After an enjoyable (though exhausting) experience tracking the Texas county- and district-level convention results in a long diary last month, I thought it would be fun, and hopefully valuable to interested spectators, to post a diary tracking the results of the April 22 Pennsylvania primary as they come in--with special attention to the delegate count.
So if you're interested in knowing how the Pennsylvania pledged-delegate race is looking at the moment, this is the place to be. Now that we're into Wednesday, updates will be a little more sporadic, but I'm going to keep going until it becomes clear that USA Today has lost interest in updating its district vote totals.
Regarding comments: Previously I directed comments about the results to the front-page open threads. Those seem to have stopped, though, so now you can go ahead and say whatever you'd like here. Remarks directed at my silly tables particularly (or just designed to stroke my ego--heartily welcomed!) have of course always been welcome.
I. Latest Numbers
Update 33, 2:30 P.M. EDT Wednesday, April 30: Given that USA Today's results numbers haven't moved in more than five days, I'm guessing that they are no longer updating their site. So I believe that this will be my last update.
Another signal that USA Today has called it quits is that, for the first time (that I've noticed), the Pennsylvania Secretary of State's results page now shows more votes counted in the statewide race than USA Today does. Here are those totals:
Clinton: 1,260,889 (54.64%)
Obama: 1,046,774 (45.36%)
Total gap: 214,115 (9.27%--clearly there are some round-off issues going on with that subtraction)
In CD 7, with 6% of the precincts left to report (which means somewhere around 6,700 votes), Obama remains 487 votes behind Clinton. That means he needs to win about 53.7% of the remaining votes to pick up +2 net delegates by flipping the 4-3 delegate split into his favor.
In CD 11, the opportunity for +2 net is Clinton's: with 40% of the precincts (~50,700 votes) still out, Clinton is currently 4,183 votes beneath the threshold (70.00%) that would turn her 3-2 win into a 4-1 win. That means she needs to win approximately 72.5% of the remaining votes to pick up the extra two net delegates.
My guess is that neither of these districts will flip, but either one still could--which means that, with Pennsylvania almost finished counting, we're looking at a potential range of net pledged delegates that goes from Clinton +8 to Clinton +12.
Dist | PDels | %In | O% | C% | Dels (O-C) | O INF+2N | C INF+2N |
At large | 35 | 100% | 45.36% | 54.64% | 16-19 | X | X |
PLEO | 20 | " | " | " | 9-11 | X | X |
1 | 7 | 99% | 66.88% | 33.12% | 5-2 | X | X |
2 | 9 | 100% | 77.48% | 22.52% | 7-2 | X | X |
3 | 5 | 100% | 35.02% | 64.98% | 2-3 | X | X |
4 | 5 | 100% | 35.02% | 64.98% | 2-3 | X | X |
5 | 4 | 90% | 42.44% | 57.56% | 2-2 | X | X |
6 | 6 | 99% | 50.04% | 49.96% | 3-3 | X | X |
7 | 7 | 94% | 49.77% | 50.23% | 3-4 | 3.90% | X |
8 | 7 | 100% | 37.05% | 62.95% | 3-4 | X | X |
9 | 3 | 89% | 34.05% | 65.95% | 1-2 | X | X |
10 | 4 | 93% | 34.12% | 65.88% | 1-3 | 48.27% | X |
11 | 5 | 60% | 31.65% | 68.35% | 2-3 | 45.88% | 4.13% |
12 | 5 | 96% | 27.26% | 72.74% | 1-4 | 68.44% | X |
13 | 7 | 100% | 38.47% | 61.53% | 3-4 | X | X |
14 | 7 | 100% | 52.21% | 47.79% | 4-3 | X | X |
15 | 5 | 100% | 39.36% | 60.64% | 2-3 | X | X |
16 | 4 | 99% | 54.21% | 45.79% | 2-2 | X | X |
17 | 4 | 100% | 45.67% | 54.33% | 2-2 | X | X |
18 | 5 | 100% | 34.33% | 65.67% | 2-3 | X | X |
19 | 4 | 100% | 45.28% | 54.72% | 2-2 | X | X |
Totals | 158 | | | | 74-84 | | |
Swing | | | | | C+10 | | |
Clinton's 214,115-vote win in the statewide vote is equivalent to 29.86% of the popular-vote lead (717,086) that Barack Obama, according to RealClearPolitics, carried into the Pennsylvania primary. Note that this number does not include any votes from Florida or Michigan (because the DNC considers those states' primaries to have been illegitimate) or from Iowa, Nevada, Maine, or Washington (because, to RCP's reckoning, popular-vote totals have not been reported for caucuses in those states). I believe the number disregards the Texas caucuses as well.
