Well, there they go again.
Polls should measure public opinion, not attempt to sway it.
What makes for a good poll these days?
Well, the same as always. A good sample size, I prefer at least 1,000. And most importantly, a party makeup that realistically reflects the public at large. These days, I accept a poll that shows Republicans and Democrats somewhere around 30 percent. And the parties are so close that any sample size needs to be within one or two points of each other. 31% or 30% Democrats and Republicans (or vice versa) and the balance Independents. And the respondents should be Likely voters at this point.
As one who watches many polls, I consider it a waste of time to even read a poll with the mix of R’s D’s and I’s outside of the parameters I just shared without doing the math and properly adjusting the numbers.
But the narrative in the “Mainstream Media” has been the supposed big “bounce” that Obama has received in the polls.
The polls have not changed. They did not move a single point by my assessment. There was no post convention bounce for either party. All that moved was the number of Democrats polled vs. the number of Republicans. And some really skewed numbers with Independents – who favor Romney by double digits in numerous polls.
The top three polls this morning on Real Clear Politics have one thing in common (other than showing Obama with a lead). They all sampled far too many Democrats and 2 of them fudged the Independent voters.
Here’s the breakdown:
The CNN Poll shows Obama ahead by +6 points. They polled 43% Democrats and only 39% Republicans and a measly 18% Independents. The fact is, Independents are going to decide this election. Republicans and Democrats are pretty evenly split. So unless the Independent voters are trending 50-50, this 18% sample is far too small to reflect reality. But the Independents in this poll are not going 50-50, the poll shows that Independents are going for Romney by a large margin 54% to 40% for Obama.
So, if we adjust this CNN Poll to reflect reality, we would get a totally different picture.
This poll shows that 97% of Democrats plan to vote for Obama and 96% of Republicans plan to vote for Romney. And we already know the leaning of the Independents.
This poll surveyed 1022 voters. Let’s reset the numbers to reflect the optimum ratio mentioned above. And we will go with a slight advantage for Democrats just to be fair. Let’s adjust the numbers to 31% Democrats, 30% Republicans and the balance, 39% Independents. That will give us 316 Democrats, 307 Republicans and 399 Independents.
And using the same ratios from the poll, 97% of the Democrats will vote for Obama. That is 307 Democrats for Obama, 9 for Romney. For Republicans, 295 will vote for Romney and 12 for Obama. And the Independents – 215 will vote for Romney and 160 for Obama. (Some Independents would go for neither.)
479 for Obama (47%)
519 for Romney (51%)
So when we include the correct ratios from within the poll itself, Romney has a 4 point lead.
The CNN Poll is severely flawed because it under-samples the Independents who – according to the poll – are going for Romney by a 14 point margin. Instead of being up 6 points, Obama is actually down by 4. A whopping 10 point mistake!
ABC News/ Washington Post
This poll shows Obama up by only 1 point. But again, lets look at the numbers.
ABC-WaPo sampled 32% Democrats and only 26% Republicans. Independents were at 37%, which is fine. We only need to adjust the Republicans and Democrats.
The actual numbers were 321 Democrats, 261 Republicans and 371 Independents for a total of 953 voters.
Unfortunately, this poll does not break down the percentage of Independent voters preferences. But we can calculate the Independent ratio given the numbers we already have. We know 49% of all those sampled indicated they would vote for Obama and 48% said Romney.
Which means 49% or 467 said Obama and 457 said Romney. If we consider the cross party votes a wash, out of the 467 votes for Obama, 321 were Democrats which means 146 were Independents for Obama. And out of the 457 votes for Romney, 261 were Republicans meaning 196 Independents went for Romney. (A small percentage said ‘other’ or ‘neither’ which accounts for the remaining votes.) So out of 342 Independents who preferred either Obama or Romney, the split was 57% of the Independent vote for Romney and 43% of Independents said Obama. This gives Romney a 14% edge with Independent voters – which is exactly the same as the CNN Poll above.
Now that we have all the numbers we need, let’s correct this poll.
We should have sampled 305 Democrats (32%) and 295 Republicans (31%). And since the cross party voting is both small and pretty much equal, we can ignore those few votes as a wash.
So the adjusted Obama vote would be 305 Democrats plus 146 Independents were for Obama giving him 451 Votes. And Romney would get 295 Republican votes plus 196 from the Independents for a total of 491.
The corrected totals would be 49% for Romney and 45% for Obama.
This is exactly the same lead Romney has in the adjusted CNN Poll above.
This poll shows Obama leading by 2 points. But let’s look at the numbers behind the poll.
They sampled 36% Democrats and 31% republicans. The remaining Independent voters were evenly split for Obama and Romney at 47% each, which is not reflective of what other polls have shown. They under-sampled Independents – only 30% – but since the candidates polled even among Indy’s, we will ignore this – although I believe their 47% each finding is incorrect and inconsistent with all the other polls out there showing Romney doing well among Independents.
So if we adjust the number of Democrats to be a more realistic 32%, we would see Obama’s 2 point lead turn into a 2 point deficit.
This poll is pretty suspect and lacking in enough numerical data to fact check the conclusions.
But still adjusts to a 2 point Romney lead.
The polling companies are pulling out all the stops to make Obama look good. Most of these are either intimidated and compromised by the Obama Administration like Gallup (as we exposed here) or simply in the tank for Obama.
Realistically, this is a tight race, but when the polls are properly adjusted to reflect the actual makeup of the country, it is pretty that Romney has 2 to 4 point lead.
Campaigns always say to ignore the polls – unless they favor them – and the Romney Campaign is no exception. But in this case, I believe the Romney folks are correct. Overall, Romney looks to have about a 4 point lead over Obama when you use a correct sample size. Wherever possible, we tried to use the actual numbers within the polls, or at least extrapolate them where we had sufficient data.
But the one conclusion that seems consistent is that the polls are intentionally skewed!