You miss my point, Johnny. I am not saying the survey is flawed. A critical part of any survey is its margin of error, and in the case of this survey for the 16-24 year olds the margin of error in something like 14%. That is, whatever the stated figure it could be higher or lower by roughly 14%. Because of that range, it would be unwise to make any deductions from the headline value. I would need more information to calculate it precisely, but you can get a rough idea of the margin of error by noting that with 14 people each individual vote represents approx 7% of the final result. So changing a yes voter to a no or vice verse gives something like a 14% swing.

That is a completely different issue from whether a 16-year-olds vote is more or less significant than a 70 year olds.