Defending My Projection: Clinton Can Win the Popular Vote

My post projecting the possible outcome of the remaining Democratic contests has provoked a lot of commentary, most of it—pro and con—quite thoughtful. As I explained, I was making projections that were optimistic from Hillary Clinton's point of view, and I found, to my surprise, that the results showed Clinton ahead of Obama in popular votes (whether Florida, Michigan, and imputed caucus results for four states were included or not) but Obama still ahead in delegate votes.

Some commenters have argued that I was too optimistic (again, from the Clinton point of view). I'm not convinced. Remember that I was projecting the percentage of the two-candidate vote for each candidate. I projected a 60 percent to 40 percent win for Clinton in Pennsylvania. Current RealClearPolitics.com average poll numbers show her leading the two-candidate vote 57 percent to 43 percent. I projected a 55 percent to 45 percent win for Obama in North Carolina. Current RealClearPolitics.com average poll numbers show him leading the two-candidate vote 57 percent to 43 percent. Current polls may be off, or those currently undecided could swing heavily to one candidate, or turnout may be heavier among Obama than among Clinton voters. But I think my estimates are defensible as outcomes that are possible.

I've also been criticized for projecting that Clinton would carry Indiana and that she would carry Kentucky and West Virginia by wide margins. As I explained, I was influenced in all three cases by the wide margins—75 percent to 80 percent of the two-candidate vote—Clinton won in many southern Ohio counties. These counties, I think, look and feel a lot like southern Indiana and most of Kentucky and West Virginia. My projections that Clinton would get 60 percent in Indiana, 65 percent in Kentucky, and 70 percent in West Virginia may be high. And don't be distracted by talk that most of Indiana gets Chicago media, which has been full of positive news about Obama for nearly four years. Only the northwest corner of the state does—about two congressional districts out of nine. Hoosiers elsewhere in the state don't take their lead from Chicago. And some of the Chicago news about Obama has been negative—the Tribune and Sun-Times have covered Tony Rezko much more than other media.

In Oregon, I gave Obama 60 percent of the two-candidate vote, even though he got only 53 percent of the two-candidate vote in the nonbinding primary in neighboring (and fairly similar) Washington on February 19.

To put my projections in perspective, I went back to RealClearPolitics.com's summary of the popular vote in previous primaries, and calculated Clinton's percentage of the two-candidate Clinton-and-Obama vote. The two-candidate vote is relevant, since no other candidate will get over the 15 percent threshold that Democrats tend to require for a candidate to win delegates. Although Clinton hasn't gotten more than 60 percent of the total vote in any state but Arkansas, where she lived and worked for nearly 20 years, she gets 60 percent or near to it of the two-candidate vote in many. To wit,

Arkansas              73
Oklahoma             64
Florida                   60
New York               59
Rhode Island        59
Massachusetts     58
Michigan                58
Tennessee            57
New Jersey           55
Ohio                        55

(In Michigan, I have shown the Clinton percentage of the Clinton-Uncommitted vote.) Is it unrealistic to project that Clinton would get about the same percentage of the two-candidate vote in Kentucky as she did in Oklahoma? Or that she would get about the same percentage of the two-candidate vote in Indiana as she did in Tennessee? Or that she would get about the same percentage of the two-candidate vote in Pennsylvania as she did in New York, Rhode Island, or Massachusetts? Or that she would get about the same percentage of the two-candidate vote in West Virginia as she did in Arkansas? I don't think so. Sure, she wasn't the first lady of West Virginia for 12 years. But the Mountain State seems to have an aversion to Obama. SurveyUSA's 50-state polls, released March 6, showed Clinton beating John McCain in West Virginia 47 percent to 42 percent, while Obama loses the state to McCain by 53 percent to 35 percent. Some 57 percent of West Virginia voters are registered Democrats, eligible to vote in the primary, but only 35 percent are prepared to vote for Obama against McCain.

