A critical caveat: I reject the notion of the popular vote as a values-based basis for superdelegates selecting one candidate over another.
In the general election, every state but Maine and Nebraska run their electoral vote the same. But in the nomination process, the states are given the ground rules by the DNC, and then they decide whether to hold a caucus or a primary. Either method is a time-honored way of selecting delegates to the national convention. This isn't something radical and new.
However, if state parties believed that their voices would only be proportionally represented based on a popular vote number, which state would select caucuses as its method of voting? You can't be intellectually honest and argue that any state party would rationally choose the caucus method if it knew its input on the national level would be fractionally diluted.
Moreover, there is no uniform national rule for counting popular vote - closed vs. open primaries change the calculus. If Clinton argues that indies and Republicans voting in some states but not others pollutes the purity of the popular vote total, it simultaneously undermines her national electability argument, since indies and Republicans will vote in the general. But this inconsistency, along with the caucus argument, further shows how illegitimate the popular vote argument is.
That's essentially what the popular vote argument seeks to do - to punish states after the game has been played under rules everyone knew going in. It's an argument that requires a lack of integrity to make.
HOW TO PROJECT POPULAR VOTE
Now that the caveat is out of the way, I'm going to give you some parameters of what to expect, turnout-wise, in the upcoming primaries.
One of the helpful concepts to help predict popular vote is projecting by congressional district. Each congressional district represents an average of 697,000 people.
Some congressional districts have more Dem voters than others, and CDs are subject to restrictions like closed versus open primaries. These two factors will affect the popular vote turnout in a state.
It is difficult to use past cycle turnout to base a projection of primary turnout this year. Numbers are up all across the board. Interest is at an all-time high, for one thing. For another, comparing turnout for a hotly-advertised and contested election versus turnout for long-settled nominations in past cycles is difficult.
So, to get these projections, I used the data from other states to get an average turnout per CD in closed states, in open states, and semi-open states. Then I applied some rough guesstimation given the Dem-ness of a state. For example, Maryland (8CDs, closed) had 110,000 votes per CD while Arizona (8CDs, closed), had around 56,000.
The end prediction is not going to be exact as far as total turnout, but it will be a useful number. And once you have a useful number in total turnout, you can legitimately play with different win percentage scenarios to see what kind of numbers we are talking about for Hillary Clinton to make up 813,000 votes.
CLOSED PRIMARIES
There have been 6 closed primaries so far: Arizona, Connecticut, Delaware, Maryland, New York, Oklahoma.
State - #CDs - Avg pop/CD - Total 2008 turnout (avg turnout/CD)
AZ – 8 – 792K - 452K (56K per CD)
CT – 5 – 700K - 350K (70K per CD)
DE - 1 - 865K - 96K (96K per CD)
MD – 8 – 702K - 878K (110K per CD)
NY – 29 – 665K - 1.839M (63K per CD)
OK – 5 – 723K - 417K (83K per CD)
Total - 56 - avg 72,000 per CD
The next set of numbers works like this. To get a sense that will help us predict the turnout in upcoming closed primary states, it's useful to get a feel for how different states have varied against the average.
So the first number is how a given state's population per House district compares with the national average of 697,000. The closer the first number is to 1.00, the closer that state's average to 697,000 population per CD.
The second number is how that state's actual turnout compared with the across-group average of 72,000 votes per closed primary CD, with 1.00 indicating exactly 72,000.
The third and final number is the second divided by the first, to give us an overall intensity factor.
AZ - 1.14 - 0.78 - 0.69
CT - 1.00 - 0.97 - 0.97
DE - 1.24 - 1.33 - 1.08
MD - 1.01 - 1.53 - 1.51
NY - 0.95 - 0.88 - 0.92
OK - 1.04 - 1.15 - 1.11
Let's use Arizona as an example to make the concept more concrete. The national average population per House district is about 697,000. Arizona's average population per CD is slightly higher than the national average at 792,000 - that's the 1.14 number you see.
The next number is, knowing that the average voter turnout among these 6 closed primary states was roughly 72,000 per CD, what was that state's comparison average turnout per CD? Arizona's was around 56,000. That's slightly under the norm - or 0.78.
The final number is the second calculation divided by the first calculation. I call this the intensity factor. If Arizona, with its slightly higher population per CD, had turned out a slightly higher number than 72,000 per CD in its actual vote, its intensity factor would have been roughly 1.00. But while Arizona has more than its average share of population per CD, it had a lower turnout compared to the average turnout among closed primary states. That gives it the low intensity factor of 0.69.
Maryland had an almost perfectly average population per CD 702K compared to 697K, or 1.01. But its Dem turnout was way above the 67K average, at 110K. This gives it a high intensity factor of 1.51.
