One Example of a Questionable Poll, Update: And One More

These polls are all too common.  After days of a much closer match-up from polls like Rasmussen in Iowa, a new poll was released by Gravis Marketing which simply does not add up when you examine the data.

In it, it reflects President Obama with a 4-point lead in the state.  In addition, it has early voter tallies of 63 for Obama and 28 for Romney.

Let the unskewing begin:

First, the poll reflects 49% for Obama with 45% for Romney, with 6% undecided.  According to most pollsters, the undecided voters always tend to go for the challenger and not the incumbent.

Second, the early-voting numbers are mysterious.  While Romney leads voters who are voting on Tuesday by 54-42, Obama leads with early voters by 63-28.

According to the National Journal which gets its numbers from the Iowa Secretary of State, requested ballots total 557,432 ballots broken down by the following party percentages:

43.3% / 32.2% / 24.4% (D/R/Other)

If you give Obama all of the Democrats: 43.3% plus all of the Independents: 24.4% this would equal about 68%.  The Gravis poll however has Romney leading Obama among independents by 5 whole points in the overall poll.

Now, I understand that maybe they picked a larger sample of the Democrats among early voters alone.  There are different ways to achieve percentages.  But you would think that there would be some kind of a correlating pattern between the actual percentages of absentee ballots sent out by party and the actual returns.

Finally, the poll over-samples Democrats by +6 points.  Yet, Obama only comes out with a 4-point lead.  That’s what most people see and move on.  Yet, it suggests that 2% of Democrats won’t vote for Obama while the early-voting results suggest that perhaps some Republicans may be voting for Obama.

At the end of the day, something clearly stinks or someone has a broken calculator.


And here is another.  The latest NBC/WSJ/Marist poll shows Obama leading Romney in Ohio by six points as of today.

Their polling sample?

Strong Democrat, Not Strong Democrat, and Democrat Leaning Independent = 48% of their sample

Strong Republican, Not Strong Republican, and Republican Leaning Independent = 41% of their sample

(1302 Posts)

Leave a Reply