People put way too much stock in polls..
Well what do you think polls are? Random numbers?
Polling is just a small random sample of a larger group of people. Sometimes the smaller sample match the larger group and sometimes it doesn't.
No, like I said, I'm not very smart when it comes to this stuff, but to me it just seems unreliable to put a lot of stock in polls.
It's true that sometimes the smaller sample matches the larger one and sometimes it doesn't. But statistics gives us a way of quantifying how often, and how closely, the sample is likely to match the larger population (assuming the
population sample is truly random). For example, suppose a poll says that A is favored over B 53%-47%, with a margin of error of 3% (with a 95% confidence interval, which is standard.) This means that 95% of the time, the larger population is within plus-or-minus the margin of error of the values measured in the samples: that is, A is between 50% and 56% (53% +/1 3%) and B is between 44% and 50%. The other 5% of the time, the true values are outside that range.
This means that even with carefully constructed polls, 1 out of 20 is likely to be a true outlier. Some people latch onto this as indicating that polls are inaccurate. This is true -- but we have a good idea of just how accurate or inaccurate they really are! The counterpoint is that 19 of the 20 (carefully constructed) polls are within the MoE. And when you have multiple polls that indicate close to the same result, the confidence goes way up. If you have a single poll that shows A up 53-47, it could be an outlier, and even if it's not the true result could be anywhere within the MoE. But if you have a dozen polls that all cluster near the same result, the confidence that they truly represent the larger population is MUCH higher.
Edit: changed "population" to "sample" (...is truly random)
