This is a real problem in survey design. There are various ways of trying to extract this information, but all very high uncertainty. For example, one way is to ask how maybe people do you know that have done various innocuous thing you have good population level data for. Then ask how many people do you know that have done x, undesirable to admit to thing.
I think it would be hard to get anything useful from such a survey though. In particular because it's not clear to me it would answer questions like:
- How many people who are medically unfit from diving are deterred by the form?
- How many people who are medically unfit from diving are not deterred by the form, but cleared by someone with limited knowledge of diving?
- How many people who are medically fit from diving are deterred by the form?
- How many people who are medically fit for diving are are unable to be cleared, because they seek a signature from someone with limited knowledge of diving? Someone who is familiar with diving but doesn't want the liability?
- Is the form designed to protect participants in diving or CYA
...