Is there a margin of error with a vote like this? Isn't it a direct count of people who want their opinion known? It isn't sampling and trying to figure out what everyone wants.
While a margin of error is required for polls due to inherent limitations of accuracy and scale if your democratic institutions have a margin of error you're not doing democracy right.
That's assuming that people correctly know and express the opinion that they hold. Asking people directly is still somewhat of a noise signal. Also, the act of viewing the result of the vote seems to have changed people expressed opinions at least.
I don't know about 'error' but there's definitely room for valid decision changes. You can take most things back within a grace period after signing the dotted line, after all.
But that isn't what this measured. Like how the US census is suppose to be an exact count of everyone on April 1; it may change due to births and deaths, but on that day (at some time) it is suppose to be exact.
A vote is an exact count of the will of the people at a given time. There is no error. People can change their minds, but that isn't what the vote represent. That is, a vote is what people who chose to vote said they wanted, not a probabilist representation of what the people as a whole want.
Given that people apparently keep changing their minds right up to voting day (indicated by polls), I think there is actually a lot of noise in the result.