Delaware has a higher population per CD than average, because the whole state is 1 CD and Delaware has about 865K. 865K/697K multiplied by 72K would project about 89K as an expected turnout. But 96K turned out, so the intensity factor is above average, at 1.08.
So now that we have this sense, let's look at the upcoming closed primaries: Pennsylvania, Kentucky, Oregon and South Dakota.
State - #CDs - average per CD - state avg/697K
Pennsylvania - 19 - 654K per CD - 0.94
Kentucky - 6 - 707K per CD - 1.01
Oregon - 5 - 750K per CD - 1.08
South Dakota - 1 - 796K per CD - 1.14
Now, if we multiply 72,000 by the number of CDs by the state's variation from the average, we can come up with a starting number.
PA - 1,286,000
KY - 436,000
OR - 389,000
SD - 82,000
Now, we have to adjust these numbers slightly, based on a projected intensity factor. First, we cannot just declare that all these states will be above average, because the 72,000 number factored into the calculations is a number in this highly energized campaign context. At the same time, compared with the other closed primaries, which all came on Super Tuesday with the exception of Maryland, there will be a media and focused campaigning factor in each state that is higher than the February 5 states received.
Here are the states projected turnout with a 1.10 factor over the 72,000 per CD average:
PA - 1,415,000
KY - 480,000
OR - 428,000
SD - 90,000
Here's 1.20:
PA - 1,543,000
KY - 523,000
OR - 467,000
SD - 98,000
Here's 1.30:
PA - 1,672,000
KY - 567,000
OR - 506,000
SD - 107,000
Here's 1.40:
PA - 1,800,000
KY - 610,000
OR - 544,000
SD - 115,000
And here are my guesses. I think Oregon will be the highest turnout of the bunch. It's the most reliably culturally Democratic state within the group, and its vote-by-mail system means all registered Dems automatically get a ballot, adding to the ease of voting. I think turnout of 544,000 in Oregon feels about right.
Kentucky and South Dakota probably won't be lower than 1.00 due to the focus, but these are not traditional culturally Democratic states, meaning the overall number of registered Dems is going to be lower. Given the higher individual focus and field resources per voter that can be spent in these states, I will simply predict the 1.00 model - 436,000 in Kentucky and 82,000 in South Dakota.
Pennsylvania now has around 4 million registered Dems, and so an intensity factor of 1.20 feels about right, because that projects to a high primary turnout of 1,543,000 (about 38%).
PocketNines' Actual Turnout Predictions:
PA - 1,543,000 (1.20)
KY - 436,000 (1.00)
OR - 544,000 (1.40)
SD - 82,000 (1.00)
Hat tip to Stephen Colbert and "the gut."
SEMI-OPEN PRIMARIES
There are two semi-open primaries left, West Virginia and North Carolina. Dems and independents can vote in these states. Let's go through the same exercise as above for closed primaries.
This one is slightly trickier, because a site might label a state "semi-open" when it is essentially open. I think of California as a good example of a semi-open state, because if you are an unaffiliated voter you don't need to do any overt advance work to switch your registration in order to vote, but registered Republicans would need to switch. A state like Ohio is often listed as "semi-open" but I am calling it "open" since you could simply change your registration at the polling place. Illinois is also listed as semi-open, but the barrier is the issue. Signing one extra form while you are already at the polling place is not comparable to West Virginia and North Carolina, which prevent Republican cross-overs on election day.
Previously held semi-open primaries on the Democratic side: New Hampshire, California, Massachusetts, New Jersey, Utah.
State - #CDs - Avg pop per CD - 2008 turnout (avg per CD)
NH - 2 - 658K - 284K (142K per CD)
CA - 53 - 690K - 5.067M (96K per CD)
MA - 10 - 645K - 1.254M (125K per CD)
NJ - 13 - 668K - 1.108M (85K per CD)
UT - 3 - 882K - 428K (143K per CD)
Total - 81 - 101,000 per CD
Running the same process as I used above:
NH - 0.94 - 1.41 - 1.50
CA - 0.99 - 0.95 - 0.96
MA - 0.93 - 1.24 - 1.33
NJ - 0.96 - 0.84 - 0.88
UT - 1.27 - 1.42 - 1.12
Now, for the upcoming states:
State - #CDs - average per CD - state avg/697K
North Carolina - 13 CDs - 697K per CD - 1.00
West Virginia - 3 CDs - 604K per CD - 0.87
Using 101,000 as the average, we start with baselines of:
NC - 1,313,000
WV - 264,000
Now we have to list a range of turnouts based on intensity. I will not list ones below 1.00, because while New Jersey and California (barely) were below 1.00 in the intensity column, they weren't by very much and there will be more intense focus in these states than there could be in big Feb 5 states.
