Poll Watch: Rasmussen (R) 2016 Democratic Nomination Survey

[Stuckinthebush]

18. Why would you say it is not "scientific"?

I see DUers say this a lot and that's inaccurate.

A survey sample of 536 has a margin of error of around 4.2
A survey sample of 1000 has a margin of error of around 3.1
A survey sample of around 1500 has a margin of error of around 2.5

So, with a sample of 536 if you took 100 separate samples then the overall percentage of people responding the same way would be within 4.2% of the reported results in 95 of these.

The margin of error is a little bigger for a sample of 536 than 1500 but it has nothing to do with being scientific. It has to do with probability. HOW they sample (methodology) would speak to the issue of how "scientific" the poll is. If the methodology is sound then a sample size of 536 can result in illustrative results.