Data: 1999 Household income versus voting preference for election 2000.

Sources: exit polls conducted by ABC News: http://www.abcnews.go.com/sections/politics/2000vote/general/exitpoll_hub.html.

Census bureau data provided by the Census Bureau, specifically "Money Income in the United States, 1999."

According to exit polls conducted by ABC News, it is clear that the higher your income, the more likely you were to vote for Bush. This was true for every income stratum in a stepwise way, as is illustrated by the following table.

Total Family Income, 1999:

• voters with family incomes greater than $50,000 preferred Bush to Gore;
• voters with family incomes less than $50,000 preferred Gore;
• the margin of support for Bush versus Gore (Gore Minus Bush) increased in a stepwise way as income increased;
• surprisingly, Nader and Buchanan had more support among lower income voters than among higher income voters.

Total Family Income, 1999:

I then pulled data from the Census bureau website to determine what proportion of the United States population lives within these income strata:

Total Family Income, 1999:

It seems clear that the election was decided by 2 factors: 1.) the disproportionately high turnout of voters in the middle class; and 2.) the dispoportionately low turnout of voters in the lowest income stratum:

• those in the $50-75,000 family income stratum made up only 21.6% of the population, they made up 25% of all voters, meaning they showed up 1.16x as much as you would expect given their distribution in the population;
• This group preferred Bush over Gore by 5 percentage points;
• those in the $30-50,000 range showed up 1.06x as much as expected.
• This group preferred Gore over Bush but only by 1 percentage point.

The worst turnout was in the stratum that preferred Gore the most, those earning less than $15,000:

• although they made up 9.6% of the population, this group made up only 7% of voters;
• they preferred Gore to Bush by 20 percentage points, or a 3:2 ratio.

Over half of Americans (50.8%) fell in income strata (> $50,000) that preferred Bush over Gore. (However, the magnitude of preference was higher for Gore versus Bush in the lowest income stratum (20% net) than it was for Bush over Gore in the highest income stratum (11% net)). In other words, although one could make a case that the Republicans on balance were the party of the wealthy (although 43% of the top money-makers preferred Gore), one could make a stronger case that the Democrats are the party of the poor (although over a third of the lowest stratum voted for Bush). It wasn't so much that the middle class and wealthy liked Bush as that the poor liked Gore.

It is interesting to compute the election results if everyone voted proportion to their distribution in the population:

Of course, this assumes that those who did not vote would have voted in the same proportion as the members of their income stratum who did vote. Given the low aggregate voter turnout, it is not at all certain that those who voted are representative of the general population.

Note that it still would have been a close race, with Gore beating Bush 48.5%-47.9%, or by .6%. (The actual Bush victory was 48.3%-48.2%, or a .1% difference.)

Another interesting question to answer is to determine how each stratum contributed to the final outcome of the election. To calculate this, I took the proportion of each stratum that voted, multiplied this times the difference within that stratum between Bush and Gore, and came up with the actual contribution to the net .1% net margin of victory.

Total Family Income, 1999:

This is a weighted average and reflects the true net effect voters in each income group had on the outcome of the election. The value of each cell is a function both of the strength of the preference (the percentage difference between how much voters preferred one candidate over the other) and the number of voters having that preference. A strong preference with poor turnout, for example, might have less net impact than a stratum with a weak preference but good turnout.

• The under-$15,000 group, although they preferred him by a wider margin, constituted less than 10% of the population and were less likely to vote; their net contribution to his victory was therefore less than that of the next higher stratum;
• Twice as many Americans were in the next higher (15-30,000) income stratum, and more of them voted. Gore's biggest net support came from this group, who contributed a net 2.1% of votes to his campaign.
• The largest net contributions to Bush came from the 16% of the American population earning over $100,000 earners (+1.7% net), followed closely by the 50-75,000 earners (+1.3% net). Had the top earners turned out in proportion to their numbers in the population, then this would have added another .12% to Bush's victory.

If voting preference is a function of income, then Gore in the end was hurt by the strong economy. 67% more Americans fell in the $100,000 and up category than in the $15,000 and below (16.1% versus 9.6%).

Other Factors

Gender

 Gender:

• Men preferred Bush to Gore by 11% net.
• Women preferred Gore by 11% net.
• Women outnumbered men among voters.

Age

Age:

• Voters under 60 were pretty evenly split;
• Those over 60 or under 30 preferred Gore; voters in the middle preferred Bush by a narrow margin.
• The margin was greatest among the elderly, with a net 5% preferring Gore; perhaps this can be explained by fears over social security and Medicare.
• Nader attracted a significant minority of voters under 30 (5% of this group voted for him); it is unclear how these voters would have voted if forced to choose between Gore and Bush, but a preference for Gore is not unimaginable.

Education

• Those at either education extreme preferred Gore to Bush;
• The net preference for Bush was highest among college graduates (6% net) or some college (6% net).
• Gore's strongest net support came from those with the least education; high school dropouts showed a 20% net preference for Gore. 59% of those who failed to complete high school preferred Gore.
• However, Gore also found support, albeit not as strong, among those who had received postgraduate education (8% net). 52% of this group voted for him.
• Bush's greatest absolute support was among those who attended (51%) or graduated (51%) college; his lowest support was among high school dropouts (39%).

A strong correlation was found among education and propensity to vote:

• although they made up 16% of the population, high school dropouts made up only 5% of total voters.
• At the opposite extreme, almost twice as many voters had postgraduate degrees as were in the general population (1.9x).

Perhaps the reason this was such a close race was that although few high school dropouts voted, they showed a strong preference for Gore, while on the other extreme, the overeducated who make up less of the population made up 18% of voters.

The breakdown by education may explain why higher income families preferred Bush to Gore, since education is strongly and positively correlated with income. Of all Americans over 25:

Source: www.census.gov, Money Income in the United States, 1999, extracted from Table 2, "Selected Characteristics – Households By Total Money Income, 1999", Page 6.

• included in Postgraduate study – pulled out only for illustrative purposes.

The link between education and income is powerful and stepwise.

• Over 84% of Americans have at least a high school education.
• The 16% who dropped out of high school have the lowest median income, earning less than 20% of what those with professional degrees earn;
• One fourth of the American population (26.8%) graduated college or higher; this group earns at least three times as much as high school dropouts.
• Almost one out of every 10 Americans has completed some postgraduate study; this group enjoys a median $82,298 income, or almost double the median family income of $42,038.

Race

Race:

* caveat: small number; large error possible.

• Blacks preferred Gore by a 10:1 ratio but were less likely to vote than whites (although they were more likely to vote than Hispanics or Asians).
• All racial groups preferred Gore to Bush except for Whites, but Whites made up 75.6% of the population (according to the Census Bureau) and 81% of the voters (according to ABC exit polls). Whites were slightly overrepresented among voters, meaning the average White citizen was more likely to vote than the average member of any other ethnic group. The net preference for Bush among Whites (8% net) was much lower than the net preference for Gore among Blacks (81% net).
• Asians also strongly preferred Gore (by 14% net) but had the worst turnout, relative to their proportion of the population, than any ethnic group (2% turnout v. 3.2%). Caveat: given the small number of Asians in the population and the fact that ABC did not provide more than one significant digit, the possibility of an error is large).
• Interestingly, although Asians had the highest median household income ($51,205, 20.4% higher than the median White household, and 83.5% higher than the median Black household), they still preferred Gore Americans who earned over $50,000 per year preferred Bush.