statoryBETA
Pricing
Back to home

How to phrase better questions?

Your question can be short. statory works better when you also mention what you want to know, which variables matter, or which theory you have in mind.

AI-Based Analysis (AI Orchestra)

Question → first LLM design → four-axis collection/support → refinement → final LLM design → analysis

You can start without data. statory looks for official statistics, news, academic sources, and video materials, then chooses methods that fit the goal.

Good question examples

  • "Why is Korea’s fertility rate so low?"
  • "How does youth unemployment affect marriage rates?"
  • "Analyze the relationship between education level and happiness"

Questions to avoid

  • "Analyze it" — no variables or goal.
  • "Anything good" — unclear what to compare or explain.

Recommended pattern: abstract question + one to three variables + goal such as comparison, relationship, or prediction.

Theory-Based Analysis

Choose a theory → theory-based first design → four-axis collection/support → refinement → LLM theory revision → analysis

Use this mode for academic research or hypothesis testing. statory starts from 17 seeded theories, drafts variables and pathways, then adjusts the design to the data.

Good question examples

  • "Analyze young adults’ marriage intention using the Theory of Planned Behavior"
  • "Explain fertility-rate differences using social capital theory"
  • "Analyze depression differences in Korea using Big Five"

Recommended pattern

  • State the academic purpose.
  • Add the hypothesis or theoretical frame you want to test.

If you know the theory, name it first. If not, describe the goal and key variables.

Data-Based AI

Upload your data → first AI design → four-axis collection support → refinement → final AI design → analysis

AI expands your own data with official statistics, news, academic sources, and video materials when those sources can strengthen the analysis. This matches §22.10 entry point 4: AI Orchestra branch B with partial user data.

Good question examples

  • "Add news and statistics from the same period to this survey data"
  • "Attach industry averages to my sales data and compare them"
  • "Add relevant academic sources to interpret this depression-scale data"

Questions to avoid

  • "Look at this file and the web too" — unclear which external sources are needed.
  • "Do everything automatically" — no comparison target or analysis goal.

Recommended pattern: use this when your own data is not enough. AI can fill variable gaps or time-series gaps with supporting sources.