[Yashcheritsiy]

I dunno. I’m pretty sure that n=575 should yield around 4-4.5% at a 95% confidence level. Doesn’t even depend too much on what the total population size is.

[longtermmemmory]

I do not like this averaging different polling methodologies and different categories of subjects.

(ie adults vs registered voters vs registered republicans vs previous voters vs likely voters vs evangelical vs non-evangelical vs so on and so on)

We forget the media polls are for the public, all these candidates have PRIVATE INTERNAL poling which tells the truth and that is never public.

(ie McCain, Romney, GHWBush all knew they were going to lose long before election day)

[xzins]

An oversample of registered voters who are Republican was interviewed for a total of 512 interviews with Republican registered voters. The results were then weighted in proportion to the adult population. The margin of error for the Republican registered voters is 7 percentage points.

The above is an odd paragraph. I plugged the 575 into my calculator, and it's about 4.18 at 95%.

So, what is the above paragraph saying that got them to 7%? That their weightings got their sample size down to 169 people?

That is so odd.

[xzins]

An oversample of registered voters who are Republican was interviewed for a total of 512 interviews with Republican registered voters. The results were then weighted in proportion to the adult population. The margin of error for the Republican registered voters is 7 percentage points.

The above is an odd paragraph. I plugged the 575 into my calculator, and it's about 4.18 at 95%.

So, what is the above paragraph saying that got them to 7%? That their weightings got their sample size down to 169 people?

That is so odd.