Avoid leading questions

Overview

A leading question nudges a participant toward a particular answer, without them necessarily noticing it's happening. This produces data that reflects how the question was worded, not what the participant actually thinks. Avoiding leading questions is one of the most important things you can do to get honest, usable results.

What a leading question looks like

A leading question often contains a built-in assumption or a loaded word that pushes the participant in one direction. For example:

"How much did you enjoy our new, improved checkout experience?"

This assumes the participant enjoyed it, and assumes it's an improvement, before they've said either. A neutral version simply asks:

"How would you describe your recent checkout experience?"

Watch for these common patterns

Assuming a positive experience. Words like "improved," "convenient," or "easy" baked into a question tell the participant what answer you're expecting. Let the participant use their own words to describe something, rather than agreeing or disagreeing with your framing of it.

Loaded or emotionally charged wording. "Don't you think this feature is a waste of time?" pushes much harder than "What do you think of this feature?" Neutral wording lets the participant land wherever they actually land, instead of reacting to the emotional charge of the question itself.

Unbalanced scales. If a satisfaction scale runs from "Good" to "Excellent," there's no room for a genuinely negative response, so people who are unhappy are forced into answering more positively than they actually feel. Make sure scales are balanced, with an equal number of positive and negative options around a neutral midpoint.

Double-barrelled questions. A question that actually asks two things at once, "How easy and enjoyable was the process?", forces someone to answer a single question for two different judgments, which may not even agree with each other. Split these into two separate questions.

Order effects. A neutral question can still lead a participant if it comes right after a question that primed them to think a certain way. If question order might be shaping responses, consider whether Group & Randomization could reduce that effect, see Sequential vs monadic design for how randomizing order works in Decode.

A simple check before you publish

Read each question and ask: could someone tell what answer I'm hoping for, just from how this is worded? If yes, rewrite it in more neutral language before your study goes live.

What to do next

See Stimulus preparation checklist if your study includes an image, video, or other stimulus alongside your questions.