METHOD (i.e. If you just want the results, skip this part):
I decided to base this projection on racial demographics, because I figured that those demographics would vary the most from district to district in Ohio, and because it is the demographic that has the best available data. In order to estimate the racial breakdowns by district, I had to make a few preliminary calculations. First, I used census data to estimate the demographics in each district. The sUSA poll only included "White," and "Black", racial categories, so I combined the "White," "Asian," and "Hispanic" groups from the census into one group that would use the "White" cross-tab in the sUSA poll. I did this because all three of these groups have strongly favored Clinton in past races. In order to translate the population numbers from the census into an estimate for the racial breakdown on voting day, I calculated a turnout factor for the two broad groups used by sUSA ("White" and "Black"), based that group's state-wide percentage of the population and its state-wide percentage of the expected voters in the sUSA cross-tabs. This normalizes the census data to this year's primary, which has a high turnout history, an African-American candidate, and a contentious election. Finally, I used the cross-tabs and expected turnout for the two broad groupings to predict the votes for the two candidates. These votes didn't add up to 100%, since Kucinich, Edwards, Gravel, and "Other" are still getting a few % of the vote, so I extrapolated things out to 100% before assigning delegates. I report the extrapolated numbers below.
RESULTS:
District 01:
This district, which includes Cincinnati, has a high African-American population, and has a high percentage of college graduates. However, its a 4 delegate district, so despite the 54-46 lead I'm projecting for Obama, its unlikely he'll add delegates in this district. He needs to obtain 62.5% of the vote to earn a 3-1 split, which he could do so if he continues to gain support as election day nears, but it would require gaining ~9 points in the polls before election day. Its possible, but I don't see it happening.
Prediction: 2-2 split.
Running delegate total: Clinton 2, Obama 2
District 02:
This district includes the suburbs of Cincinnati and borders Kentucky. Its less than 5% African-American, but it has quite a few residents with college degrees. The race-based data suggests a 57-43 victory for Clinton. Its another 4 delegate district, and I expect another split here. Its going to be nearly impossible for Clinton to reach 62.5% of the vote, given the large amount of college graduates in this district, Obama's momentum, and the distance she'd have to cover to get there.
Prediction: 2-2 split.
Running delegate total: Clinton 4, Obama 4
District 03:
The Ohio 3rd represents Dayton and a good deal of the land between Columbus and Cincinnati. It is our first district with an odd number of delegates, with 5 of them. I'm projecting 23% of the Democratic primary voters will be African-American, and that Obama will narrowly lose this district 51-49. That would also give Clinton a 3-2 delegate victory here. But this prediction is tenuous, given the small predicted margin of victory, the large number of highly-educated voters (who generally prefer Obama), and Obama's momentum. I think Obama will win, but for the sake of consistency, I'll stick with the what the demographic numbers are suggesting.
Prediction: 3-2 Clinton victory in a "swing district"
Running delegate total: Clinton 7, Obama 6
District 04:
This district covers the ground between Dayton, Toledo, and Columbus, and is demographically similar to district 2. Over 90% of its population is white, and Clinton should win another 4 delegate district. The main difference in this district is that it has about half as many college graduates, meaning Clinton's chances of gaining the 3-1 split are a little bit better here.
Prediction: 2-2 split
Running delegate total: Clinton 9, Obama 8
District 05:
District 5 includes land South of the cities on Lake Erie This in the NW part of Ohio, and includes a lot of border territory with Michigan and Indiana. The 5th is the district with the highest White and lowest African-American populations. It has three times as many Hispanics as African-Americans, and should be Clinton's biggest popular vote victory in the state, 59-41. Unfortunately for her campaign, it will probably represent another "moral victory" that will not bring with it any advantage in delegates. This is her best shot at a 3-1 split, as my projection has 62.5% just within the margin of error for the poll, but I don't think she'll get there.