As for North Carolina, consider Clinton's percentage of the two-candidate vote in the states where she was weakest.

Delaware             44
Alabama              43
Utah                      41
Wisconsin            41
Vermont               39
Louisiana             38
Maryland              37
Mississippi           37
Virginia                 36
Illinois                   34
South Carolina    32
Georgia                 32
D.C.                        23

Why did I not project North Carolina (21 percent black) to vote like Virginia (19 percent black)? Because North Carolina whites are different from Virginia whites. According to the 2007 Census Bureau population estimates, the Northern Virginia counties and independent cities in the Washington metropolitan area have 2,454,486 people, 32 percent of the state's population. The North Carolina counties in the Raleigh and Durham metropolitan area have 1,527,253 people, 17 percent of the state's population. These metro areas are where you find the bulk of the liberal, upscale white voters, and in Virginia they form twice as large a share of the population as in North Carolina. The coastal regions of east Carolina have nothing like the Hampton Roads metro area; Democratic voters here tend to be conservative and tradition-minded, and there are still plenty of registered Democrats in those parts. There's also a hot Democratic primary race for governor that will be drawing them to the polls. The upshot is that I expect North Carolina's results to look more like Alabama's and less like Virginia's.

About Puerto Rico, I confessed I was simply guessing. Commenter Liahona makes some good and some not-so-good points:

To be valid, Hillary would have to win Puerto Rico by a huge margin—almost as much as she wins Pennsylvania by. That won't happen. I think it is probably based on the ffact [sic] that she has done well with Hispanics in Texas and California. Well, your reasoning is flawed. All Hispanics are not alike. I used to live in Puerto Rico. Puerto Ricans are a Caribbean people (Antillanos). Most Puerto Ricans either look like Barack Obama or have a relative who does. Obama has the support of the governor (who's indicted but still in office). The Obama team is also well-organized. He is likely to do extremely well, if not win. So, since your argument is based on a false assumption, it is wrong.

She (or he) may well be right about Obama being well organized, but remember that this is a primary, not a caucus, and turnout in Puerto Rican elections typically is higher than anywhere on the mainland. So a campaign's ability to get supporters to the polls may turn out to be not very important. I agree heartily that all Hispanics are not alike. But very few look like Obama; few have ancestors from either Kenya or Kansas. And I don't think they're likely to be swayed, as another commenter says, by Bill Richardson, who grew up in Mexico City and now lives in Santa Fe, N.M.—both a long way from the Antilles (and with quite differently accented Spanish). But many Puerto Ricans do have close ties to New York, Clinton's home state now, and some may have close ties to recent migrants from the island to Florida, where Clinton won 60 percent of the two-candidate vote. In Osceola County (Kissimmee, Disney World), which has had a lot of Puerto Rican migrants, she won 72 percent of the two-candidate vote. If Puerto Ricans in Puerto Rico vote like Puerto Ricans in New York and (probably) Florida voted, they'll give Clinton a solid majority of the two-candidate vote.

I projected South Dakota and Montana to vote for Clinton. Many commenters noted that neighboring states voted heavily for Obama. Yes, but those were caucuses, not primaries. I think almost everyone expects that the South Dakota and Montana primaries will produce lower Obama percentages than the Idaho, Wyoming, North Dakota, or Nebraska caucuses, in which Obama got 82 percent, 62 percent, 63 percent and 68 percent of the two-candidate vote, respectively. How much lower no one knows. Anyway, if you give Obama rather than Clinton 60 percent of the two-candidate vote and keep the rest of my projections the same, Clinton still ends up with more popular votes than Obama.

Am I guaranteeing that Clinton will win the popular vote? Certainly not. I maintain that these projections are optimistic from Clinton's point of view but not wildly unrealistic. I agree with almost every analyst that Clinton can't win a plurality of delegates selected in primaries and caucuses. But I do think it's possible that she can win more popular votes than Obama.

Note: I've left Guam, which elects four delegates, out of my calculations. Uh-oh.