depends on what you want to do. If it's an unimportant (high-school) science project then it should be A-okay to ask fellow students and people on the internet
you should also collect demographic data, like age, gender, etc.
you certainly have to know what you want to find out
for instance, "Are religious people more likely to have lucid dreams?", or "Do people under stress/with sleep deprivation experience abnormal sleep related phenomenons, such as lucid dreaming, OBE, et al."

don't worry too much about the surveys, you can also have a project without a lot of surveys and experiments, as long as you collect good data from other scientists.