Probing Questions

The defining feature is dependence: a probe reacts to the participant's previous answer (Robinson, 2023). Probes can be laddered, each following the answer to the one before, so you move from "a bit much" to a specific moment and then to a reason.

A probing question is open and uses the participant's own words. A closed question can be answered yes or no. A leading question contains the answer you expect.

• Opener: a prepared main question, such as "Tell me about the last time you ordered food online."
• Closed question: "Was the price too high?" It invites "yes" and supplies the cause.
• Leading question: it introduces an emotion, reason or brand the participant did not use. "Did it make you feel bad?" leads; "How did it make you feel?" does not.
• Clarifying probe: asks what a word means. It is one kind of probe, not a separate thing.
• Silence: a pause after a short answer. It helps, but it is a pause, not a probe. A quiet gap can send the participant to a new topic rather than deeper into the last one.

Participants answer the words they are given, and sometimes remember those words later as fact. Loftus and Palmer's 1974 experiment showed students films of car crashes and asked how fast the cars were going. Mean estimates were 40.8 mph when the verb was "smashed" and 31.8 mph when it was "contacted". A week later, 16 of 50 who had heard "smashed" said they saw broken glass, against 7 of 50 for "hit" and 6 of 50 with no speed question. There was none. The study used American students and films, so it shows the mechanism, not the size of the effect in a consumer interview.

Run four tests on each follow-up before it goes into a discussion guide, the list of questions a moderator plans to ask. It is a probe only if it passes all four.

1. Does it depend on the last answer? Pass: it refers to something the participant just said. Fail: you would ask it in the same words to everyone.
2. Is it open? Pass: it starts with what, how or when, or asks them to describe something. A "Can you..." request is a pass only if it is an invitation to describe, not a yes/no question. Fail: "yes" or "no" answers it.
3. Does it use only their words? Pass: every emotion, reason, brand and number in your question came from the participant. Fail: you introduced one.
4. Does it avoid offering a reason? Pass: it asks what led to the answer. Fail: it suggests the cause ("because of the price?").

The same fix applies to openers: Nielsen Norman Group's interview-guide advice replaces "Do you work in an office?" with "Describe a typical day at work."

Carry a short bank of neutral probes grouped by job. Ask descriptive and example probes early and "why" probes later, once rapport has built. The section "The DICE approach" in Robinson's paper also places descriptive and idiographic memory probes earlier than clarifying and explanatory ones. The five jobs below each map to one of Robinson's four types, shown in brackets.

• Elaborate (descriptive): "Tell me more about that."
• Specify (idiographic memory): "Give me an example of when that happened."
• Clarify (clarifying): "What do you mean by 'a bit much'?" Or repeat their phrase with a rising tone.
• Explain (explanatory): "What made you say that?"
• Recall (descriptive): "What was going on the last time that happened?"

Probes are not usually written in advance, so keep the bank on an index card. Two or three prepared probes under each main question, and a pause of a few seconds after a short answer, are enough.

When a study needs a moderator, Qualfacto can supply one, and sessions can run in languages other than English. It recruits participants by city for in-depth interviews, focus groups, digital diary studies, online bulletin boards, shop-alongs, ethnography, usability and app testing, in-home use tests and opt-in digital behavioural tracking. Deliverables such as recordings and transcripts depend on the scope agreed before the study begins.

Whoever moderates, ask for four things you can check:

1. A sample transcript, or a pilot transcript where the scope includes recordings and transcripts, so you can read the real follow-ups against the four tests.
2. The moderator's probe bank for your topic, to run through test 3.
3. A short briefing before fieldwork, in which you walk the moderator through the four tests.
4. The share of follow-ups that pass test 3, counted from the transcript after fieldwork: follow-ups using only the participant's words, divided by all follow-ups. You can recompute it yourself.

Pricing is by proposal for each project. Qualfacto takes briefs by email: write to support@qualfacto.io and ask about the free pilot.