18 May 2016

US Election 2016

Swing is a very British concept. The transition of voters from one party to another. Popularised by psephologists David Butler and Robin McKenzie, the swingometer has been a staple of British election nights since the 1950s. But could it be used for an American presidential election?

Taking cue from Peter Snow, I will be showing you exactly what could happen on US Election Night. The difficulty is that the swing in America is not uniform at all, so the swingometer may be completely pointless.

Indeed, Donald Trump need not do well in certain states in order to win the election. In contrast to the 150-200 seats at British general elections that change hands, only 10 out of 56 (don't ask) states have the potential to change hands. They are referred to as "swing states" and so the swing in these 10 states determines all the difference. Swing variation in these states in America is incredibly high, so it's left to us, the Brits, to act like sheep, wafting from Labour to the Conservatives and vice versa in unison.

But could you actually use a swingometer for the US presidential election?


And here it is.

The first thing that jumps out is just how well Barack Obama did four years ago. There's not much realistic territory for Hilary Clinton to get into as a result. Only North Carolina could she take, otherwise she would need a swing of 3.6% to take the next state, Nebraska's 2nd District. And no, that's not a rule from Mornington Crescent.

Because Obama won every "swing state" save for North Carolina, Donald Trump has a mountain to climb. Trump needs a uniform swing of 3.7%. After the Obama landslide in 2008, Mitt Romney could only muster up a swing of 1.7%, and so Trump really needs to do well.

That said, because Obama did so well, it will be difficult for Clinton to obtain a swing to the Democrats. A swing to Clinton of 1% would take North Carolina, giving her an extra 15 seats in the electoral college (don't ask). 

In fact, the target list for Clinton is quite thin, and it's more that she will be looking to hang on rather than attack. Resultantly, Trump's target list is quite expansive, and there are some winnable targets on here:

The nine swing states that Romney didn't win are listed as Trump's targets. Florida might just go to Trump because the majority for the Democrats was so narrow, a swing of just 0.4% would take this. The opinion polls, however, for Florida, strangely are very similar to the swing we are currently seeing in the opinion polls nationally - so maybe there may not be such huge swing variations this time around. Applying the current swing we see nationally to Florida, we are within the margin of error for that state's latest poll, which has Clinton on 43% and Trump on 42%; the uniform swing has 42% for Clinton and 41% for Trump.

But if Florida stays blue, then Trump has no chance. As well as holding North Carolina, he needs to win Flordia, Ohio, Virginia, Colorado, and Pennsylvania (in order of difficulty) to get into the White House. The other four "swing states" of New Hampshire, Iowa, Nevada, and Wisconsin are only worth 26 electoral college votes, and so if Trump loses Florida, Clinton is the US president.

If Trump takes Florida (as well as holding North Carolina), then he still needs Ohio, one of the first key marginals to declare. Losing this would be a disaster because he would still then need to go down as far as Wisconsin on the board even if he takes Florida.

Trump won't particularly mind if Virginia goes blue, because he then won't need to take Wisconsin. Similarly, if he wins those top three but loses Colorado, he need only take New Hampshire (4) and Iowa (6) to compensate. Pennsylvania is the other key state and, like Florida, is worth north of 20 votes. Consequently, lose this and he will need to take the other four states.

However, lose two of those five states and Trump is toast. If disaster occurs for Trump and he loses both Florida and Pennsylvania (but takes the other three) then he'll need a miracle and ridiculous swings to win. He'd have to go deep into Democrat territory to win. On top of the remaining "swing states", he'll need to take (in order of difficulty): Maine's 2nd District (4.3% swing neeeded); Minnesota (4.4%); and Michigan (4.8%).

Trump's only chance, then, would be for Gary Johnson, the (probable) Libertarian candidate, to take votes off the Democrats. Can that happen? Gary Johnson is what the Brits might call a "Thatcherite" (and is my personal choice for president), and so it seems more probable that he will take votes off many of the centrist or slightly-right-of-centre voters that voted for Mitt Romney in 2012. Johnson is polling at north of 10% currently though. What we don't know is who he's taking these votes off. Swing time! The last national poll to include Johnson has Clinton on 42%, Trump on 34%, and Johnson on 11%. Using that, here are the swings:

So it's evidently quite clear - Johnson is taking more votes off Republicans than he is off Democrats. So that sinks Trump.

But can Johnson win anything in terms of the electoral college?

Johnson cannot win any votes for the electoral college with these kind of swings, though, because of just how badly the Libertarians did in 2012. The best result was in the not-too-safe state of New Mexico, where the Libertarians polled 3.5%. Now, if that swing was replicated in New Mexico, Johnson would have around 15%, but the winning candidate (which should be Clinton) will still have about 44%. Trump would have around 35%. Again, this matches (near enough) the state opinion poll for New Mexico, so maybe America will behave more uniformly this time around. 

And after all that, it's time to ask the question for the other side - can the Democrats do any better than they did in 2012?

We've gone down to #6 on this list because Barack Obama managed to take Indiana in 2008 - but Romney won this back in 2012. It's not considered a "swing state", but as Obama had this in 2008 it's worth using as a barometer for a Clinton landslide.

North Carolina is the first example of non-uniformity - and so this may get shoved into Democrat territory so much so that whoever the Republican candidate is in 2020 will struggle to win it. Here's the uniform projection versus the state poll - and the swing for the state poll:

If the swing we see in the state poll of a whopping 7% from Republicans to Democrats is typical across all of America, then Clinton is looking at whatever a "landslide" is in America - 414 electoral college votes. She'd win Nebraska's 2nd (3.6% swing needed), Georgia (3.9%), Arizona (4.6%), Missouri (4.7%), Indiana (5.1%), South Carolina (5.3%), Mississipi (5.8%), and Montana (6.9%), giving her an extra 82 on the 332 Obama had four years ago.

How do the state polls in those seats shape up? Georgia shows a swing to the Democrats of 2%, Arizona 2.5% to the Democrats, Missouri's polls are too wide-ranging to be conclusive (so this could fall on the night to the Democrats!), Indiana 1.5% to the Democrats, and Mississipi 4% to the Democrats. In the other states listed above no polls have been done. But there's a pattern here - that the Republicans seem to be doing badly in their own states and matching the uniform swing in the Democrat-held swing states.

So if North Carolina goes, Trump needs to win those five states we talked about earlier as well as the next three. 

The uniform national swing shown on the swingometer is this:

It's oh-so-close to taking Florida, but not quite.

Looking at all the evidence above, I'm calling the election for CLINTON. The rise of Johnson to take votes away from the Romney-ites, the failure of Trump in his own states, and the likelihood of Clinton taking Florida gives Trump virtually no chance.

No comments:

Post a Comment