Pew Polling released a new poll tabulating data that they collected after the Republican debate. The poll was taken between 10/4 and 10/7. The results show Mitt Romney with a 4 point lead on Barack Obama 49% to 45%.
The previous poll by Pew showed Obama with an 8 point lead over Romney 51% – 43%.
So did one debate actually see a 12 point swing from +8 Obama to +4 Romney?
Absolutely not. (Sorry to disappoint my Republican friends!)
A companion site started by Virginia Right! several weeks ago called Political Poll Check was founded just for polls like this. What does a 12 point swing in a poll when compared to the previous poll from the same outfit mean? It must be apples to apples, right?
I am sorry to report is is not apples to apples.
When we analyzed the previous Pew poll taken 9/12 to 9/16, the poll received a grade of ‘C’ and scored 74 points. The poll sampled 37% Democrats and only 31% Republicans. The skew was +6 Democrats and the Skew Factor, which is the deviation from our standardized baseline, was +9 Democrats.
After our Normalization process to account for this skew factor, we found that the poll actually showed Mitt Romney up by 2 points.
(Normalization, to put it simply: “Say you had 3 recipes for soup and you wanted to compare the ingredients. One makes 25 servings, one makes 50 and one 100. They have exactly the same ingredients, but for some reason, when you make the recipe for 50 servings, it tastes a lot better and is everyone’s favorite. Normalization converts all the recipes to 1 serving so you can easily see that the one you like most has more butter and salt.)
So, looking at the latest poll released by Pew, they now show Romney has a 4 point lead over Obama. But there were several distinct differences in these two Pew polls. The 9/12 poll sampled 37% Democrats and 31% Republicans. The latest poll sampled 32% Democrats and 35% Republicans, which means Republicans were sampled with a 3% advantage. This PEW Poll is almost dead on accurate with the Normalization sample we use which is the average of a party affiliation poll conducted by Rasmussen for 2012.
There was absolutely zero skew factor.
This poll received an 88, which is a B+ and the highest poll grade to date (out of 30 polls).
When we normalized this poll to our standard set, it showed Romney with the same 2 point advantage as the normalized sample of the 9/12 Pew Poll.
The Gender Factor:
We do not normalize for race or gender. Most of these polls are pretty good on getting these demographics close enough to the census data we use to grade the polls. The United States has approximately 48% men and 51% women (rounding drops 1%). There is a difference in both of these polls. The older poll sampled 7% more women than men and the latest poll favored women by 9%. The older poll showed women voting for Obama 56% to 38%. The newest poll shows Obama and Romney each draw 47% of women, so the gender gap has closed as far as females. However, the older poll also shows Romney with a 2 point advantage with men, and the latest poll shows that is not an 8 point advantage to Romney.
If we were to factor in the gender sampling errors, the latest poll would probably show a point or two swing more for Romney. So there was a bump in the polls, but there was no 12 point swing.
But the gains with both sexes and a large bump in enthusiasm for Romney shows that Romney not only has a lead on Obama nationwide, but that he is trending up while Obama is trending down.
These next two debates will be very telling.