In late April, Insider Advantage shocked us with their prediction that Clinton was slightly ahead in their poll 44% to 42% in North Carolina. Because if they were right, Clinton may be able to sweep the primaries today! But their latest poll results have Obama ahead 48% to 44%. That suggests Obama has made a 6% gain over Clinton in the past week but it is all a fallacy.

In their 4/29 poll, the African American turnout that they predicted was 25% but on 5/5 they increased that to 35%. Perhaps they undersampled the AA turnout in their 4/29 poll. It would be slanderous to assume that they are doing this on purpose. Perhaps they learned of something that indicated African Americans were going to turn out in record numbers.

To compare, Survey USA’s latest poll has black turnout at 32% and American Research Group has black turnout at 36% so 35% seems to fit in nicely with the other pollsters. But what bugs me is the media interpretation of the instantaneous change in the results due to the tweaking of the demographics… Do they understand that demographic weightings can considerably alter the final results of a poll?

Here is one such media report that I obtained with respect to the instantaneous switch without any reference to why/how it happened.

InsiderAdvantage, … shocked everyone … with a poll showing Clinton leading in NC, now shows Obama back on top, 49% to 44%.

If Insider Advantage had not tweaked the demographic turnout in their poll, we would see Clinton actually making slight gains. But the illusion is different.

It is all about turnout among African Americans in North Carolina.