If few enough people vote, arguably yes (though that's a fairly complex question and depends a lot on what your conception of democracy is). But in essence, an election is a full-population poll: like a random sample with probability 1 of being chosen. So it's statistically maximally valid. We (political scientists) tend to squint a bit and say the opportunity to vote is equivalent to actually voting, so non-voting doesn't harm the legitimacy of elections. I don't personally buy that, but it's a tricky problem to fix.
Because I, for one, didn't know it was happening and therefore could not participate. My chance of having my opinion count was zero. So pretty much by definition the poll is not representative of the universe of people-who-include-me. And by induction of all the other people who had no chance to participate.
There's a single democracy. There's no shortage of random polling websites. Telling me "You are ruled by this person, who was selected in a process you knew about since childhood" is one thing; telling me "This is likely true because people on this website you've never heard of say it's true" is another.
(And yes, governments where there are ad-hoc rationalizations for who gets to vote or where the procedures aren't publicly announced are in fact illegitmate and widely recognized as such.)
I claimed a thing was true. You said that the results of a debate disagree with me. Either you meant that my position was likely to be untrue as a result, and in turn the position advocated by the "winner" of the debate was likely true, or you didn't mean anything relevant at all. Which is it?
Isn't our democracy built upon "self selected polling" , people who chose to go excesrice their right to vote. Its all meaningless?