In several previous posts, I noted that the Democrat and media polling (nearly interchangeable terms) historically undercounts Republican voters in the spring and summer before a general election, only to add them back in as the election nears. Rasmussen and a small number of other polls are the exception in that they ask respondents under which party they are registered to vote and then reweight their respondents to match the partisan makeup of the electorate.
This election season appears to be no different.
Huff Post Pollster operates an nifty interactive chart which offers the average partisan distribution of the polls over time. I have created a chart that shows the partisan distribution of the media polls without Rasmussen from April 2012 through the present, which you can find here.
As you can see on the Pollster chart, the media polls reduced the number of GOP voters in their samples by 4 points from 29.6% in April to 25.4% by August 5, 2012, moving from a 5% bias for Democrats to a 10.6% bias.
Then, the media polls began to miraculously find millions of missing GOP voters after the conventions until Republicans again make up 30% of the polling electorate today and the polls are back to an average 5% bias for Democrats.
You can see this shift in individual polls over the past week.
Last week, the ABC News/Washington Post poll found more Republicans and halved their Democrat bias to 3%.
Republican voters increased their share of voters by five points from 30.3% to 35.2% of the electorate.
Democrat voters decreased their share of voters by four points from 36.6% to 32.5% of the electorate.
Independents remained largely unchanged, sliding slightly from 30.1% to 29.5% of the electorate.
In the space of a month, the Pew poll electorate shifted from looking like the 2008 to strongly resembling the current Rasmussen partisan breakdown of registered voters.
I have a theory that we Republicans migrate out of the country like a flock of birds during the summer of election years, only to return during the fall to be found again by our intrepid Democrat and media pollsters.
Or maybe the Democrat and media pollsters intentionally game their polls to provide the results their Democrat media clients would like to see and then shift back as the election approaches to offer roughly realistic polls that basically resemble the final election results in order to maintain their credibility.
What do you think?
UPDATE: Gallup will shift its polling from registered to likely voters tomorrow and a depressed Susan Page from USA Today told PBS that she expects large shift to Romney because “Republicans are more energized and more likely to actually go and vote.”