You are not logged in.
If in the first town, there are exactly 19 eligible voters, all of whom voted, but in the third town, there are 100,000 eligible voters, so only 19% of them voted, then that 30% vote for A in the first town is perfectly true and reliable, while in the third town, the 3000 votes for A is a much *less* reliable indicator of the will of the people.
Percentages by themselves are meaningless. It is the context that gives them meaning. You haven't provided enough context here to know which of your three scenarios is more reliable.
As bobbym and Angishom have said, through, simply as measures of the relative amount of votes received, the percentages are equally reliable in all three cases.