General Discussion

The reason that surveys (specifically, things that are said to be surveys because they don't qualify as polls) are regarded as less reliably predictive of the opinions of entire populations (which I assume is what you mean by "scientific" ) is because respondents are often self-selected, whereas polling inherently involves (among other tenets) specific efforts to choose respondents randomly, strongly increasing their predictive quality afa the whole population is concerned.

Once you, as the survey taker, have to stipulate that "these results reflect the opinions of the people who elected to participate in the survey" you are essentially admitting (to people in the know, at least) that your survey's predictive strength is weak (e.g. as you say, have limited value).

It appears from his response that Donnie Dipshit is not someone who is in the know, though, with a pathological liar, such things can be difficult to determine conclusively