Skip to content

The Hypothesis Ladder

Why It Matters

Investigations fail by jumping to specific accusations before establishing general problems, or by staying too general when evidence supports specific claims. The ladder provides structure: you need different evidence to show "this company has financial irregularities" (show patterns in public filings) versus "this company is committing fraud" (show intent and specific violations) versus "CEO knew about fraud" (show their knowledge and involvement). Attempting to prove the highest claim without establishing lower rungs wastes effort and risks libel.

The Key Move

  1. Start with the observable: can you show something is unusual or concerning without claiming wrongdoing? (Abnormal patterns, expert skepticism, complaints.)
  2. Escalate to problems: can you show the concerning pattern causes harm or violates norms? (Document specific harms, cite standards violated.)
  3. Escalate to responsibility: can you connect problems to specific actors and their decisions? (Show decision chain, who knew what when.)
  4. Escalate to intent: can you show it was deliberate not accidental? (Show pattern across instances, evidence of concealment.) Only advance when you have sufficient evidence for that level. Publish at the highest level you can prove.

Classic Application

Opioid crisis reporting. Early investigations showed:

  1. Observable: opioid prescriptions increased dramatically (CDC data).
  2. Problem: increased prescriptions correlated with overdose deaths (public health data).
  3. Responsibility: pharmaceutical companies marketed opioids as non-addictive despite internal research showing addiction risk (internal documents from lawsuits).
  4. Intent: companies continued aggressive marketing after becoming aware of addiction epidemic (emails, meeting notes showing executives aware of harm but prioritizing sales). Each level required different evidence and different publication decisions.

Surprising Application

Performance review documentation. When documenting employee performance issues:

  1. Observable: specific behaviors and outcomes (missed deadlines, error rates).
  2. Problem: impact on team and projects (delayed launches, customer complaints).
  3. Responsibility: connect to decisions and actions under employee's control (not external factors).
  4. Intent: patterns showing negligence vs. good faith effort (repeated issues after coaching vs. isolated incidents). Only escalate to termination when higher rungs are proven. Publishing (documenting formally) too early or too late both create problems.

Failure Modes

Stopping too early (you have evidence for higher rungs but publish only the safe observable claim). Jumping rungs (claiming fraud when you've only shown irregularities). Treating all rungs as equally difficult (intent is much harder to prove than observables - adjust effort accordingly). Not publishing at lower rungs when that's what evidence supports (perfect is enemy of good). Confusing rung level with importance (sometimes just showing the observable is highly consequential).

Go Deeper

Meyer, "Precision Journalism" on evidence standards for claims. Protess et al., "The Journalism of Outrage: Investigative Reporting and Agenda Building in America" on escalation strategies. Ettema and Glasser, "Custodians of Conscience" Chapter 6 on proof standards.