Wednesday, May 21, 2008

Rating Obama’s Jedi Master

Eons ago, back in February, Bloomberg News obtained a detailed spreadsheet that was inadvertently released by the Obama campaign. The spreadsheet was the topic of this diary at the time. The Bloomberg reporters posted it at this location. The prognosticator was Jeffrey Berman, the “Obama campaign’s unsung hero" and Jedi master.

In it, the campaign analyzed about twelve scenarios for all the state-by-state races after the February 5 Super Tuesday, including projected splits in the popular vote and the delegate totals. Only one scenario is in this spreadsheet, but you should take a look at it. It is fascinating to see what the Obama campaign predicted at the time and where they saw the race going.

Now that Obama has clinched the lead in pledged delegates and nearly clinched the nomination, let’s check out exactly how good Jeffrey Berman's wizardry actually was.

It is remarkable how accurate the predictions were, and it’s interesting to see where the results deviated significantly. I put together the following chart to illustrate the difference between the forecast on February 6 and the actual results of the primaries and caucuses.

The chart shows the differences in both popular vote percentage and number of delegates earned. The delegate numbers show the results for caucus states through the local and state conventions that have been held so far. For example, on May 16, Nevada provided an extra delegate for Obama at Clinton’s expense. The actual results are based on The Green Papers tallies.

The Green Papers and this chart do not include the one extra delegate that the Obama campaign claims for each Louisiana and Texas (although the extra Obama caucus delegate(s) in Texas may yet appear on June 7). (The popular vote shown for Texas in the chart is for the primary.)

The Obama results also exclude the Edwards delegates that have now endorsed Obama, as well as the two Clinton pledged delegates from DC and MD that switched to Obama.

State W-L
On February 6, the forecast for the remaining 27 contests was for 19 wins and 8 losses. So far, Obama has 17 wins and 7 losses, matching the predicted number of wins and losses to date. MT, SD and PR will go 2 and 1, for a final record of (drum roll, please) 19 wins and 8 losses. The Obama campaign did not project to win Maine, but he won it anyway. On the other hand, Indiana went from his winner column over to a narrow loss to Clinton.

Popular Vote Variance
Obama had double-digit over-performances in the popular vote percentage in VA, DC, HI, ME, and VI, all winning contests for him. In fact, the day of the Potomac Primary was exceptional for exceeding the forecast not only in VA and DC, but also in MD (by 7%). Most notable was the winning vote margin in Virginia. It had been forecast to be tight at 1.2%, but it ballooned to a 14% margin. In Hawaii, total turnout for the caucuses was a magnitude higher than it had ever been, and the result was a 24% blowout for favorite son Obama.

Obama under-performed in the popular vote in eight states, although by less than 1.7% each in four of them (OH, RI, MS and PA). He lost OH, RI and PA, and even though he garnered a slightly smaller percentage of votes in Mississippi than had been predicted five weeks earlier, his winning margin over Clinton there was nearly 25%.

In Guam, Indiana, West Virginia and Kentucky he received, respectively, 4%, 3%, 17% and 12% less of the popular vote than had been forecast by the campaign. Nevertheless, he won Guam (by 7 votes). In Indiana a switch of 7210 Limbaugh-influenced Clinton votes would have given him the state. As for WV and KY, what else is there to say? As DHinMI has often noted, "Appalachia has an Obama problem."

Delegate Count Variance
Leaving aside a half-delegate under-perform in the Democrats Abroad primary (where Obama crushed Clinton anyway), Obama under-performed the campaign forecast in Ohio, Pennsylvania, Indiana, West Virginia and Kentucky, all of which turned out to be very tough for him to win. Of the 22 fewer delegates garnered in those states, 19 were lost in Indiana, West Virginia and Kentucky. Excluding Indiana, the campaign on February 6 had not even expected to win the other four states, each of which later exhibited varying degrees of the Appalachia syndrome.

Obama more than made up for it in Maryland, Virginia, Texas, and North Carolina, where he gained 28 more delegates than had been forecast. And for a one-day over-performance, it’s hard to top the day of the Potomac Primary, when VA, DC and MD produced 19 more delegates than the campaign had expected (exactly offsetting the IN, WV and KY losses).

Of the 24 contests held since February 6, the Obama delegate numbers for nine of them were within one delegate of what had been forecast (including five that were spot on). Another seven were within three delegates.

Overall so far, Obama has over-performed by 28.5 delegates compared to the February 6 prediction.

If it were I that had received that forecast on February 6, and then witnessed the kind of results that we see here, I’d say Jeffrey Berman deserved a big, fat raise — and then some time off on August 29.

P.S.: Here's a snapshot of part of the famous spreadsheet:

Jeffrey Berman Spreadsheet

As many of us remember, Ben Smith at Politico offered this comment on the campaign spreadsheet at the time it was “released” in February:
The scenario considered in the Bloomberg story foresees the widely anticipated Obama roll through the Potomac Primary, and Clinton victories in Texas, Ohio, and Pennsylvania. Crucially, it doesn't forecast any blowouts with huge delegate margins; Washington State's caucus, by this projection, offers Obama his widest margin, 20 delegates.

The scenario "projects Obama will end up in June with 1,806 of the delegates who select the party's nominee to 1,789 for New York Sen. Hillary Clinton." That's a count that gives her a 50-delegate lead among superdelegates, and him a 67-delegate lead among pledged delegates.

Isn’t it nice to be able to breathe easier now, compared to then, with Obama’s relatively large lead?

No comments: