I agree with you, surveys can be very wrong depending on how they are done and then what is done to extrapolate the data. I have a simple example. If you ask how many people voted for a green party politician in a neighborhood where mostly people live who support this party, and then extrapolate this you will get very bad results. I never really liked any way of estimating crypto users that I have come across over the years. There are many bad approaches from active wallets addresses to number of users of a particular platform or platforms and services. I would say that a very low percentage of people own or use crypto still.