I'm on the reverse end of the spectrum. When researchers came in to ask questions about violence and weapons in schools, I truthfully tried to help them as best I could. It was only later that I understood that, when they asked if I'd seen a student fire a weapon at school in the past month, they hadn't meant for me to include the latin club's trebuchet.

I helpfully told them that I used theobromine weekly (in the form of chocolate).

Even if you have honest answers, it's a minefield to phrase questions so that you get the data that you think you are getting.

For example, there's a statistically significant amount of men that have sex with other men, but who self-identify as heterosexual. That group might end up in either camp if the survey isn't aware of the issue and phrases the questions badly, which makes quantitative studies much harder to do.