Design a survey that isn't secretly leading people toward an answer
Question wording steers responses more than most survey writers realize.
- Use it for
- Anyone writing a survey who wants results they can actually trust.
Survey questions can lead respondents toward a particular answer through wording choices the writer doesn't even notice — a loaded premise ("how much do you love X"), an unbalanced scale, or an order effect where earlier questions prime later ones. Someone writing their own survey is the worst-positioned person to catch this, because the bias usually reflects what they already believe or hope to find.
When not to use this
For an informal, low-stakes poll where precision doesn't matter, this level of scrutiny is overkill. This is for a survey whose results you actually need to trust and act on.
Did this work?
A neutral third party couldn't guess from the question wording what answer you were hoping for.
Tested on claude-opus-5. Evidence status is draft; it moves to battle-tested only on recorded runs, never by hand.