Today's prediction....Here is the one from last week:
Posted on 09/03/2016 9:42:53 AM PDT by SoFloFreeper
Winning the White House depends as much on who comes out to vote as which candidate Americans prefer. Each week, we poll more than 15,000 people, then factor in likely turnout among key demographics to see how voting would play out in the Electoral College.
(Excerpt) Read more at reuters.com ...
Today's prediction....Here is the one from last week:
an interesting interactive, as this thing allows you to change the “turnout” prediction for different demographics.
All of this is irrelevant if more American taxpayers and gun owners show up to vote than do freeloaders, foreigners, Muslim terrorists and illegal alien invaders.
Hillary’s EC edge keeps shrinking.
Even Reuters/IPSOS knows its only a matter of time before Hillary slips below 270.
Inevitable Hillary, huh? Dream on.
Not difficult to see who Al Reuters is pulling for.
The samples for the smaller states are very dated.
Consider Florida, a large state, the current sample is 8/26-9/1 and the prior 8/19-8/25. So, those are weekly samples. I figure they’re about 900 (spreading 15,000 across the states by population).
Trump halved the gap, from 7 to 4 points.
Now consider Michigan. Michigan is a middle-sized state. The current sample is 8/19-9/1, and overlaps with the prior sample of 8/12-8/25. I figure there’s 400 to 500 in any one week’s sample, and to juice sample size, they use two weeks of surveying. This may be reasonable, but it has the result that the Michigan figure is a bit stale.
BTW they have Trump leading by a point in Michigan.
Now consider New Hampshire. N.H. is a small state. The current sample is 8/12-9/1, and considerably overlaps the prior sample of 8/5-8/25. I figure there are less than one hundred surveys in any one week. Even aggregating over three weeks results in a small survey.
BTW they have Trump leading by a point in New Hampshire.
BOTTOM LINE: Abstracting from the bias against Trump in the Ipsos poll (which Nate Silvers puts at 2 points), it only gives us useful information for thelarge , non-battleground states; e.g., California, New York and Texas.
The 7-Eleven straw poll may be more useful (via extrapolation) for the small to medium-sized non-battleground states. Seriously.
i have no reason to disbelieve you, but where do you get those numbers as sample sizes? I’ve heard they are small but cannot find them.
sff
15,000 per week, which I assume by state is approximately in proportion.
In the network exit poll, they have a sample size of about 14,000, but focus it on the battleground state and the state with competitive Senate and Governor races. They then jump through some hoops to derive inferences for the national popular vote (e.g., how men voted relative to women).
* What if only people that turned out for rallies voted? Trump at over 100 percent.
* What if only people that hide behind masks and attack people for political purposes voted: Close match between Hillary and Stalin as a write in.
re: if more American taxpayers and gun owners show up
What is this concept of “show up”? The election will probably be won by the side where the committed not only “show up” but “push up” their non-political neighbors to vote their way. The reality is that most potential voters are not political junkies like FReepers or DUmmies.
Most voters are busy with family, work, church, struggling to pay their bills. They watch the ballgame, sitcoms and reality TV. They only vote when someone urges them to vote.
So the real question is whether their political neighbor will knock on their door?
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.
