Some interesting things to consider.
Mitt Romney holds large lead in UnSkewedPolls.com average of polls
While Mitt Romney enjoys a 47 percent to 46 percent lead over President Obama in the Rasmussen Reports Daily Presidential Tracking Poll released today, he has taken a 7.8 percent lead in the UnSkewedPolls.com average of unskewed polls. The UnSkewed Average includes the results of “unskewed” analysis of several well-known national polls, that typically over-sample Democrats and/or under-sample independent voters. Here is an example illustrating the process of “unskewing” a skewed poll that over-sampled Democrats.
The current UnSkewedPolls.com average reported today from the website includes eight recent polls, each one the latest conducted by that polling or media organization. When unskewed and averaged, they show a Mitt Romney lead of 7.8 percent.
The most notable of the polls is the CNN/ORC poll released last Monday that reported Obama leading by a six percent, 52 percent to 46 while it massively over-sampled Democrats and under-sampled independent voters. Unskewed, that poll's data calculated to a 53 percent to 45 percent margin in favor of Romney.
The last NY Times/CBS News poll was equally as skewed. It reported the presidential race 49 percent Obama to 46 percent Romney while the unskewed analysis of that poll's data revealed a 51 percent Romney to 44 percent Obama result. This poll relied on an unusually large sample of Democratic voters.
The UnSkewedPolls.com average included six other polls, all of which were similarly skewed due to the sampling issues mentioned above, and produced results showing Obama finishing stronger as a result. Unskewed, the data from these polls indicated a substantial Romney lead in every single instance.
In the past many have relied on the Real Clear Politics average of polls. In past election cycles it might have seem that there were one or two polls that seemed to favor the Republicans, and maybe a few that somewhat favored the Democrats, and averaging all the major polls produced a number that seemed fairly accurate. In past election cycles, the RCP average was fairly accurate for that reason.
During this election cycle, it seems we have many more polls that are so much more skewed and biased in favor of the Democrats, primarily a result of heavily over-sampling Democrats in the samples and not weighting them in the analysis to counter that effect. As a result, the RCP average itself has become quite skewed from being dominated by so many skewed polls. The current RCP average includes 10 polls, of which eight of them are heavily skewed and have been unskewed for inclusion in the UnSkewedPolls.com average of unskewed polls.
The purpose of unskewing the polls is to arrive at accurate numbers, not to show one candidate or the other ahead in the presidential race. The analysis process in unskewing polls relies on the Rasmussen Reports partisan data measured from hundreds of thousands of voters by Rasmussen Reports, which measures the partisan percentages at 37.6 percent Republicans, 33.3 percent Democrats and 29.2 percent independents.
Journalists and analysts can honestly debate this premise and assert their own educated prediction of what the partisan makeup of the electorate on election day will, but within some degree of reason and reasonable probability. While 2008 was a heavily Democratic election, the percentage of Democrats voting was about six percent more than that of Republicans voting in the actual election. The 2010 election that saw Republicans gain 65 seats in Congress was about the opposite of that.
During this year, most states have seen large increases in Republican voters registrations while registration of Democrats has declined. In many instances where both parties held state and local primaries on the same day, the Republican primary typically had more voter participation. There is overwhelming empirical evidence of higher enthusiasm and participation levels of Republicans than Democrats this year. The 4.3 percent Republican edge that Rasmussen Reports is projecting would appear to be a fairly small-c conservative projection given the available evidence. Rasmussen Reports has invested far too much in their reputation for accuracy to gamble that way taking chances in projections of this sort.