I am questioning the rationale for scaling down the "good chance" numbers using the name ID numbers. It is my understanding that the "chance would vote for" numbers are already BASED ON THOSE WHO HAVE HEARD of the candidate. Now, how about the proportion of those voters who have NOT heard of the candidate? For instance, the 27% of the voters who have not heard of Obama. Once they hear more about him, would it be more likely for them to think that there's a chance they would vote for him? Or less? The prudent view, based on statistical theory, is that they will not act any differently from those who have already heard of him when the poll was conducted. Therefore, of the 27% who don't know of him yet, 41% of them will probably say "good chance", conforming to the current statistics.
THEREFORE, the "chance would vote for" numbers should be expected to remain the same, barring any change in electorate or other factors. The scaling down of the Clinton numbers is statistically INCORRECT, and it seems the authors of this blog are intent on BRINGING DOWN her numbers and PUMPING UP everyone else, no matter the accuracy of the math. People, don't read beyond what the numbers say. Clinton has an overwhelming amount of support across the democratic electorate, and this has to be acknowledged. Fudging the numbers does not erase this fact.