Please note that I strongly endorse PocketNines’ article here urging Obama supporters to fervently question the relevance of the whole “popular vote” thing to the nomination race. As a resident of (and recent Obama caucus precinct captain in) Minnesota, I’d like to smack anyone who seriously contends that we’re 1/4 as important as Missouri, or any other state of comparable size.
II. A Little Background Context
According to the delegate counter on barackobama.com, before Pennsylvanians started voting, Obama was ahead of Clinton 1,420 to 1,249 in pledged delegates, for a lead of 171.
The “magic number" for pledged delegates in the Democratic race (presuming the delegates from Michigan and Florida remain unseated because of those states' DNC rules violations) is 1,627: as soon as either candidate hits 1,627 pledged delegates, he or she is mathematically assured of having the majority of the pledged delegates headed to the national convention in Denver.
Including the Pennsylvania delegates that were earned on April 22--and the 18 more from Iowa, New Hampshire, and South Carolina that John Edwards can still lay claim to--there are 584 pledged delegates, or just under 18% of the national total, outstanding. On that premise, Obama needs 207 more pledged delegates (35.45% of what's left nationwide pre-Pennsylvania) in order to clinch the national pledged-delegate win. Clinton, in contrast, needs 378 more pledged delegates (64.73%) to clinch. This is not unrelated to the claims voiced by more than a few people that the nomination race is over.
There are 158 pledged delegates at stake today in Pennsylvania (plus three add-on superdelegates, who will presumably go 2-1 or 3-0 to today's pledged-delegate winner). 103 of the 158 pledged delegates will be allocated according to the individual voting results in Pennsylvania's nineteen Congressional Districts. The other 55 will be portioned out on the basis of the statewide vote, in two separate pools: 35 "at-large" delegates and 20 PLEOs ("Party Leaders and Elected Officials").
It's worth noting that all of the vote percentages in this entire diary are calculated by including only votes for Clinton and Obama. Pennsylvanians are of course casting votes for other presidential candidates in the Democratic primary today, and those votes may have some effect on the Obama and Clinton percentages reported by CNN and other media outlets. Nonetheless, for the purposes of allocating delegates, the only number that matters is each candidate's share of the total votes cast for viable candidates in a given district. As long as (say) Joe Biden's vote total is under 15% in any particular district, votes for him are disregarded when we calculate the delegate hauls for the viable candidates.
III. The Prognosticators
I've found five pundits on the Web (three of whom post here on DKos) who have posted district-by-district projections for the delegate totals from today's primary. One of the five projections--the one posted last week on CQPolitics.com--made no clear prediction, oddly, about the result of the statewide race.
Put together, though, the five sets of predictions show a significant number of similarities, and they make it somewhat clear which districts we can expect to be the "close calls" tonight. I've listed all five sets (along with a sixth column that carries the average of the five in each district) on the table below.
I'm hoping these projections will present a expected "par" of sorts, to determine whether either candidate is doing considerably better than educated observers had thought (s)he would. (The delegate splits below are all stated as [Obama delegates]-[Clinton delegates] because, well, I'm biased.)
