In a truly tied race, you'll have polls on both sides of the flat line. I think that's what we're really seeing.

Mary Burke must be doing something right.

That's enough for a lot of us.

if she is going to get them to vote.

They have a history of over-polling Republicans. If they suddenly go the other way I would suspect it was an effort to build credibility for the big games.

Perhaps, not very, ergo:



https://www.google.com/?gws_rd=ssl#q=gravis++marketing+scam

for someone to run a bogus polling operation without actually collecting data. Why over-sample Republicans when you can just invent them?

In my imagined scheme, the poll would show the paying candidate as doing better than in the rest of the polls right up to the time of the election, at which time you would bring your invented results in line with the findings of the more legitimate polls. You'd be selling the PR value of showing your client ahead in the poll until the very end, at which time you'd "detect" a sudden change in the mood of the electorate and end up with a polling result that, because it is simply based on the average result of the real polls, would produce a respectable predictive track record and retain the appearance of legitimacy.

There have been instances where polling firms have been accused, with good evidence, of just making shit up.


I subscribe to Hanlon's razor; never attribute to malice what best can be attributed to incompetence. I suspect Gravis accumulating the raw data but is clueless when it comes to collating it.