Today, I decided to test the cost-benefits of swinging an election across six scenarios:
- an acceleration of the current very favorable-to-Obama momentum (409-129; this entails winning states like Louisiana, Georgia, Arizona and Alaska)
- the current trends continue through Nov. 4 (372-166; Montana goes blue scenario)
- a slight fallback off the current trending in John McCains favor (which appears to my eye to be worked into 538.com's forecasting) (369-169; West Virginia and North Carolina at risk)
- a noticeable swing that places McCain within striking distance but it's still a long shot (349-189; Ohio, Nevada, Missouri at risk)
- significant momentum change in McCain's favor, but not enough (313-225; Washington, Maine, New Mexico, Florida at risk)
- very strong charge by McCain in the last few weeks, widespread reversal of the election dynamic (McCain win in a squeaker, 272-266)
(more below the break)
The first rule of election tampering, obviously, is that doing so has a chance of influencing the final outcome. Obviously, if Obama is on course to win by 290 points in the Electoral College, it's probably not going to do much good to fiddle with absentee ballot mailings, swap out good voting machines for bad ones, send out misleading 'helpful' notices to scare off voters, etc. Likewise, the 206-point scenario isn't conducive to fiddling with the ballot boxes. On the other hand, it's not like we are talking about big stakes here - it's only control of the greatly-enhanced power of the American Presidency. Hardly worth the effort of cutting corners and bending the rules.
But just in case someone doesn't reach the same conclusion we do, let's see if we can utilize statistics to guess where they might try their luck.
Repeating the Caveat
So you know: I'm not adjusting for the particulars of state election laws or infrastructure, so discount the findings accordingly. This is meant to be a high level exercise, designed to illustrate to raise awareness of, all other things being equal, which states would produce the return on cheating investment due to a combination of (1) their electoral vote count (2) how close the race is expected to be in each of the scenarios listed above. We use a plug-in 3% 'successful swing and we don't get caught either' assumption just to keep things simple. I will leave questions of voting security to the many experts who check in regularly here and, I hope, will have commentary to add to this discussion. Again, this is all about identifying which states interested persons might want to keep an eye on, depending on how the race goes.
Forecasting Methodology
I use Nate Silver's data from 538.com, specifically the current polling averages, by state to get what I assume is a current-moment data point for each state, then the regression forecast for each state, to get a simple linear trendline. In my observations over several weeks, this Obama trend slope (strongly favorable in almost every state), is pulled back by 50% in order to derive the daily estimate.
When I use this approach on the data I use from 538.com, the results that add up closely to Silver's daily Electoral College vote count, though obviously it's all-or-nothing by state (and not Monte Carlo'd) so some variation is inevitable. For example, using this morning's inputs, I get a 369-169 result. The most noticeable differences I see with Silver's map are that I award West Virginia's EV's to Obama and while Obama wins North Carolina in this scenario, it's a close call.
Deriving Scenarios
Personally I think that handicap will disappear rapidly as we are now 20 days away from the election. However, as past presidential elections have demonstrated - as this one has several times! - momentum shifts can be sudden and strong. I am also of the belief that there is little more wrong that the Republicans can do from a purely political standpoint to further harm their chances of retaining the White House. Regardless, I leave them a little room to surprise me.
Each scenario is discussed separately; still, I wanted to have an idea how likely each was, so I estimated some probabilities assuming a rather 'fat' single-tail distribution
16% Score +0.5 momentum change
50% Score, Baseline "mo"
14% Score -0.5 mo
12% Score, -1.0 mo
8% Score, -1.5 mo
6% Score, -2.0 mo
Question: When Does Cheating Prosper?
Answer: When the rules of fair play aren't working out for you, time is not on your side, and you think your chances of success are more than good enough to offset the consequences of being caught cheating.
Now recall the margins of victory listed in the introduction. All of the top four scenarios are at least 150 point victories in the Electoral College for Barack Obama. This is based on independent national survey numbers, not the Obama campaign. Fair play is working just fine for the blue team.
Recall the election is 20 days away and polls have Obama close to 10 points up. Time is on Obama's side.
There is no upside for the Obama camp to cheating by, say, engineering the theft of the electoral votes of, oh, let's come up with a mildly credible case... Arizona...just to show up John Mccain. (That jest made, I am not convinced Arizona is a shoo-in for McCain. Neither is Alaska assured for Palin.) Obama could SPOT McCain California in the Electoral College in four most favorable (and likely) scenarios and still win.
It's just not a gambit that there's any reason whatsoever for the Democrats to contemplate. Vast leaps of derogatory faith have to be made, in order to establish a rationale that 'liberals just can't help but cheat, it's in their nature', etc., etc. Nasty stuff, almost as bad as the screed poured over the hate radio. Regardless of what Obama's detractors maintain, there is still the matter of rational self-interest: Why take the risk? It's a sucker bet for someone who is at the moment 6-8 points up in the projected popular vote and 200 EVs up in the forecast Electoral College count.
