Last-minute prediction: Obama by a landslide! No, wait… a 269-269 electoral college dead heat

After all my poll-watching, I’ve decided to throw my hat in the rang one last time, and engage in that most dangerous of sports: the last-minute prediction. Last minute predictions give the predictor the least time to live in comfort in the time between prediction and outcome, and most danger of subsequent accusations of being wildly wrong despite having all the best poll data to hand. Hence, they’re not that popular.

Still, they are fun. So here goes. Let the ridiculing of me start in about 24 hours!

McCain has 134 votes from sixteen states ‘in the bag’, all unsurprisingly Bush states from 2000 and 2004: Texas, Tennessee, Arizona, Alabama, Louisiana, Kentucky, South Carolina, Oklahoma, Arkansas, Kansas, Mississippi, Nebraska, Utah, Idaho, Alaska and Wyoming. Of these, only Arizona even looks like wavering, and I think home bias will win through there. A further four states are going McCain’s way, worth 26 votes in total: Georgia, West Virginia, Montana, South Dakota. These twenty states would give McCain 160 votes.

Obama has 218 electoral college votes ‘in the bag’, it seems at this stage: all the Kerry-Gore states from the last two elections, plus Iowa (which went to Gore last time out) and New Hampshire (which went to Bush). That is, armed with little else but a range of polls and past voting records, I think Obama will easily win the following 18 states:

  • California, New York, Illinois, Michigan, New Jersey, Massachusetts, Washington, Maryland, Connecticut, Iowa, Oregon, Hawaii, Maine, New Hampshire, Rhode Island, DC, Delaware and Vermont

I also think it’s very likely that another 41 votes across three states are Obama’s, bringing him to 259 and just an Indiana or Missouri away from the US presidency: Pennsylvania, Minnesota, Wisconsin, again all Kerry-Gore states since 2000. (Although with Maine splitting its electoral college votes, technically he may only have 257 from these 21 states, but will more than likely make up for this by picking up two in Nebraska, which operates the same system.)

So Obama’s 21-20 victory in states would leave him 259-160 ahead.

What makes it ominous for McCain is that a further 47 electoral college votes look like they could be going Obama’s way – although these are most susceptible to a Bradley effect swing of say 5%, as his lead is smallest in these four states: Ohio (20), Virginia (13), Colorado (9), New Mexico (5). Of all these four states, only New Mexico in 2004 did not vote for Bush in the last two elections. If the Bradley effect or some other gap between poll and reality were to deny him even two of these states, Obama would still win the presidency.

And that’s before factoring in the six true toss-ups worth 72 votes, which interestingly are all Bush 2000-2004 states: Florida (27), North Carolina (15), Indiana (11), Missouri (11), Nevada (5), North Dakota (3). Current polls suggest this will probably split down the middle, maybe 32 to Obama and 40 to McCain. Taking these at face value, and assuming either (a) no Bradley effect, or (b) that the Bradley effect and the no-landline effect cancel each other out suggests a win for Obama in the region of 338 to 200.

If we take states like Arizona as safe and call this margin the top end, what’s the bottom end of the spectrum? A 5-point Bradley effect, or something similar, could cost Obama Florida (27), Ohio (20), Virginia (13) and possibly Colorado (9), worth together 69 votes. The mathematically sharp-minded amongst you will have spotted that a 338-200 outcome with 69 votes redistributed gives us: 269-269, a tie!

Personally, I think we’ll be closer to the upper end for Obama, but the fact that the polls leave us the possibility of a tie is certainly one to watch, if some states start falling unexpectedly on McCain’s side. Failing that, it may bring some electoral attention away from Florida and Ohio and Pennsylvania to Colorado for a change!

Advertisements

Eight weeks work yields seven states for Obama, just Montana (maybe) for McCain

Between mid and late September, the big story in terms of the state-by-state polls was McCain’s loss of his ‘medium’ states – i.e. those where he had some lead, but not a large one – such as Florida and North Carolina into ‘toss-up’ territory. Obama had managed to garner a few extra ‘strong’ electoral college votes, but nothing seemingly irreversible.

So, what’s happened in the last four weeks? Well, Obama’s overall increase in the national polls is very much old news at this stage, so what can state-by-state polls tell us above and beyond? Obama’s national performance is unsurprisingly replicated in RCP’s state-by-state poll statistics: he has managed to increase his vote by 3% or more in 22 of the 35 states covered by RCP, while McCain has only done similar in two states: Maine, which is solid Obama country, and Texas, which is of course safe McCain territory.

Of most note, from the state by state results, are the following:

  • Pennsylvania (21 votes), exceeded in electoral college votes by just four states, now looks like solid Obama territory, where his lead has increased from 48-44 to 51-41.
  • Ohio, worth 20 votes, has also moved from a toss-up towards Obama – four weeks ago, McCain had 1 point lead. Now, he trails 50-44.
  • Michigan (17) now looks unassailable for Obama, while the much-mooted Georgia (15) upset remains a possibility: in the last four weeks, McCain’s lead has shrunk from 12% to 6.8%.
  • There have been significant gains for Obama in other Tier 2 states, such as Virginia (13), and in particular Minnesota (10) and Wisconsin (also 10).

They are the most important changes – most other gains haven’t really had an impact. So where does that leave us all with only a couple of weeks left?

