Reply #10: How about: [View All]

[Condor]

"First off, the football game comparison is wrong, we are talking about percentages in this case, percentages can definitely stay the same throughout the day, even though people are still voting. If intermediate results are off, you still should be able to account for the fact that according to your poll the majority of democrats voted in the morning, while an even larger group of republicans voted in afternoon and early evening."

Examples below:
This being a percentage increase, this really is something that needs to be accounted for.

Suppose in the morning 53 percent voted Kerry, and 47 percent voted Bush. In the afternoon this is supposed to be brought to a 51-48 for Bush, so lets examine some numbers:
Suppose 100 voters in the morning, 53 for Kerry, 47 for Bush.
If the turnout remains constant throughout the day, expect 200 voters at the end of the day. For 200 voters, 51-48 to Bush would mean 102 Bush voters and 96 Kerry voters, or an afternoon in which 55 Bush votes and 43 Kerry votes were cast.

That is quite a swing, from 53-47 to 48-51 may seem like a swing of 5 points, but in the final exit poll, all results from the midday exit poll will be included, which means the afternoon has to go a lot more to Bush then to Kerry, in the case of equal votes cast before and after the midday poll. The midday swing would then be a 9 point swing, something to be explained.

Suppose turnout was larger in the second part of the exit poll (quite probable since this covers a larger time period, and all people returning from work), lets assume 200 afternoon votes were cast in our example.
The required vote totals would be 153 Bush, 144 Kerry, or 106 votes for Bush and 91 for Kerry. This still means in the afternoon and early evening, people voted 53-46 for Bush in the afternoon, still a reverse of the morning voting trend, still something to explain.



There is nothing erroneous about midday exit polling numbers. It shows what the standing is up until that moment. If the exit polling strategy is correct, then there should be some explanation offered as to why the morning voting trend and the afternoon voting trend were so different. Saying the midday exit poll is erroneous means the exit poll itself is erroneous.

And you might add that it is not actually that hard to get a percentage from raw numbers, but maybe that will lower your chances of getting a reply. ;)