Complete the prompt below and paste the text into Claude or Google NotebookLM.

<aside>

Goal:

Assess and improve the hypothesis below so that it fully meets the requirements of a researchable hypothesis for investigative journalism.

Instruction:

  1. Assess the hypothesis against the nine criteria below.
  2. For each criterion, give a brief rating: meets / partly / does not meet, with a short explanation.
  3. Rewrite the hypothesis into one or more improved versions that satisfy all the criteria.
  4. Use neutral and factual language:
  1. Deliver the output in two parts:

Assessment criteria:

  1. Actor first: Does the hypothesis begin with the person or organisation that has the greatest influence on the wrongdoing or problem?
  2. Correct structure: Does the hypothesis follow the form: [Actor] did/does [specific action] after [factual event], causing [concrete effect]?
  3. Brevity: Is the hypothesis a maximum of two sentences long?
  4. Action concrete and observable: Is the verb testable and does it refer to an actual action or omission (not to an opinion or intention)?
  5. Relevant action (unlawful, undesirable or hazardous): Does the verb refer to behaviour that may conflict with law, policy, public duty, safety or justice?
  6. Event factual: Is it clear which concrete incident preceded it (time, place, context)?
  7. Effect concrete and testable: Does the hypothesis describe a visible or measurable effect in the environment (policy, decision, harm, benefit, change)?
  8. Language neutral and factual: Does the text avoid subjective words (such as benefit, refuse, guilty) and negatives (not, no)?
  9. Comparisons factual: Are comparative terms (more, greater, more often) used only with an explicit comparison or reference point?

Output:

Part 1 – Assessment

No | Criterion | Rating | Explanation

Part 2 – Improved hypothesis/hypotheses

  1. [New version 1]
  2. [New version 2]

Hypothesis:

[add here]

</aside>

⚠️ AI’s answer says nothing about the likelihood that your hypothesis is correct. AI may suggest things you know are untrue.

🕵🏻‍♀️ Assess for yourself, based on your knowledge and experience, which AI suggestions you want to adopt.