Of course, there is another person in this foot race, someone who is not winning by the rules of fair play, for whom time is running out. Questions of ethics, character and legality set aside, there is regardless an incentive to investigate alternatives to letting good sportsmanship take its course.
If you aren't part of the campaign whose candidate is by all accounts way up and very much likely to win in three weeks, that is.
The At-Risk States
How we decide this is pretty basic:
- big states
- with lots of EVs
- and close projected margins of victory
- are vulnerable to electoral shenanignans
- so keep yours eyes open, how 'bout it?
Mathematically, it's sort-of basic:
- assume a flat 3% shift in your candidate's favor by .. whatever it is you do.. as the base unit of activity. Multiple shenanigans are possible.
- compare this to the projected victory spread for the other guy
- use a normal distribution to rate your chances of overcoming this projected spread
- multiply by the number of electoral votes for the state being targeted by your mischief
- rank the states using this score
- repeat for all the scenarios.
- share the Top 10 for each scenario with the other kids in the class.
And here we go
Cheating to defeat the First (Obama Tsunami) Scenario
Your Goal As Cheater: Swing 140 electoral votes from the Democrats to the Republicans
Your Challenge: You need to pick off a whopping 12 states that would otherwise go to Obama
- Georgia
- Alaska
- Louisiana
- North Carolina
- Missouri
- Arizona
- Florida
- Virginia
- Ohio
- Montana
- Colorado
- Virginia
That's in order of increasing difficulty.
Assessment: Ain't gonna happen.
Next to what I think is the likelihood scenario...so far.
Cheating to defeat the Second (Obama Landslide Plus) Scenario
(Not really that much more than the third scenario, the only difference being Montana.)
Your Goal As Cheater: Swing 103 electoral votes from the Democrats to the Republicans
Your Challenge: You need to pick off 8 states that would otherwise go to Obama
- North Carolina
- Missouri
- Montana
- Florida
- Ohio
- Virginia
- West Virginia
- Colorado
Assessment: Ain't gonna happen.
Cheating to defeat the Third (Obama Landslide) Scenario
(This is the current orthodox model forecast.)
Your Goal As Cheater: Swing 100 electoral votes from the Democrats to the Republicans
Your Challenge: You need to pick off 8 states that would otherwise go to Obama.
(They just happen to be the same ones, save that Montana is replaced by Nevada.)
- North Carolina
- Missouri
- Florida
- Ohio
- West Virginia
- Virginia
- Nevada
- Colorado
Assessment: Ain't gonna happen.
Cheating to defeat the Fourth (Obama Big Win) Scenario
Your Goal As Cheater: Swing 80 electoral votes from the Democrats to the Republicans
Your Challenge: You need to pick off 6 states that would otherwise go to Obama.
(They just happen to be the same ones, save that Montana is replaced by Nevada.)
- Missouri
- Florida
3.Ohio
- Virginia
- Nevada
- Colorado
Assessment: It'd take an awful lot of work and luck and nobody looking and very favorable courts and DA's.
Moving right along...
Cheating to defeat the Fifth (Obama Closer-than-we-preferred Win) Scenario
Your Goal As Cheater: Swing 44 electoral votes from the Democrats to the Republicans
Your Challenge: You need to pick off 3 states that would otherwise go to Obama.
(They just happen to be the same ones, save that Montana is replaced by Nevada.)
- Florida
- Virginia
- Washington
At this stage we should probably start listing alternate targets as well:
- Colorado
- Michigan
- Minnesota
- Maine
- Pennsylvania
- New Mexico
- New Jersey
Assessment: Sufficient to warrant dedicated resources in place ahead of this scenario which, given how close the race was during the summer, are available.
("Relax, people! I got this.")
Cheating to defeat the Sixth (Obama At-the-wire Win) Scenario
Your Goal As Cheater: All you need is 9.... EVs. 8 is not enough, as that will toss the election into the House of Representatives.
Your Challenge: You need to pick off exactly 1 state with 9 or more electoral votes.
In this final scenario, the focus will be on one of five states that (a) have sufficient EVs to matter and (b) are projected in this scenario to be rather close.
(Recall this is a full-blown breakdwon scenario for the Obama camp, the very least likely.)
- Virginia
- Minnesota
- Colorado
- Michigan
- New Jersey
Assessment: If we are back in a situation where Michigan and New Jersey are projected by anyone to be close races, we probably have bigger concerns. But that's just me thinking.
Wrap
For most of the scenarios, cheating on the part of the Democrats is utterly untempting and unrewarding; in all the scenarios Obama is projected to win if things go as expected...and the process isn't contaminated by electoral shenanigans. In all but the lower two scenarios, the level of exertion and risk of exposure to the Republicans required to thwart such large projected margins of victory in the popular vote would obviate any decision for a reasonable person.
It is only in the close contest, the final two scenarios, where there is credible opportunity for mischief, if its purpose is securing possession of the White House.
We've known the goal all along - don't even let it think of being a close race, not just for our sake, but for the Republicans... because in the long run they benefit from an electoral process that people trust, also.