How do we know this? To begin with, the core Democratic blocks in big cities didn’t change their votes according to the National Exit Polls. The NEP showed that black, Latino, and Jewish city voters went for Kerry at rates of 95%, 66%, and 80%.
The National Exit Poll gathers data each year on various campaign activities. The vital activity accompanying any significant vote increase is called GOTV - get out the vote. According to the NEP, only 1 in 10 urban residents contacted received Republican GOTV calls. Media buys in big cities were not even close to those in the suburbs where Bush only netted an extra 3% vote share over 2000, and campaign events for these urban groups were virtually non existent.
Scrutiny of the National Exit Poll was not the focus on election night. Hurried analysts looked at the vote totals coming in and offered explanations that we now know were not even remotely accurate. While we were told that this was a red-versus-blue election on November 7, it was also noted that Kerry’s GOTV strategy was working, based on increased votes/turnout. Senior analyst Charles Cook assumed that Bush made major inroads in black and Latino city voters but that assumption was not supported by the splits.
Then where did that Bush Urban Wave Originate?
There was a minimal Republican GOTV campaign in big cities and, in general, a minimal presence in the form of advertising and special events. Thus the basis for converting any shift in sympathy to Bush was lacking. Only the National Exit Poll, the revised edition the day after the election, had the special lens necessary to note surges of white big city voters who comprised the Bush victory margin. White voters had to account for the margin and the NEP analysts already knew that. There was no apparent general cause for a shift in loyalties and voting in big cities. The black, Latino, and Jewish voting blocks there had remained essentially unchanged since 2000.
For large urban areas, Latino votes doubled and went from 14% to 16% of the total vote compared to 2000. A small decline in absolute numbers, plus the increased Latino vote pushed the black urban share from 29% to 19% yet the national exit poll showed overall black turnout up 40%.
So the question remains: how do we account for the election winning Bush increases in big city vote share? We know that black, Latino, and Jewish voters in the big cities were strongly in favor of Kerry. The votes came from the only remaining big city voting block. There had to have been an unprecedented out pouring of white voters in large urban areas.
FIGURES 6 & 7. According to the NEP, white voters contributed less than 5 million votes to the big city segment in 2000 but almost 9 million in 2004. This is worthy of the term “surge.” They accounted almost exclusively for the increase from 2.3 million to 5.9 million big city votes for Bush from 2000 to 2004. Where did they come from? We may never know but they “won” the election.
The Nature of the Bush White Urban Wave
The white urban wave was shy, reluctant to show its true form and was perhaps a ghost in the machine. White turnout was supposed to go up as part of an overall increase. No one anticipated that it would materialize for Bush in the big cities in the way that it did. Nor does it help that these white voters were apparently “shy” about talking to pollsters.
Figure 8. The Black and Latino vote in big cities didn’t shift much from 2004. Bush lost big again. But he did much better among big city whites than his 2000 performance.
Where then did the Bush swing in the urban wave come from? The simple answer is that it was weighted into existence. The act of reconciling the exit polls to the official vote count created it. The Bush urban voters came into existence because they had to… otherwise the official vote count would be wrong.
Weighting is a practice used by the US Census, political consultants, public health officials and others who conduct large scale survey research. If you collect data on a population, Latino voting patterns in the 2004 election for example, and your data is unrepresentative of a subset of that population, you can weight certain responses by a multiplier greater or less than one to make your poll consistent with the population measured. The problem though is when weighting is used to reconcile polling data with a “known fact” that may not be known at all. The NEP assumes that the official vote total must be accurate and weights accordingly.
'
The impossible election of 2004
Generating the Bush urban wave was effortless. Only 10% of urban voters required a call. They were not required to attend rallies or watch television ads. In fact, many of them didn’t even need to vote. That was taken care of by the weighting process conducted when the national exit poll was found to be inconsistent with the announced vote tallies. After all, how could the unintentionally released Election Day NEP be right in showing a 3% Kerry overall victory margin when the vote tabulators showed a 3% Bush win? Rural Americans didn’t produce that margin. Neither did the small towns or the suburbs. Even the improvement in the smaller cities wasn’t enough. The big cities, according to announced totals, delivered the vote for Bush.
