We've heard ad nauseum about the differences between Democratic delegate allocation and the Republican method, and between delegate allocation and electoral votes. Bill Clinton has suggested that his wife would have the delegate lead if Democrats chose delegates as Republicans do, and Evan Bayh attempted to use electoral votes as a comparative measure.
Suppose, though, that instead of changing the Democratic system to approximate the GOP or the general election, we changed the distribution of electoral votes in the general election to the approach used by the Democratic Party to allocate convention delegates. What would be the outcome of that change?
Follow me below the fold for a look-see...
I'm using the most recent presidential election as the substrate for this analysis. Taking the Bush-Kerry results as displayed by Wikipedia, I allocated each state's electoral votes under the general rules of the Democratic delegate selection plans for 2008:
- Only the votes for candidates receiving at least 15% of the vote are considered
- The allocation of those electoral votes is proportional
Consider, for example, Maryland. In 2004, Kerry's vote total was 1,334,493 and Bush's was 1,024,703; none of the minor candidates came anywhere close to viability, so we ignore their 25,000 or so votes. That 56.6% to 43.4% distribution allocates the states EV as 6 for Kerry, 4 for Bush. If Kerry had been held under 55.0%, the outcome would have been 5-5, and it would have taken 65.0% for him to reach a 7-3 split.
Applying similar logic to the entire nation gives the following results (state-specific margins of three or more are bolded):
State | EV | Bush | Kerry | Margin |
Alabama | 9 | 6 | 3 | 3 R |
Alaska | 3 | 2 | 1 | 1 R |
Arizona | 10 | 6 | 4 | 2 R |
Arkansas | 6 | 3 | 3 | 0 |
California | 55 | 25 | 30 | 5 D |
Colorado | 9 | 5 | 4 | 1 R |
Connecticut | 7 | 3 | 4 | 1 D |
Delaware | 3 | 1 | 2 | 1 D |
District of Columbia | 3 | 0 | 3 | 3 D |
Florida | 27 | 14 | 13 | 1 R |
Georgia | 15 | 9 | 6 | 3 R |
Hawaii | 4 | 2 | 2 | 0 |
Idaho | 4 | 3 | 1 | 2 R |
Illinois | 21 | 9 | 12 | 3 D |
Indiana | 11 | 7 | 4 | 3 R |
Iowa | 7 | 4 | 3 | 1 R |
Kansas | 6 | 4 | 2 | 2 R |
Kentucky | 8 | 5 | 3 | 2 R |
Louisiana | 9 | 5 | 4 | 1 R |
Maine | 4 | 2 | 2 | 0 |
Maryland | 10 | 4 | 6 | 2 D |
Massachusetts | 12 | 4 | 8 | 4 D |
Michigan | 17 | 8 | 9 | 1 D |
Minnesota | 10 | 5 | 5 | 0 |
Mississippi | 6 | 4 | 2 | 2 R |
Missouri | 11 | 6 | 5 | 1 R |
Montana | 3 | 2 | 1 | 1 R |
Nebraska | 5 | 3 | 2 | 1 R |
Nevada | 5 | 3 | 2 | 1 R |
New Hampshire | 4 | 2 | 2 | 0 |
New Jersey | 15 | 7 | 8 | 1 D |
New Mexico | 5 | 3 | 2 | 1 R |
New York | 31 | 13 | 18 | 5 D |
North Carolina | 15 | 8 | 7 | 1 R |
North Dakota | 3 | 2 | 1 | 1 R |
Ohio | 20 | 10 | 10 | 0 |
Oklahoma | 7 | 5 | 2 | 3 R |
Oregon | 7 | 3 | 4 | 1 D |
Pennsylvania | 21 | 10 | 11 | 1 D |
Rhode Island | 4 | 2 | 2 | 0 |
South Carolina | 8 | 5 | 3 | 2 R |
South Dakota | 3 | 2 | 1 | 1 R |
Tennessee | 11 | 6 | 5 | 1 R |
Texas | 34 | 21 | 13 | 8 R |
Utah | 5 | 4 | 1 | 3 R |
Vermont | 3 | 1 | 2 | 1 D |
Virginia | 13 | 7 | 6 | 1 R |
Washington | 11 | 5 | 6 | 1 D |
West Virginia | 5 | 3 | 2 | 1 R |
Wisconsin | 10 | 5 | 5 | 0 |
Wyoming | 3 | 2 | 1 | 1 R |
Total | 538 | 280 | 258 | 22 R |
In the real 2004 election, the Bush-Cheney ticket ended up with 286 electoral votes, so this approach would have narrowed, but not reversed, the outcome. More to the point, it would have vastly changed the pattern of states "in play". That's because it might have been possible to change the EV distribution within a state without flipping the plurality in the state. In fact, the patterns of "in play" states, and the strategies for working on them, would be different for the two parties.