Here's 1.10:
NC - 1,444,000
WV - 290,000
Here's 1.20:
NC - 1,576,000
WV - 316,000
Here's 1.30:
NC - 1,707,000
WV - 343,000
Here's 1.40:
NC - 1,838,000
WV - 369,000
For North Carolina, I'm going to just say 1.20, like Pennsylvania. That projects to 1,576,000.
I think West Virginia is a heavy Clinton state, and I think independents favor Obama, so I see turnout tending to be closer to 1.00 than 1.40 in intensity. I am going with 1.14 as a prediction. Again, a gut feel. 301,000 in West Virginia.
One of the biggest revelations in the popular vote game that should come out of this diary is that while Pennsylvania is 6 CDs bigger than North Carolina, turnout in those two states is going to be approximately equal, since North Carolina's baseline starts out a little bit higher. That's very bad news for the Clinton popular vote racket.
PocketNines' Actual Turnout Predictions:
NC - 1,576,000 (1.20)
WV - 301,000 (1.14)
OPEN PRIMARIES
Montana and Indiana are the remaining open primaries.
Open primaries, that have already gone are: South Carolina, Alabama, Arkansas, Georgia, Illinois, Missouri, Tennessee, Louisiana, Virginia, Wisconsin, Vermont, Rhode Island, Ohio, Texas and Mississippi.
State - #CDs - Avg pop/CD - Total 2008 turnout (avg turnout/CD)
SC - 6 - 735K - 532K (89K per CD)
AL - 7 - 661K - 543K (78K per CD)
AR - 4 - 709K - 311K (78K per CD)
GA - 13 - 734K - 1.061M (82K per CD)
IL - 19 - 676K - 2.037M (107K per CD)
MO - 9 - 653K - 822K (91K per CD)
TN - 9 - 684K - 623K (69K per CD)
LA - 7 - 613K - 384K (55K per CD)
VA - 11 - 701K - 986K (90K per CD)
McCain says "I'll be the nominee," "eloquent but empty" night
WI - 8 - 700K - 1.114M (139K per CD)
VT - 1 - 621K - 154K (154K per CD)
RI - 2 - 529K - 186K (93K per CD)
OH - 18 - 637K - 2.233M (124K per CD)
TX - 32 - 747K - 2.878M (90K per CD)
MS - 4 - 730K - 421K (105K per CD)
Total - 150 - 95,000 average
Total after McCain had it locked - 55 - 127,000 average
I'm going to use the second number because that's much closer to the reality we have today.
WI - 1.00 - 1.09 - 1.09
VT - 0.89 - 1.21 - 1.36
RI - 0.76 - 0.73 - 0.96
OH - 0.91 - 0.98 - 1.07
TX - 1.07 - 0.71 - 0.66
MS - 1.05 - 0.83 - 0.79
For Indiana and Montana, we have:
State - #CDs - average per CD - state avg/697K
Indiana - 9 CDs - 705K - 1.01
Montana - 1 CD - 957K - 1.37
Intensities of 1.00 would project:
IN - 1,154,000
MT - 174,000
Right now, it's tough to say whether intensity will be lower or higher than 1.00. Indiana is a red state, and Montana is probably red vis-a-vis two Dems who aren't in the Schweitzer/Tester farmer/rancher mold. I personally know how much Clinton is hated in Montana, but also how much Republicans have malice in that state, and would be willing to meddle with strategic Clinton crossover voting.
I see Indiana having more Ohio-like turnout, at an intensity of around 1.07. That would be 1,223,000 votes. Montana feels closer to a 0.94, or around 164,000 votes. Those are gut predictions.
PocketNines' Actual Turnout Predictions:
Indiana - 1,223,000 (1.07)
Montana - 164,000 (0.94)
PUERTO RICO
Puerto Rico doesn't lend itself as well to this kind of prediction. It was a caucus, and now seems to be an open primary, but it doesn't have congressional districts that fit this analysis. Population is 3,994,000. The primary also seems to have more limited hours, from 8am to 3pm. So until we get more details, it's unclear what turnout to predict in Puerto Rico, although somewhere between 500K and 1 million seems like a decent guess.
OVERALL POCKETNINES' Turnout Predictions:
Pennsylvania - 1,543,000
Indiana - 1,223,000
North Carolina - 1,576,000
West Virginia - 301,000
Kentucky - 436,000
Oregon - 544,000
South Dakota - 82,000
Montana - 164,000
Puerto Rico - unknown (roughly the size of KY and OR, but those are closed while PR appears open, yet may be no voting past 3pm)
So, that was an exhaustive breakdown, but we haven't started playing with win margins. So let's do that now.