Dist | PDels | PsiFighter37 | Elec Insp | Poblano | Al Giordano | CQ Politics | Consensus |
At large | 35 | 16-19 | 15-20 | 16-19 | 17-18 | N/A | 16-19 |
PLEO | 20 | 9-11 | 9-11 | 9-11 | 10-10 | N/A | 9-11 |
1 | 7 | 4-3 | 5-2 | 4-3 | 4-3 | 4-3 | 4-3 |
2 | 9 | 7-2 | 7-2 | 6-3 | 7-2 | 7-2 | 7-2 |
3 | 5 | 2-3 | 2-3 | 2-3 | 2-3 | 2-3 | 2-3 |
4 | 5 | 2-3 | 2-3 | 2-3 | 1-4 | 2-3 | 2-3 |
5 | 4 | 2-2 | 1-3 | 2-2 | 2-2 | 2-2 | 2-2 |
6 | 6 | 3-3 | 3-3 | 3-3 | 3-3 | 3-3 | 3-3 |
7 | 7 | 4-3 | 4-3 | 3-4 | 4-3 | 3-4 | 4-3 |
8 | 7 | 4-3 | 4-3 | 3-4 | 4-3 | 3-4 | 4-3 |
9 | 3 | 1-2 | 1-2 | 1-2 | 1-2 | 1-2 | 1-2 |
10 | 4 | 1-3 | 1-3 | 2-2 | 2-2 | 2-2 | 2-2 |
11 | 5 | 1-4 | 1-4 | 2-3 | 2-3 | 2-3 | 2-3 |
12 | 5 | 1-4 | 1-4 | 2-3 | 1-4 | 2-3 | 1-4 |
13 | 7 | 3-4 | 4-3 | 3-4 | 4-3 | 3-4 | 3-4 |
14 | 7 | 3-4 | 3-4 | 3-4 | 3-4 | 4-3 | 3-4 |
15 | 5 | 2-3 | 2-3 | 2-3 | 2-3 | 2-3 | 2-3 |
16 | 4 | 2-2 | 2-2 | 2-2 | 2-2 | 2-2 | 2-2 |
17 | 4 | 2-2 | 2-2 | 2-2 | 2-2 | 2-2 | 2-2 |
18 | 5 | 2-3 | 2-3 | 2-3 | 2-3 | 2-3 | 2-3 |
19 | 4 | 2-2 | 2-2 | 2-2 | 2-2 | 2-2 | 2-2 |
Totals | 158 | 73-85 | 73-85 | 73-85 | 77-81 | 50-53 | 75-83 |
Swing | | C+12 | C+12 | C+12 | C+4 | C+3 | C+8 |
DemConWatch also has an interesting page containing projections of a sort regarding pledged delegates in Pennsylvania, but it leaves considerably more "unprojected" than the above five prognosticators do.
IV. Delegate Split Thresholds
On primary nights like this one, it's frequently important to know what percentage of the vote a candidate needs in a given district to gain additional delegates. The percentages needed are a matter of simple arithmetic, and they can be reflected in a more or less set-in-stone matrix, as shown in the next two tables.
To read the two tables, first find the column bearing the number of delegates in the district that you're curious about. Then find the row with the number of delegates you'd like to find the threshold for. The cell at the intersection of that row and column contains the percentage of the district vote a candidate needs to exceed in order to earn the number of delegates at the left end of the row.
| 9-del dist | 7 | 6 | 5 | 4 | 3 |
Win 1 | 15.00% | 15.00% | 15.00% | 15.00% | 15.00% | 16.67% |
2 | 16.67% | 21.43% | 25.00% | 30.00% | 37.50% | 50.00% |
3 | 27.78% | 35.71% | 41.67% | 50.00% | 62.50% | 83.33% |
4 | 38.89% | 50.00% | 58.33% | 70.00% | 85.00% |
5 | 50.00% | 64.29% | 75.00% | 85.00% |
6 | 61.11% | 78.57% | 85.00% |
7 | 72.22% | 85.00% |
8 | 83.33% |
9 | 85.00% |
For example, the italicized cell above shows that, in a 6-delegate district (such as Pennsylvania's Congressional District 6), a candidate must get more than 58.33% of the viable-candidate votes in that district in order to receive four, rather than three, delegates from that district. (In a two-person race such as this one, rising above the 58.33% threshold means moving from a 3-3 delegate tie to a 4-2 delegate win. Note that that's a difference of +2 net delegates, not just one.)
