Okay. I can't argue with someone who's simply being unreasonable. All science starts somewhere. All science builds on what came before it. If you want to argue that you can't start anywhere except the finish, there's nothing more to be said.
Feel free to be right in your own mind. Every major university and research organization differs with you, though:
https://oqi.wisc.edu/resourcelibrary/uploads/resources/Survey_Guide.pdf
http://archive.randi.org/site/index.php/swift-blog/1862-are-surveys-scientific.html
https://explorable.com/research-and-surveys
https://www.nbrii.com/our-process/validity/
Read your links, I believe they support Texas Yankee's position as much as your own in that valid, reliable surveys are extremely difficult to construct, apply and interpret. I say read the links, the link from "Explorable" was so poorly written as to bring it into question.
While I personally believe most surveys to be virtually useless due to the aforementioned complexity and expertise required as well as the unreliability of the survey targets (I often simply click through the center choice of surveys primarily because I don't like to be surveyed), it is possible, if appropriately constructed, to achieve somewhat of a "guess" as to the respondents position assuming no personal investment.
I simply don't believe a conversation as described to bear any relevance due to the bias and self-interest of the participants on such a topic. [Gun rights advocate versus 30.06/30.07 poster, potential customer versus business owner/associate, etc.] In this regard I question the reliability of even anecdotal "evidence", both sides have a vested interest, they are innately biased.
I perceive any instrument that purports to scientifically measure something as subjective as human opinion to be inherently inaccurate regardless of the integrity of it's construct.