Skip to content

Credibility Assessment

Why It Matters

In adversarial contexts, you can't assume information is reliable just because someone asserts it. Witnesses have biases, memories fail, experts are hired to support specific positions, documents can be selectively produced. Yet you must make decisions based on this imperfect information. Credibility assessment is the discipline of extracting signal from noise when sources are potentially compromised. It's not cynicism (assuming everyone lies) but calibrated skepticism (adjusting confidence based on source reliability).

The Key Move

For any information source, assess four dimensions:

  1. Capacity - did they have the ability to observe/know this information accurately?
  2. Bias - what incentives do they have to distort?
  3. Consistency - does this account align with their other statements and independent evidence?
  4. Corroboration - is this confirmed by independent sources with different biases? High credibility requires high marks on most dimensions. When credibility is mixed, weight the information accordingly rather than accepting or rejecting wholesale.

Classic Application

Witness impeachment in cross-examination. Witness testifies Defendant ran red light. Cross-examination establishes:

  1. Capacity - witness was 200 feet away, at night, with obstructed view;
  2. Bias - witness is plaintiff's brother-in-law;
  3. Consistency - witness gave different account to police at scene;
  4. Corroboration - no other witnesses or physical evidence supports this claim. Each impeachment dimension reduces credibility incrementally. All four together devastate the testimony without ever calling the witness a liar.

Surprising Application

Evaluating startup pitches. Founder claims product will reach $100M revenue in three years. Capacity assessment: has this founder built a comparably-scaled business before? (Usually no - reduce confidence in projection methodology). Bias assessment: founder has obvious incentive to inflate projections for funding. Consistency assessment: do the unit economics in the pitch deck actually support the revenue claim, or are they contradictory? Corroboration: do independent market analyses suggest this market can support this growth rate? None of this proves the founder is wrong, but it calibrates how much weight to give the revenue projection versus other available information.

Failure Modes

Confusing credibility with likability - charming people can be unreliable; unlikable people can be truthful. Binary thinking - treating sources as completely reliable or completely unreliable rather than adjusting confidence levels. Motivated credibility assessment - believing sources that support your preferred conclusion and disbelieving those that don't. Over-weighting single dimensions - perfect consistency doesn't overcome terrible capacity to observe. Ignoring your own credibility - if your assessment is based on biased evaluation, it inherits those biases.

Go Deeper

Wellborn, Olin Guy. "Demeanor." Cornell Law Review, Vol. 76, 1991; Fisher, George. "The Jury's Rise as Lie Detector." Yale Law Journal, Vol. 107, 1997.