There are two kinds of polls.
Those intended to “take the pulse” and get an accurate picture, and those skewed to attempt to influence thoughts and opinions.
Political candidates hire pollsters to get an accurate picture of how they are doing, where they are doing well, and where they need improvement.
These are the important ones.
Others run polls and intentionally skew the results to give the perception that their side is winning.
A recent NBC News-Marist Poll is a case study for polls trying to change the public perception that one candidate is falling behind.
A few days ago, Roanoke College came out with a poll showing George Allen rising to an 8 point advantage over Tim Kaine.
I am skeptical of all polls until I look at the numbers behind them. I am not interested in knowing the results of a poll that is not truly reflective the voters. I want the truth.
The Roanoke College poll was pretty well balanced. I like polls to have over 1,000 as a sample size, but if not, the ratios must be tightly controlled. Roanoke College passed my smell test. Based on the sample, I agree that Allen looks to have an 8 point lead, perhaps a bit less, but at least 5 points.
So when I saw a poll this weekend showing Tim Kaine had jumped to a 10 point lead, either something horrible had happened with the Allen that I had missed, or Kaine was Sainted.
Well, neither had happened. So I looked at the numbers behind the Marist Poll.
The quick answer came in the number of Democrats vs. the number of Republicans sampled.
In the Roanoke College poll, the mix of Republicans (30%), Democrats (28%) and Independents (31%) was pretty good. Virginia is pretty evenly split these days with the past couple of elections leaning a bit more to the Republican side.
And how evenly did the Marist Poll split the leanings? Well, not good at all. Those identifying as either a Democrat or leaning Democrat totaled 50%, Republican 36% and Independent just 14%.
So what gives with that? I can promise you that if this was a representative sample the Virginia Senate would not be in a 20-20 tie!
But even skewed polls tell us something.
Each candidate will get the majority of votes from their own party as well as a percentage of the Independents.
With 50% Democrats sampled in the poll, you would expect that Kaine would get most of the Democrats and some of the Independents. You would also expect the same from Allen. And if we consider the Democrats who will vote for Allen a wash with the Republicans who will vote for Kaine, we would expect Kaine to get 50% plus a few Indy’s and Allen to get 36% plus a few Indy’s.
So how did our Senate Candidates do?
Allen did pretty much as you could predict. 39%. All of the Republicans and some of the Indy’s.
And Kaine? He only got 48% which indicates that he does not even get all of the Democrats and none of the Indy’s.
So if this poll were properly weighted to reflect the actual political landscape of Virginia, Allen would have at least a 6 point lead over Kaine.
NBC News releasing such a skewed poll is no surprise. The fact that Marist would release such a skewed poll and risk damage to their reputation as a polling firm either means they are also in the Democrat’s tank or NBC paid them a lot of money for this.
Marist Polls normally doesn’t play politics this much.
They should really be ashamed of conducting such a biased poll and not at least attempting to weight the poll to reflect reality.
I have saved this poll to my archives. I will be watching Marist Polls more closely in the future.
And if they continue this pattern of skewed results, I might suggest a slight name change.
MarXist Polls sounds appropriate.
Article written by: Tom White
