One Voter, One Vote… Black or White

I have been struggling for a while with collecting all the data necessary for this blog, so this is me giving up and admitting I do not have it all. But what I do have is pretty darn persuasive, so here goes…

Anybody who has been following the primaries in the US must have noticed that Barack Obama has established a lead over Hilary Clinton in the race to be the Democrat Presidential candidate. Seeming concerns about the electability of a black man, or a woman, have been pushed to one side as many polls have concluded that race and gender is not that important to most electors. However, you should always be wary of polls. People have a nasty habit of making pollsters draw the wrong conclusions, especially when they are asked questions where there is an obvious “right” answer. Asking someone if race influences the way they vote is a pretty much asking them if they are a racist. That means that some people know to respond that race does not matter, even if it does to them. Before we go too far, let me state that I have no opinion on whether racism will influence the outcome of the elections in the US. What I am more interested in is that black voters may have obtained a disproportionate influence over the Democrat’s selection process, and that may lead to shockwaves on many fronts – racial, political and cultural – if their preferred choice, Obama, wins the candidacy but turns out to be a lot less popular with the wider electorate than is currently supposed.

After doing lots of research, I eventually stumbled across someone who had preempted my arguments. Earl Ofari Hutchinson nailed the main elements of the argument in this blog for The Huffington Post. In short, black voters have backed the Democrats over Republicans by a ratio of about 9 to 1 in every Presidential election in the last few decades. That level of support may have helped Bill Clinton win his campaigns, but was not enough to elect Al Gore or John Kerry. Black voters are hence not a key swing demographic if the Democrats are going to take back the White House. They cannot swing much more behind the Democrats than they already do. For the Democrats to win, they need more white and Hispanic votes; some estimates suggest George W. Bush carried almost 40% of the Hispanic vote against Kerry, and typically about a third of Hispanics have backed the Republican Presidential candidate. Those are the voters that the Democrats need to aim for.

Whilst the commentators have worked hard to present both Clinton and Obama as having their natural followings, with Clinton doing best with women, Hispanics and the over 60’s, there is no doubt about which subset of the populous is the most unified in support of one candidate. Blacks who have voted in the primaries have cast over 80% of their votes for Obama. If you take black votes out of the equation, Clinton is the clear leader in the contest. The solid and unwavering support for Obama from African-Americans has effectively reversed the position of the two main candidates, turning the underdog into the clear leader. That is interesting, but not the main point I want to make here. The really interesting question is whether the influence of the black vote is higher in the Democrat selection process than it would be in a national election.

Obama recently informed his backers that his goal in the upcoming Pennsylvania primary, set to be won by Clinton, is to keep the losing margin to less than 10%. Some of the polls put Clinton ahead in Pennsylvania by nearly 20%, though others are around the 10% mark. African Americans are forecast to vote for Obama by the same kind of 80 point margin in Pennsylvania as he has enjoyed elsewhere, and they are expected to account for 18% of the likely Democrat voters in that primary. Here comes the key point, though: blacks only account for 11.4% of people in Pennsylvania. Not only are blacks heavily backing Obama, they are also turning out for the Democratic primaries in disproportionate numbers. Compared to a national election, they are heavily overweight in the samples being used to decide the Democrat contender. A little maths should explain the importance this has to the results. If Clinton was to win the Pennsylvania primary on a 55:45 split of the vote, that would she would close the delegate gap by about 18. If the black vote is consistent with the poll projections, Clinton needs to win the non-black vote in the state by a margin of 65 to 35 in order to win by 10 points overall. However, if the black share of the vote was only 11.4%, i.e. in line with the actual racial split in the state overall, then all things being equal, Clinton would do much better. If Clinton won Pennsylvania non-blacks in the ratio of 65:35 and Obama won Pennsylvania blacks in the ratio of 90:10, then the total result would be a win for Clinton with 58.7% instead of just 55%. In other words, the over-representation of blacks in the primary relative to the state’s population would be worth nearly 4% to Obama, which is equivalent to about a 7-delegate swing. If we re-run the equations on the assumption of a 60:40 overall win for Clinton with the state as is, and with the black vote as predicted, then that would require Clinton to win the non-black electorate by 71 to 29, which in turn would make the over-representation of black voters worth over 5% to Obama.

Have black voters been over-represented in other Democratic primaries? That is where I have so far struggled to get comprehensive data. News stories from individual states do suggest that is what has happened. In South Carolina, for example, which was key to generating positive momentum for Obama, a reported 55% of voters were black. Clinton did better with black voters in this state, picking up 19% of black votes compared to Obama’s 78%, and the remaining 2% going to Edwards who was still in the race back then. However, 55% of the primary’s electorate is severely at odds with the actual racial composition of South Carolina. According to the most recent census data, over two-thirds of South Carolina is white. That means in South Carolina there was a similar numeric over-representation of blacks to that which is being predicted will occur in Pennsylvania, with blacks accounting for almost double the share of Democratic primary votes than would be proportionate for their numbers in the state.