Prediction: 2-2 split, with "swing" potential for a 3-1 Clinton win
Running delegate total: Clinton 11, Obama 10
District 06:
District 6, which stretches from the outskirts of Youngstown and follows the SE border of Ohio all the way down to the Ohio-Kentucky-W. Virginia tri-state area. District 6 is similar to district 3 in that it has 5 delegates, and should lean heavily towards Clinton. I'm predicting a 56-41 victory for Clinton, which is in safe territory for now but could swing to Obama if his momentum continues.
Prediction: 3-2 Clinton win
Running delegate total: Clinton 14, Obama 12
District 07:
The Ohio 7th includes the outskirts of the Columbus and Dayton areas, and a lot of rural land surrounding both cities. This is another safe 56-44 "win" for Clinton in another 4 delegate district, that carries another minuscule chance of e 3-1 split.
Prediction: 2-2 split
Running delegate total: Clinton 16, Obama 14
District 08:
Hey, look! This district, which wraps around Dayton and strethes the Iniana-Ohio border, is predominately White! And... look at this! Its got 4 delegates! Clinton should win big here! And earn... wait for it.
Prediction: 2-2 split
Running delegate total: Clinton 18, Obama 16
District 09:
The Ohio 9th, home of Tommy Callahan, is an industrial district that includes Toledo, Sandusky, and... Port Clinton. The district has a fairly high amount of African-Americans (13.7%), and a fair share of college graduates (19.8%). However, I'm guessing the economy will be a primary concern in this area, and the economy is Clinton's 2nd best issue for Ohio voters (next to battling terrorists). The numbers suggest a narrow, 52-48 win, within the margins of error for both an Obama victory in the popular vote and a 6-4 delegate split in Clinton's favor. If Obama can keep hitting Clinton hard on NAFTA, he could gain ground here, but all he really has to do is play defense and prevent Clinton from getting the 58% she'd need for the 6-4 split. So, yeah... this should be another meaningless Clinton win.
Prediction: 3-3 split
Running delegate total: Clinton 21, Obama 19
District 10:
This district sits in the Western and Southern parts of the greater Cleveland area, and has 6 delegates. Its similar to district 9, but has far fewer African-Americans a less industrial economy, and more college graduates. This represents another good opportunity for the Clinton camp to pad their delegate lead, but the numbers and my gut both say she'll fall just short.
Prediction: 3-3 split with a 4-2 Clinton win within error margins
Running delegate total: Clinton 24, Obama 22
District 11:
Hoo, boy. Did I get this one wrong last time, when I predicted a narrow Clinton victory in the popular vote here. The majority of people in this district are African-Americans, and as a result Obama should easily win this district. In fact, the method I'm using in this projection have Obama down for a 64-36 win, good enough for a comfortable 5-3 delegate victory. Its also noteworthy that Obama is closer to a 6-2 win (the magic number is 68.75%) than Clinton is to a 4-4 split.
Prediction: 5-3 Obama victory
Running delegate total: Clinton 27, Obama 27
District 12:
The Ohio 12th includes the Eastern half of Columbus and the areas North and East of city. Over 20% of this 5-delegate district is African-American, and about a third of the population has a college degree. To put things in perspective, the population of those with advanced degrees in this district (10.3%) is higher than the state-wide percentage of those with a college degree (9.7%). Oh, and did I mention that tOSU sits next door in district 15? This is a young, well-educated, racially diverse area. I have Obama down for a narrow, 51-49 victory here, but think the acual margin will be bigger than this.
Prediction: 3-2 Obama victory
Running delegate total: Clinton 29, Obama 30
District 13:
This district includes Akron and runs to the NW, along the areas to the SW of Cleveland. It is a fairly well-educated, fairly industrialized, and fairly racially diverse area of the state. I think this state will have a tight race for the popular vote victory, and have Clinton winning 53-47. I don't see either Clinton or Obama earning a 4-2 split unless things change a lot between now and March 4th.
Prediction: 3-3 split
Running delegate total: Clinton 32, Obama 33
District 14:
This district runs form East of Cleveland to the Ohio-Pennsylvania border. The district is 94.8% white, and has as many combined Hispanics and Asians as it has African-Americans. It is well-educated, but also has an industrial economy. The racial demographics point towards a 58-42 Clinton win. This district has 6 delegates, and Clinton is right on the verge of getting a 4-2 split.
