Quote:
Originally Posted by
icfireball 
First, when I saw SDW's red responses, I actually thought that it was rather clever. The drawback of course is that it's hard to quote. But your post just proves you are a forum formatting whore.
Thank You Very Much! 
On to real substance, you referenced a so called . This is
absolutely erroneous. The Zogby poll was correct. Barack Obama WAS ahead in California by about 13 points on election day. But in California, "early voting" is allowed, and well over a million votes had been bagged before Feb 5... most of those breaking for Hillary because she is the establishment candidate. That's why she did so well in states like Florida and Michigan where no one campaigned. To corroborate my point, let me direct your attention to a SurveyUSA that was released the same day as the Zogby poll that showed Clinton ahead by 10 percentage points. Why this discrepancy? SurveyUSA decided to include people who had already voted before Feb 5 in the poll results.
Now onto the general election match-up polls. Yadda, yadda, yadda, ...
At this point in time all this means is GIGO.
Let's see what the polls say after both parties have nominated their candidates.
As to your Zogby comment, after doing
my own first hand statistical analyses, I can truthfully say;
You're full of (sh)it!
So to prove the above statement, I'll first post a few links;
CA SoS Democratic Primary Results
So basically, HRC received 51.92% of the
total votes cast, and Obama received 42.40% of the
total votes cast, or overall both HRC and BO received 94.32% of the
total votes cast, a difference of 9.52% in HRC favor.
So now let's go to the Zogby Poll;
California - 2008 Presidential PollsReuters/C-SPAN/Zogby Poll: McCain Dominates GOP Races; Dems Locked in Tight Contests As Super Tuesday Polls Open
See page 6 of the Zogby PDF file for their California Democratic Primary prediction of February 3-4, 2008.
BO is polled at 49% and HRC is polled at 36%, so immediately a red light starts to flash (in my mind anyway), as these two percentages add up to only 85% versus the 94.32% from actual vote totals, or Zogby underestimated HRC + BO by 9.32%, this under prediction is one of three critical steps to understanding the Zogby polling errors. Note that the sample size was n = 895, and MOE was ± 3.3% (a standard textbook calculation (although the actual number is 3.24%)) with the underlying assumption that the sample taken is truly random (which it clearly was not, but more on that later) which represents the 95% confidence interval, or to put it another way there is a 95% chance that the Zogby estimates correctly reflect the actual voting population within their stated MOE bounds.
The final link is the CNN exit poll data;
EXIT POLLS: California
Just looking at the male/female BO/HRC table at the top of that page (i. e. taking a weighted average of the exit polling data), BO received an exit poll estimate of 41.52% of the total vote and HRC received 52.56% of the total vote, or HRC + BO = 94.08% versus the aforementioned 94.32% from the actual votes. The sample size, n = 1908, MOE = 2.23%.
So the second critical step is to realize that
on February 5th, 2008 the exit poll data agrees quite closely to the actual total votes cast (early ballots
and primary day ballots combined). Thus, if correct, the exit polls suggest a voting pattern that changed very little between primary day ballots and early ballots cast.
I could rest my case there, but I'll take it one step further (the third critical step), I used the above links, the aforementioned Zogby polling percentages and the actual voting population data, and simulated an early voting population (x) versus the actual voting population (X). I used increments of 100,000 (from 0 up to the total voting population) and entered the data into an Excel spreadsheet. I suggest you repeat this exercise for yourself to see how absurd the Zogby polling numbers really are.
Anyway using the original Zogby polling percentages (BO = 49% and HRC = 36%), to match the Zogby numbers requires an early voting population of 2,646,440 votes cast, at which point HRC receives 61.63% of the early votes and BO receives 38.37% of the early votes, or a difference of 23.25% (Note: I used the Solver function in Excel to give me the closest rounded integer solution).
Now you say that 1M+ early voters were cast, yet in fact early ballots must = 2,646,440 (or greater) in order to get a valid (positive) solution using the Zogby percentiles.
And trust me, I tried all types of alternatives (deviations from the Zogby percentages), and any valid solution that gets me close to 1M+ pre-primary day voters gives me numbers that fall well outside the Zogby MOE.
Therefore, I've concluded that the Zogby polling data for the California Democratic Primary was full of (sh)it, since the Zogby numbers fail 3 critical statistical criteria. For the Zogby numbers to be randomly representative of the actual voting population, would require a larger sample size that has statistical significance in all demographics, clearly there are dozens, if not hundreds of demographic groupings, thus each demographic group within the Zogby poll must actually offer a fair (and random) metric of each percentile of each demographic that actually voted.
So, if you don't mind, please provide some factual links supporting your position mathematically and statistically, TYVM.