Reply #21: Consider any RV sample N == 100% of the electorate.

[tiptoe]

In the LVCM Scores from 1-to-7 are given for answers to a set of questions pertinent to characteristics of a "Likely Voter".

A distribution of scores within the sample N determines, for example, that a total score of X1 corresponds to 60% of the electorate "making the cut", a score X2 to 65%, X3-->70%...X4-->75%...X5==80% etc. etc

The Pollster then presumably makes a "projection" of the turnout for the November election. The basis for that projection seems rather arbitrary, a mere guess (or maybe a preference of perhaps a "target" for fraud-dependent pollster?). But, if he does as political analyst Bruce Gyory did in 2008, just days before the election, he might investigate EARLY-VOTER turnout. High % Turnout of the electorate favors Dems. Gyory, upon seeing that numbers from certain states were relatively very high, warned other analysts in an Oct 29 2008 Newsday article (which is linked to in Charnin's TRUE MATH post, but shortly thereafter became "unavailable", and still is, at Newsday's website) to "forget about the LV polls" (which exclude by "LV"-criteria a high proportion of Dems va Reps) and, instead, to heed the RV poll samples for projecting Obama's share accurately. Not many did, and 131 million turned out to vote, 89% of the Electorate! That's a lot.

How to analyze the accuracy?

THe US CENsus Bureau does a post election survey NOT asking WHO one voted for, merely "Did you vote?" (and how: "In Person"? "By Mail"?)

But, whereas the Natinal exit poll in 2004 polled 13,047 random-selected exiting voters and in 2008 more than 17,000, both reflective of a less than 1% Margin of Error, the Census Bureau interviews 70,000 people and determines with an accuracy of 0.3% MoE the size of the voting age population that actually voted (at a 90% confidence level).

You can tell how "accurate" the average pollster "projection" (i.e. "guestimate") was by comparing the average of the sizes of their LV samples relative to the RV samples from which they "qualified" as "Likely VOters".

TruthIsAll shows an example of determining the accuracy of an aggregation of 5 pollsters's LV subsamples and their corresponding RV full samples for 2004.

You can see them here, by scrolling down to the 2004 Pre-electionn Poll sections.

The Census Bureau figure of "Turnout" (88.5%, link provided) can be compared to the aggregated pollster's LV size as a % of the aggregated RV samples (See "Projected Turnout"). THE CEnsus BUreau also gives an absolute figure of "Votes Cast". The difference between the Census Votes Cast and the Recorded Vote COunt is, natually, the "Uncounted VOtes".

Anyway, have a look.

You'll see in Charnin's FOrecast Model that CNN/TIme provides 16 RV samples with LV subsamples. The % of the LV size compared to corresponding RV size == the current Projection IF THE ELECTION WERE HELD TODAY.

I believe Gallup's current LV to RV sample sizes is 1883/3000 == 62.8 % projected turnout...for Gallup...at this time.

A feature of the preelection polling scam as unfolding is the gradual DISAPPEARANCE of the RV Samples.

And, Rasmussen, the Republican affiliated pollster who's polls shows 8.1% margin advantage for GOP vs 1.8% against 124 others, DOESN'T PROVIDE RV SAMPLES AT ALL!!! We can't even say if his "LV" samples are drawn from a RV sample "representative" of the electorate.

But the GOP Election Fraud System must surely appreciate Rasmussen, for the margin of "Cover" he affords *pre-election* for their fraud manipulation on election day and the post-election ultimate cover-up of the "Forced" Final exit polls.

Read about the RV and LV polls in Charnin's 10-15 post.

tt



TruthIsAll actually did a beautiful analysis of the candidate shares among the "Unlikley to Vote"