Since my Swingometers are using terminology more often used in discussion of British elections, not American, it’s not surprising that some are unclear on just how they work.
Here’s an explanation.
The Swingometer doesn’t look at the raw election results. Instead, it subtracts all votes for candidates other than the Democrat and the Republican. From what’s left, the Democrat’s percentage and the Republican’s percentage of the two party vote are calculated. Those percentages will always add up to 100.
The difference between the two is then taken as the margin of victory. So if a Democrat took 55% and the Republican took 45%, the result is D+10.
The swing is then applied to that margin. A swing of 10 would take the election from D+10 to a tie. In a two candidate race with no third parties, the swing is simply double the amount one candidate gains or loses. Every vote switched has a doubled impact on the swing because vote switches are zero-sum. The two-party vote by definition goes to a Democrat or a Republican, and never anywhere else.
What’s interesting about the two-party vote is that we have a poll that in mid-term elections is decently predictive of that for the House of Representatives. The Gallup generic ballot is useful when there’s no Presidential poll included with it to make the poll less accurate for the House questions.
2010 being a midterm election, we can use the swing from the actual 2008 House results to the 2010 Gallup poll to make a crude projection of the final House results.
It beats trying to analyze 435 individual races, to be sure.