Would you buy a brand because a friend likes it?

Would you buy a brand because you saw a commercial for it on TV?

Would you buy a brand because you found out they endorsed an important event?

Just about every survey about a product or service (as opposed to a person or idea) includes a question about purchase. In some ways, purchase questions are pretty easy to answer. The soap is in your cupboard because you purchased it. The cheese is in your fridge because you purchased it. Sometimes, we even know why we have purchased something (though in many many cases we have absolutely no clue).

But when you switch the question style away from past purchase to future purchase, you introduce massive error. Why? Because you’ve just asked your survey responder to guess about something they haven’t yet done and might never have thought of doing until you brought it up. You’ve just introduced an irreparable confound.

So please. Stop asking people to hypothesize about what they will do and ask them about something where they have a reasonable chance of knowing at least some of the answer. Ask if they like the product, ask if they like their friend who recommended it, ask if they like the event, ask if they like the concept.

Just don’t ask them to attribute a reason to it.