, partly supported, unverified or contradicted. A partly supported mark is common and honest: the source confirms a narrower statement than the one being made. The Verdict then rests only on what the marks allow, and the Human Decision is made with those marks in plain view.

One supplier, three sources

A buyer is told a new supplier delivers within 21 days. The supplier's brochure says so, which is a claim, not evidence of it. A trade magazine profile repeats the brochure, which adds nothing. A shipping record the buyer obtains shows eleven of the last fourteen orders arrived within 21 days, and three took up to 30. That record is direct, independent and capable of disappointing, so it decides the matter: the claim is partly supported, and the plan should allow for a month.

Questions

Is a famous source always stronger?
Reputation helps you guess whether a source is careful, but it cannot make a source speak to a question it never addressed. A well known publication writing about a whole industry is still context for a claim about one firm.
How old is too old?
That depends on how fast the thing measured changes. Prices and lead times can move in months, while a law or a company's founding date stays put. Ask whether the figure could plausibly have changed since it was recorded.
What should I do when good sources disagree?
Keep both on the record and look for the reason. Disagreements usually come from different dates, different definitions or different samples, and naming the reason often settles which source fits your claim.

Bring Argumont a Decision worth challenging.

Start with a paid 60 day pilot on your own real Decisions.

Research a claim