Worked examples: from vague question to testable hypothesis

1) Laboratory research — enzymatic yield

Vague question: "Will changing the buffer help yield?"

Refined hypothesis: "If Buffer B (pH 7.4) replaces Buffer A during purification, then the mean yield (mg product / g input) after 24 h will increase by ≥20% compared with Buffer A, measured across n=6 independent runs. Success = mean increase ≥20% and a two-sided t-test p < 0.05. Assumptions: starting material quality and temperature controlled."

2) Product experiment — SaaS feature

Vague question: "Will users like the new onboarding?"

Refined hypothesis: "If new onboarding flow A is shown to new users, then 14-day retention (percent active users on day 14) will increase by at least 5 percentage points compared with current onboarding B, measured via randomized assignment of new signups over 30 days. Success = difference ≥5 points with 95% CI excluding zero. Assumptions: no concurrent product changes affecting retention."

3) Program evaluation — community outreach

Vague question: "Does workshop X improve outcomes?"

Refined hypothesis: "Participants who complete the 6-week workshop X will report a 10-point higher mean score on the employment-readiness assessment at program end compared with matched controls, controlling for baseline scores. Success = adjusted mean difference ≥10 points; assumptions: control group comparability and consistent assessment administration."

What to notice

• Each example replaces vague verbs with measurable metrics and an explicit decision rule.
• They list assumptions and scope so interpretations are bounded and debatable rather than assumed.
• Different hypothesis types use the same underlying structure: claim → prediction → measurement → decision.

Use these patterns to translate your own questions into testable statements. When in doubt, add one more line describing how you will measure the main outcome.