I admit I have not done all the numbers for all the states, but I think it is reasonable from what I have seen to assume the skew calculated for Pennsylvania, which was worth 4 points in the context of a 55-45 win for Clinton, is roughly consistent nationwide. At the time of writing, CNN says that Obama has 1411 pledged delegates to Clinton’s 1242. If the results are rebased to eliminate the 4% skew due to black over-representation, as would inevitably be the case in a national election where blacks voters could not be over-represented to anything like the same degree, then we see Clinton gain, and Obama lose, 106 delegates. That would place Clinton as the delegate leader, with a 1348-1305 lead in pledged delegates, and a 1585-1512 lead overall. Rebasing the poll results by this level of skew would also be enough to convert Obama’s impending victory into a clear win for Clinton.

There is nothing to stop any part of the population being over-represented in the primaries. In practice, if Obama gets the delegate numbers he will have won the candidacy fair and square, even if he needed the high turnout of black voters to put himself over the top. However, it is relevant to question whether this process really guarantees selection of the strongest candidate to contest the national election. In the national election, each voter gets one vote, whether black or white. Blacks may feel very strongly about Obama, and hence be more motivated to vote in the primaries, but one strongly felt black vote counts no more highly than the vote of a white person who shows no interest in the Presidential election prior to the big day. The current head-to-head polls suggest both Clinton v. McCain and Obama v. McCain would be incredibly close. Like I said above, treat polls with caution. Democrats do not need to look that far back in history for a valuable lesson about how Americans actually vote. Nixon, a loser against J.F. Kennedy in the 1960 election, identified the “silent majority” of voters as more crucial to determining electoral outcomes than any minority, no matter how active they are. In the face of an ever-present counterculture and enormous resentment over Vietnam, Nixon won by a landslide in 1972. The silent majority Nixon identified was composed of the older generation, blue collar workers and many ordinary white citizens in the Midwest and South. That sounds a lot like the people who have been voting for Clinton, not voting for Obama, and who McCain will target as the constituency he needs to retain from Bush’s 2004 victory in order to secure another Republican win. If the silent majority ultimately decides the 2008 Presidential election with a victory for McCain over Obama, the fallout for race politics in the US could be very severe.


  1. Half thoughts? More like fully-formed think-tank briefing reports! As always, I tremble in awe of the denseness of your analysis…so rather than try and be equal in reply, I’ll leave a couple of half-thoughts:

    1) Weakly-tied Republican/non-affiliated voters

    Are these the same ‘silent majority’? Only tenuous anecdotal evidence (e.g. ) suggests they may prefer Obama to Clinton – and if this is the case, could this demographic compensate for the disproportionate effect you talk about?

    2) Republicans seem to fear an Obama/McCain election?

    And that’s led to a ‘campaign’, by none other than Rush Limbaugh, for Republicans to vote for Clinton in open primaries (cf. and from the man himself –,2933,334669,00.html ).

    Limbaugh explicitly states that he wants to ‘keep the Dems at war’, which is a rather neat tactic – why not extend the time period where Democrats spend their own supporter’s money to explain why each Deomocratic candidate is not suitable for presidential office…

  2. Hey Guy, I cannot argue with your logic, but the problem here is data, or the lack of it. This is about forecasting votes, a notoriously difficult thing to do at the best of times. Not only have I not got enough data to reach a conclusion on Republican attitudes to the Democrat nomination race, but once you entertain the possibility of bluffs and double-bluffs it is impossible to draw conclusions about why people do what they do. Republicans may be voting for Obama in the primaries because they like him, and Republicans may be voting for Clinton in the primaries because they dislike her and think she would lose to McCain. The problem is that logic could just as easily be applied the opposite way around! Is Limbaugh telling people to vote Clinton because he thinks she would be easier to beat, or is he doing it because he thinks she would be harder to beat and wants to encourage Democrats to rally behind Obama? You also seemed to have missed a rather unpalatable corollary of your argument – you imply that Limbaugh is an astute reader and player of American political games, and not just a buffoon that says and does things to attract attention to himself…

    Anyhow, the best data we have is provided by the national polls and not inferences from other kinds of contests. At present they suggest Clinton is marginally more popular Obama, both as a straight choice between the two and when each is compared to McCain. The problem for the Democrats is that the polls suggest McCain may be more popular than either Democrat candidate. Whatever happens, it looks set to be a tight race.

Leave a Reply

Your email address will not be published.