On September 1, McCain had 99 electoral college votes from 9 states of the 32 or so polled ‘in the bag’ (i.e. with a lead of 10% or more) and just one other state leaning (Georgia). Eight weeks on, all he has to show by way of gains are the three electoral college votes of Montana, where he enjoys a 5.6% lead.

Barack Obama could count, eight weeks ago, on the 156 electoral college votes of eight states, varying in size from California to Maine. Since then, he has added a further seven states – most notably Pennsylvania (21) and Michigan (17) as per above – and seventy-six electoral votes. Another 47 votes in four states are leaning towards Obama, including the sizeable Ohio (20) and Virginia (13).

So, with all that bad news for McCain, where are his glimmers of hope? Unfortunately, they don’t look like they’re hiding in the remaining undecided states. The five states that are still in toss-up territory, as per RCP’s listings, contain 69 electoral college votes – 42 of which are in Florida and North Carolina. Both of those states lean slightly towards Obama, as do two of the other three.

Perhaps his hopes lie in the possibility that people are more likely to say they’ll vote for Obama than actually will, as some have suggested. Or that the turnout from Obama demographics such as the young and African-American will be low. Then again, for each commentator saying that, there are others who believe that young voters are grossly under-represented in traditional polls conducted using landlines. Who knows?

Still, it could be worse… Just take a look at the Economist World Vote. Obama 8,957-31 McCain. Cuba’s status is surely the biggest turn-up for the books!

Economist.com World Vote, Oct 23 2008

Economist.com World Vote, Oct 23 2008

Altogether, a simple sum of all the strong, medium and weak votes for both candidates gives Obama a whopping 337-128 lead. As was famously once said, “they think it’s all over…”

Lean on who? McCain’s problem – RCP poll analysis 09/29

A couple of weeks ago, at the height of Palin-mania, I did a quick cross-check of RCP’s electoral college polls compared to how they’d been at the start of September. At that time, there was lots of bad news for Obama, in Pennsylvania, Michigan and Indiana, in particular. To compound that, McCain had turned lots of his ‘weak’ electoral votes (EV) into medium and strong votes, in places like Florida, Georgia and Virginia. On Sep 12, according to my analysis of the RCP figures, McCain was only one strong EV behind Obama (130-129, strong being 10 points or more ahead in the polls).

I wondered at the time, whether it represented a shift change in the race, i.e. was the Palin announcement setting a new average around which the numbers would shift, that average being closer to 50-50 than before, or was the announcement and its aftermath just one of the ups and downs around the existing average, i.e. that the average was going to stay something like 53-47 in Obama’s favour? It seems, if the evidence of the last two weeks or so is anything to go by, that it is more the latter than the former. Maybe it is the turmoil in the finanical markets, maybe it is the public tiring of Palin for a while as they did with Obama earlier in the year, or maybe it is something else again, but the last 15 days or so will have left the Obama-ites a lot happier than the McCain-ites.

A first glance at the polls could leave one to argue that it is not as cut and dry as one might think. McCain has actually gained ground, in a statistically significant sense (i.e. 3% or more) in more states than Obama (3-2). However, it’s the composition of those gains that is important – McCain’s gains have been non-battleground states, including New York and Maine, where his team probably doesn’t rate his chances, and Montana, which he was probably counting on anyway. Obama’s two gains have come in very interesting states: North Carolina, where he has overturned an 11% poll deficit to now have the slenderest of advantages 46.8-46.5, and New Mexico, which (by the simple rule of 5% or more means leaning) is now leaning to Obama 50.3-44.3. In addition, Michigan is now leaning to Obama (+6.6%), while Florida is back to being a toss-up state (46.0-47.6, in McCain’s favour). Ditto Missouri.

Further down the electoral vote pecking order, McCain did get some good news: Minnesota and Wisconsin have moved from leaning to Obama to being toss-ups – interesting to see how much the location of the RNC affected the former. The net of it all, in electoral vote terms, is the following (and bear in mind that the RCP analysis only covers about 480 of the 540 or so electoral votes):

  • Obama’s has managed to take 17 weak EVs and move them to strong (+14) and medium (+3). His strong EVs amount to 144, up from 130 on Sept 12 (but still below the 156 on Sep 1).
  • At the same time, he’s upped his weak EVs from 56 to 73, giving him a total of 274 – enough to win the Presidency even without the 15 smaller states missing from RCP’s analysis, were results to go exactly as the polls.
  • McCain, on the other hand, has seen a slight fall in his strong votes, from 129 to 117. It’s his medium level votes, though (that is, those states where he has a 5-10 point advantage) that have collapsed. Literally. While there are 11 solid pro-Bush 2000-04 states that McCain has ‘in the bag’ so to speak, Florida and Missouri’s conversion to toss-ups means that there is not a single state in the USA that leans to McCain, i.e. where he has a lead of between 5 and 10 points.

The upshot is that the simple sum of RCP polls gives Obama an 83-vote lead, up from just 27 votes on Sep 12. Better still for Obama, that lead is entirely made up of medium to strong poll showings, i.e. a lead of 5 points or more. Still, elections can swing on small margins and it is worthwhile noting that there are now almost 150 electoral votes in toss-up states (i.e. where neither candidate has a lead of 5 points). Were something small to happen to push the marginal voter back to the GOP, who knows what could happen? Vice-presidential debate anyone?