Research

Verified against ChatGPT · 2026-08-10

Turn a vague research question into one a study could actually prove wrong

Rewrites a broad, unanswerable research question into a specific, falsifiable version with a stated measurement and a defined population, so the resulting research effort has a real finish line.

ChatGPT (GPT-5.1)4 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

Here is a research question I want to investigate. Help me tighten it into something falsifiable — a version where a specific finding could actually prove it wrong, not just something interesting to explore indefinitely.

ORIGINAL QUESTION
Does gamification improve employee training outcomes?

WHY I CARE ABOUT THIS
We're deciding whether to spend budget adding badges and leaderboards to our compliance training platform.

WHO OR WHAT THIS APPLIES TO
Adult employees completing mandatory annual compliance training, not students or gamers in general.

WHAT I ALREADY SUSPECT
I suspect it might boost completion rates but not actual retention of the material.

Diagnose what makes the original question unfalsifiable or too broad as stated — usually it's an undefined population ("do people" — which people?), an undefined outcome measure ("does it help" — help by what measure, over what timeframe?), or a comparison with no baseline ("is X better" — better than what?). Name the specific gap in Does gamification improve employee training outcomes? before rewriting it. Then produce a tightened version that specifies the population from Adult employees completing mandatory annual compliance training, not students or gamers in general., names a concrete, measurable outcome, and states an explicit comparison or baseline. The tightened question should be falsifiable in the sense that you could describe, right now, a specific result that would count as a "no."

WHAT NOT TO DO
Do not just rephrase the original question in fancier language while leaving the same ambiguity intact. Do not narrow the question so far that it stops addressing We're deciding whether to spend budget adding badges and leaderboards to our compliance training platform. at all — the tightened version has to still answer the real thing you care about.

OUTPUT FORMAT
1. The specific ambiguity or unfalsifiable element in the original question.
2. The tightened, falsifiable research question.
3. One sentence describing what a "no" result would look like.
4. If the tightening narrowed the scope meaningfully from We're deciding whether to spend budget adding badges and leaderboards to our compliance training platform., one line flagging that trade-off explicitly.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

A question like "does gamification improve training outcomes" has no single answer because "outcomes" and "improve" are undefined variables the model has to silently pick a value for in order to respond at all — left unspecified, it will typically answer about the most commonly-discussed version of the topic in its training data, which may not be the version relevant to the actual decision at hand. Asking the model to first diagnose the specific ambiguity, rather than jumping straight to a rewrite, forces it to name what's actually missing (population, outcome measure, or baseline) instead of pattern-matching to a generically "better-sounding" version of the same underspecified question. Requiring the tightened question to support a describable "no" result is the practical test for falsifiability: a question phrased so broadly that literally any finding could be spun as a confirming answer isn't actually researchable, it's just a topic, and stating the shape of a negative result upfront is the cheapest way to check whether the rewrite actually fixed that. The trade-off flag at the end matters because tightening a question for falsifiability and preserving the original motivation are in some tension — a maximally falsifiable question is often narrower than the thing someone actually cares about — and surfacing that trade-off explicitly means the person asking gets to decide whether the narrower, answerable version still serves their real purpose, rather than discovering the gap only after the research is done.

What you get back

Ambiguity: "improve training outcomes" doesn't specify whether that means completion rate, quiz score, or long-term retention. Tightened question: Among adult employees completing mandatory annual compliance training, does adding badges and a leaderboard increase 30-day post-training quiz retention scores compared to the same training without gamification? A "no" result: retention scores show no statistically meaningful difference between the gamified and non-gamified groups.

Verified against

ChatGPT GPT-5.1 · 2026-08-10

Changelog

  • 2026-08-10 Initial publish, verified against ChatGPT GPT-5.1.

Need this built into your business?

If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.

EXPLORE WHAT SCULT BUILDS
All Research prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY