A US pendulum: it don't mean a thing if it aint got that swing

It’s going to be a big day for election watchers. In Australia, the essential accompaniment for days like this has long been the pendulum, developed and tirelessly popularised by Malcolm Mackerras. But American commentators use a different set of tools, so they require some interpretation for an Australian audience.

Fundamentally, a presidential election is not so different from a general election here. Just as victory here depends on getting a majority in the lower house of parliament, in America it depends on winning a majority in the electoral college. In each case, candidates need to win not the popular vote at large, but particular geographical areas — electorates in Australia, states in the US — some of which are more marginal than others.

Hence Mackerras last month was able to present, as he has in past years, the presidential contest in pendulum form. Let’s see how it helps tell us what to watch for today.

First the easy part: the Democrat side of the pendulum, showing states that John Kerry won in 2004 and would be at risk with a swing to the Republicans. Here are the most marginal, up to the 4% mark, with the number of votes (in brackets) that each state has in the electoral college. (Two small states, Maine and Nebraska, allocate their electoral college by congressional district instead of winner-take-all, so their districts get separate entries on the pendulum.)

Delaware (3) 3.8%
Washington (11) 3.6%
New Jersey (15) 3.4%
Maine 2nd District (1) 3.0%
Oregon (7) 2.1%
Minnesota (10) 1.8%
Michigan (17) 1.7%
Pennsylvania (21) 1.3%
New Hampshire (4) 0.7%
Wisconsin (10) 0.2%

The swing is pretty clearly going in the other direction, so there isn’t going to be much interest in these states. Since the McCain campaign gave up on Michigan a few weeks ago, the only Democrat states in any doubt have been New Hampshire and Pennsylvania. They are on such small margins that a surprise result is possible, but most unlikely.

Note that the margins are all two-party figures: for each state I have factored out the minor party vote. Since there is no preferential voting in America, people who vote for minor parties have no effect on the result; they are conceptually the same as those who stay home.

Results in the US, however, are typically reported as raw figures; hence 60-40 and 54-36-10 would appear as a 20-point lead and 18-point lead respectively, even though in two-party terms they are the same (the extra 10% could be minor party votes or, in an opinion poll, just undecideds). So when the polls put Barack Obama a little over 50%, that’s more like mid-50s in two-party terms. For comparison, John Kerry got 48.8% of the national two-party vote in 2004.

Which brings us to possible Democrat gains. Here is the Republican side of the pendulum, up as far as 11%:

Nebraska 2nd District (1) 11.0%
South Dakota (3) 10.9%
Montana (3) 10.5%
Indiana (11) 10.4%
Kentucky (8) 10.0%
Mississippi (6) 9.5%
South Carolina (8) 8.6%
Georgia (15) 8.4%
Louisiana (9) 7.3%
Tennessee (11) 7.2%
West Virginia (5) 6.5%
North Carolina (15) 6.2%
Arkansas (6) 5.3%
Arizona (10) 4.9%
Virginia (13) 4.1%
Missouri (11) 3.6%
Florida (27) 2.5%
Colorado (9) 2.4%
Nevada (5) 1.3%
Ohio (20) 1.1%
New Mexico (5) 0.4%
Iowa (7) 0.3%

Kerry won 252 of the 538 votes in the electoral college, so to win a majority Obama would need a uniform swing of just 1.1%, to pick up three states and bring his electoral vote total to 284. Ohio, with 20 electoral votes, is the median state. At 6%, which is about where the polls are, Obama would gain ten states on a uniform swing, with a total of 113 electoral votes.

Mackerras would be a rich man if he had a dollar for every time he’s had to say “swings aren’t uniform, but deviations from uniformity roughly cancel out”. That’s the case here as well. Even on a big swing Arkansas doesn’t look like shifting, and Arizona (John McCain’s home state) will be a real ask. But states further up the pendulum are looking vulnerable instead: North Carolina, maybe Indiana, possibly even Georgia or Montana.

Just as in Australia, the pendulum gives a good guide to the magnitude of gains or losses for a given swing, as well as an indication — never exact, but better than random — of which individual results to look at.