I didn't put in a column for an 8-delegate district because Pennsylvania doesn't have any 8-delegate districts.
In addition to the 103 delegates allocated to the specific Congressional Districts, Pennsylvania will be awarding 55 delegates according to the results of the statewide primary today. The delegate thresholds for a 55-delegate race wouldn't fit on the table above (and the calculation is slightly more complicated, too, because the 55 delegates are split into separate pools of 20 and 35), so below is the table for the statewide delegate race. I'm presuming that neither candidate is going to win by much more than 25 percentage points:
| PA statewide (55 dels) | Net dels |
Win 21 | 37.50% | -13 |
22 | 38.57% | -11 |
23 | 41.43% | -9 |
24 | 42.50% | -7 |
25 | 44.29% | -5 |
26 | 47.14% | -3 |
27 | 47.50% | -1 |
28 | 50.00% | +1 |
29 | 52.50% | +3 |
30 | 52.86% | +5 |
31 | 55.71% | +7 |
32 | 57.50% | +9 |
33 | 58.57% | +11 |
34 | 61.43% | +13 |
35 | 62.50% | +15 |
V. What the @#$&% Does "INF+2N" Mean?
Even when they show projected delegate splits, results tables like the top one above typically don't give the reader much idea of how close the candidates are to changing the current state of the delegate race. I concocted "INF+2N" (Improvement Needed For +2 Net pledged delegates) as a number that reflects how difficult it will be, as returns continue to come in, for a candidate to improve his/her delegate haul. A high INF+2N number (say, more than 10%) means the candidate is not very likely to do any better in the delegate race in that district than (s)he's currently projected, barring the appearance of some precincts that are overwhelmingly in his/her favor.
The number isn't hard to calculate (well, if you've got a spreadsheet application, anyway). Given the current state of the race--the candidates' respective raw vote, and the percentage of precincts that have reported--the computer calculates what fraction of the remaining uncounted votes the candidate needs in order to improve his/her delegate haul by the smallest possible number--which, in a two-candidate race, is +2 net delegates. Subtract from that fraction the proportion of the district's vote the candidate has gained so far, and you've got his or her INF+2N.
If that explanation is too confusing (and it probably is), here's an example, in a hypothetical district race between Candidate A and Candidate B:
Dist | PDels | %In | A% | B% | Dels (A-B) | A INF+2N | B INF+2N |
27 | 9 | 75% | 60.00% | 40.00% | 5-4 | 4.45% | 40.00% |
Here, with 75% of the precincts reporting in a nine-delegate district, Candidate A leads Candidate B by exactly 60%-40%. As you can see from the first delegate threshold table above, that translates to a 5-4 delegate split in A's favor. If you're wondering how good (or bad) those last 25% of precincts will have to be in order for A to get a 6-3 split, or for B to flip the 5-4 into his favor, INF+2N provides that answer.
Again from the delegate threshold table, A needs to exceed 61.11% of the total district vote, once it's all counted, in order to get a 6-3 split. B needs to push his share over 50% to win the district 5 delegates to 4. But with three-quarters of the precincts already counted,* either candidate needs to do considerably better than that in the remaining votes to shift his/her overall proportion over the relevant threshold.
As a spreadsheet can show you reasonably simply, A needs to win 64.45% of the uncounted quarter of the votes in order to push her fraction of the total district vote over the 61.11% threshold. That's difficult, because it's 4.45 percentage points better than A has performed over the first three-quarters of the vote counted. And thus A's INF+2N number is currently 4.45%.
B has it considerably tougher, though: in order to push his proportion from 40% to 50% (plus one vote), he needs to win a smidgen over 80% of the votes that currently remain uncounted. Those 80 minus B's current 40% of the vote = 40 percentage points, so B's INF+2N number is 40.00%.
* Keen observers will note that I'm using "precincts reported" as a direct proxy for "votes reported," which is... demonstrably false: some precincts, clearly, have a substantially larger number of voters than others. But I don't see any way around this problem; no one knows exactly how many votes will be counted tonight until they're counted. Obviously this means that there is a fair amount of imprecision in the INF+2N number.