In this thought-experiment, let's assume that the parties had polled each state well enough to have a pretty good idea of the eventual result. Let's further assume that a concerted campaign effort in a particular state could increase the number of voters for that party in that state, thereby changing the EV distribution in the state. I'll demonstrate what I'm talking about by going back to the Maryland example.
We've seen that the 56.6%-43.4% vote in the state resulted in a 6-4 Democratic lead. The question is: how many additional votes for a specified party would it take to change that EV allocation? By my calculations, had there been 67,155 additional Republican votes (and no added Democratic votes), the outcome in Maryland would have been 5-5. That 67,155 votes represents 2.85% of the actual 2,359,196 {Democratic + Republican} votes in the state in 2004. On the other hand, to move the distribution up to 7-3 would have required a whopping 568,527 more Democratic votes, fully 24.1% of the actual count. In other words, a moderate effort by the GOP could have helped their cause, whereas there was no way for the Democrats to improve their result (though of course they would have played some defense to assure that 6-4 split).
Carrying this analysis through all of the states and DC, there would have been 15 states where Democrats could have gained by adding fewer than 5% (on their side) to the actual vote totals. For the Republicans, the corresponding number of states was 12. But there were only two states where both parties could have changed the distribution by adding 5% to the total. Here are the two sets of "in play on offense" states:
Democratic -- AL, AZ, CA, CO, GA, IN, IA, MS, NV, NM, NY, OK, SC, TX, UT
Republican -- CA, FL, IL, MD, MA, MI, NC, OH, OR, PA, TN, TX
Not exactly the "usual suspect" swing states, is it? It's no surprise, I suppose, that the two largest states are the only ones in common. It takes a smaller percentage change to shift allocations when the number of units is larger. In many of the other states, one party or the other would be attempting to minimize a large deficit or extend a large advantage. In eight cases (AZ, CA, IA, NM, and NY for the Dems; IL, MA, and NC for the GOP), even a 1% increase in the statewide total would have changed the allocation. A mere 11,172 more Republicans would have changed MA to a 7-5 division, while the Democrats would have added an electoral vote (and taken one away from Dubya) by adding a mere 5989 votes in NM, 9990 in AZ, or 10,060 in IA. Actually, that's all it would have taken to flip New Mexico and Iowa in the real EV outcome.
The point here isn't to argue for states to allocate their electoral votes as the Democratic Party allocates its National Convention delegates. After all, had this approach been in use in the 2004 election, you can bet your bottom dollar that the campaigns would have molded their strategies with such considerations in mind. There would have been a GOP effort in California, a Democratic push in Texas. Democrats would have been trying to stay closer in the Solid South, Republicans looking to staunch the bleeding in the Northeast and the Mid-Atlantic.
So my point is that a major part of campaign success is to build your overall stratgy according to the rules of the game you're playing. It's foolish to argue that you'd be leading the race for the Democratic nomination if the Republican rules applied -- they don't. It's ridiculous to suggest that you're better because Democrats preferred you over another Democrat in states with large numbers of electoral votes -- that's irrelevant. It's stupid as hell to ignore, and then diss, states that choose their delegates with caucuses -- that's intentionally denying yourself delegates. It's absurd to use the "popular vote" (particularly a version thereof that includes some and excludes others) as an argument about your chances at the convention -- that ignores the process and is irrelevant.
Final note:
While I was at it, I calculated the 1992 presidential results using the 2008 Democratic delegate selection rules. It would have gone to the House of Representatives ... Clinton 245, Bush 204, Perot 89. Ol' Ross would have gotten electoral votes in all states except AL, AR, DC, GA, HI, KY, LA, MD, MS, NC, SC, TN, and VA.