Pennsylvania win margins:
1% - 15,430
10% - 154,000
12.5% - 193,000
15% - 231,000
20% - 308,000
Indiana win margins:
1% - 12,230
3% - 37,000
5% - 61,000
8% - 98,000
10% - 122,000
North Carolina win margins:
1% - 15,760
5% - 79,000
8% - 126,000
10% - 158,000
12% - 189,000
15% - 237,000
West Virginia win margins:
1% - 3,010
10% - 30,000
15% - 45,000
20% - 60,000
25% - 75,000
30% - 90,000
Kentucky win margins:
1% - 4,360
5% - 22,000
10% - 44,000
15% - 67,000
20% - 89,000
Oregon win margins:
1% - 5,440
5% - 27,000
8% - 44,000
10% - 54,000
12% - 65,000
15% - 82,000
South Dakota win margins:
1% - 820
10% - 8,000
15% - 12,000
20% - 16,000
Montana win margins:
1% - 1,640
5% - 8,000
10% - 16,000
15% - 25,000
20% - 33,000
Now you have all the tools to mix and match your predictions. I'll make a few, but you can use all the above work to do a plug and play. Want to assume a higher intensity factor for Pennsylvania (bigger turnout) and then assume a 20% margin for Clinton, you can do it using this method.
Here are some simulations.
Simulation 1:
PA - Clinton by 15% - 231,000
IN - Obama by 1% - 12,000
NC - Obama by 12% - 189,000
WV - Clinton by 21% - 63,000
KY - Clinton by 13% - 57,000
OR - Obama by 12% - 65,000
SD - Obama by 18% - 15,000
MT - Obama by 11% - 18,000
Clinton gains - 351,000
Obama gains - 299,000
Net - Clinton +52,000
Needed to close the 813K gap: 761,000 (Puerto Rico, shady Florida)
Simulation 2, a worst-case scenario for Obama:
PA - Clinton by 18% - 278,000
IN - Clinton by 7% - 86,000
NC - Obama by 5% - 79,000
WV - Clinton by 25% - 75,000
KY - Clinton by 18% - 78,000
OR - Obama by 8% - 44,000
SD - Obama by 10% - 8,000
MT - Obama by 5% - 8,000
Clinton gains - 517,000
Obama gains - 139,000
Net - Clinton +378,000
Needed to close the 813K gap: 435,000 (Puerto Rico, shady Florida)
As you can see, anemic Obama wins in 4 states in which he's heavily favored, and robust wins for Clinton in her 3 favored states and a solid win in Indiana still only get Clinton 378,000 votes. Even with the dishonest 295,000 number from Florida, that's only 673,000. It would take 139,000 votes from Puerto Rico (plus 1K from Guam) to give us a Florida-tainted tie. And that's her argument to superdelegates (none of whom would accept including Michigan in that argument).
I put the chances of this tiny. Miniscule. Obama is 21 points up in a new North Carolina poll today, which makes at least a 10% win margin feel pretty comfortable. North Carolina is far less vulnerable to Limbaugh-style meddling, since it's semi-open and ideal for Obama. Obama has also won crushing double digit victories in every state in the VA-to-LA southern arc.
Simulation 3:
PA - Clinton by 12% - 185,000
IN - Clinton by 2% - 24,000
NC - Obama by 15% - 237,000
WV - Clinton by 18% - 54,000
KY - Clinton by 15% - 67,000
OR - Obama by 11% - 59,000
MT - Obama by 10% - 16,000
SD - Obama by 17% - 14,000
Clinton gains - 330,000
Obama gains - 336,000
Net - Obama +6,000
Needed to close the 813K gap: 819,000 (Puerto Rico and shady Florida)
Executive Summary:
The popular vote argument is a bogus one on the merits, but as a thought experiment, the numbers show that type of primary (closed, open, semi-open) will dramatically affect the turnout in the upcoming states. For example, most people don't realize that North Carolina (a 13 CD state) will probably have bigger turnout than Pennsylvania (a 19 CD state). Also, about 2/3 of the remaining popular vote resides in PA, IN and NC combined.
By applying all the math we can see the ranges of expected turnout in each upcoming state. Making defensible projections of Dem primary turnout, we can then make projections in the overall popular vote change if we plug and play different win percentages in the individual states.
After completing this exercise, we see how huge the win margins have to be in Clinton's base states while at the same time how relatively small the win margins have to be in the Obama base states, just to get us to an optimistic projection of a tie in the popular vote.
Anyway, play with the numbers yourself. You can go back and change the intensity numbers to get different turnout, but I think it'll be very clear very quickly that even if PA is a 1.40 intensity rather than the 1.20 I project, you get 257,000 more overall votes, and 12% of that is only about another 31,000 - just not a very big number when that's your #1 state to make up the popular vote margin. And you can legitimately plug in a 1.40 intensity, but not a 1.60 (we've seen that in no state). So you also have a sense of the boundaries of this.
Hopefully this diary pushes the discussion a little further down the road so that more people see in stark reality just how monumentally unlikely it is that Clinton can catch up in popular vote.