Prediction: 3-3 split, with an excellent chance of a 4-2 Clinton win
Running delegate total: Clinton 35, Obama 36
District 15:
The Ohio 15th covers the other half of Columbus, including tOSU, and spreads out to cover quite a bit of rural land to the East and Northeast. This is the most educated district in the state, and is probably to be the youngest. However, these projections are based primarily on racial demographics, which predict a 56-44 Clinton victory. This is also supported by the 62-33 Clinton advantage in the sUSA cross-tab for the Columbus region. I expect the actual margin of victory to be much smaller, given the presence of the University I think Obama will pull out the win. Its a 4 delegate district, so regardless of who wins the district it should be split 2-2.
Prediction: 2-2 split, with an excellent chance of a 4-2 Clinton win
Running delegate total: Clinton 37, Obama 38
District 16:
This is rural land to the Southwest of Akron. This district is racially homogeneous and has an economy that relies heavily on industry. In short, this is Clinton country. She should carry this 5 delegate district, unless Obama makes a LOT of movement before the election. Right now, I have Clinton winning to the tune of 57-43.
Prediction: 3-2 Clinton win
Running delegate total: Clinton 40, Obama 40
District 17:
This is the district that I will be driving a van to this weekend, as myself and over 50 Penn State students will be heading to Youngstown (which is in this district) to canvas for Barack Obama. This district includes Akron, Youngstown, and the land between the two. It has 7 delegates (I know, its weird). The racial demographic method I'm using here projects for a 53-47 Clinton win, but her margin of victory is within the poll's margin of error. This is absolutely a swing district, and the fact that the Obama campaign decided to send a huge group of volunteers to this district points towards either good fortune or a lot of intelligence in how they're distributing their ground resources.
Prediction: 4-3 Clinton win in a swing district
Running delegate total: Clinton 44, Obama 43
District 18:
This district sits between Columbus and District 6. Its fairly similar to district 6, as it has a population that is 96.3% white, with only 11.3% of the population having a college degree, and a third of its population has jobs in construction, extraction, production, or transportation. All these things point to a big Clinton victory, and so do my projections. Based on the current data, I'm predicting a 59-41 win here, and there's a lot of breathing room on either side of that prediction before either candidate can change the delegate count.
Prediction: 3-2 Clinton win in a swing district
Running delegate total: Clinton 47, Obama 45
Well, there you have it. According to the racial demographics of the districts in the state and the racial cross-tabs of the latest sUSA poll, I'm predicting a 2 delegate win for Clinton in Ohio, despite her significant 6 point lead in that poll. The reason for the tight delegate race is that much of Clinton's lead is coming from districts that have an even number of delegates, and her lead is no longer large enough to earn 3-1 or 4-2 splits in those districts.
Its going to be very hard for the Clinton camp to add to the delegate lead I'm projecting here, but if they want to do so they should focus on districts 5, 10, 12, and 14. She's already winning districts 5, 10, and 14, but could earn 3-1 or 4-2 splits with a little upwards movement. District 12 is currently leaning Obama in these projections but could easily flip into the Clinton camp.
If the Obama camp wants to catch Clinton in Ohio, or overtake her there, they should focus on districts 3, 17, and 11. Clinton has very narrow leads in the first two of these districts, so by taking the lead Obama would get a 3-2 or a 4-3 split. In district 11, a good GOTV effort could land Obama with an impressive 6-2 delegate split.
All of the districts I mention in the last two paragraphs are within the margin of error of the poll. There are 7 districts I mention, 4 where Clinton could gain ground and 3 where Obama could gain ground. So if you want to get a range of possible outcomes for the Ohio primary, I'd put it here:
Clinton: 44-51
Obama: 41-48
It would take the best-case scenario for Clinton to win this state by a double-digit delegate margin. I think that's what she needs to do there to stay in the race, particularly if Obama carries more delegates in Texas, as matttx predicts.
