Interactive: Season the melting pot to your political taste

By Chris Wilson

Last January, at a conference of Hispanic conservatives in Florida, the former Republican Party chairman Ed Gillespie laid out the salient challenge facing his party in the clearest possible terms.

"In 2020," Gillespie said, "if the Republican nominee for president gets the same percentage of the white, African-American, Hispanic and Asian-American vote that John McCain got in 2008, a Democrat will be elected to the White House by 14 percentage points."

Concerns over the Republican Party's appeal to minority voters have been front and center since Mitt Romney lost the presidential election,but the math is not new. Gillespie's calculations, which were mentioned in a Timemagazine cover story in March on the Latino vote, are a simple matter of looking at population trends and extrapolating. The Census Bureau provides detailed estimates of population growth by race and ethnicity through 2050. The Hispanic population is expected to triple between 2008 and 2050, while the total number of white, non-Hispanic Americans will remain stagnant.

I wasn't able to reach Gillespie or precisely recreate his math, but the concept is valid. The following interactive lets you play with the figures by adjusting the predicted turnout and party preference among the four major racial and ethnic groups in America for the next nine election cycles.

The interactive starts by taking the Census projections for how many Americans will be old enough to vote from the present through 2048. You can see the total number of people eligible to vote in a given year by clicking the "Total" tab at the top.

On the default tab, "Voters," you see that population reduced down to the electorate by estimating the voter turnout for each group. Remember that the total voting age population includes many who are ineligible to vote, so it will never be 100 percent. The default values, in the controls below the graph, are the 2012 figures. Remember that exit polls figures are imprecise, so these are just ballpark estimates that you can adjust to your liking.

The last step is to divide the demographic blocks into Republican and Democratic voters. Again, by default the divides are from the past election, in which white voters supported Romney by about 20 points while minority voters trended Democratic to varying degrees. When you mouse over the "Voting Behavior" box, you see the demographic layer you're editing split into Democratic and Republican voters. Mouse over the graph to see the result for a given year.

You can't alter the past in this tool, but every time you adjust a number in the controls, it projects the values forward for the next nine presidential elections.

It's easy to poke holes in projections like this. An untold number of unpredictable events could drastically alter the demographics of the country. But population generally develops in predictable ways, and no one is questioning that the trends favor the Democrats.

The bigger question is how much we will care about racial and ethnic divides in the future. If you browse through old Census records on election demographics—what else is there to do on a Saturday night?—you quickly get a sense for the change in what characteristics we think motivate voters. Asrecently as 1972, the Census reported registration and turnout rates for Americans by "ethnic origin," with Germans and Brits clocking in as major voting blocs.

Forty years later, one hears only scattered discussion of national origin in political journalism. (There was a brief discussion, in August, of whether Mitt Romney's trip to Warsaw was an attempt to court Polish American voters, and one does hear a fair amount about Cuban Americans as they compare to other Hispanic voters.)

The best hope for Republicans may not be to improve their share of voters in interactives like the one above, but rather to make such exercises seem absurd. These are not completely different aspirations—number-crunching like this will become irrelevant when the political preferences of the groups even out and the data become uninteresting. That will happen either when both parties are equally palatable to minorities or when a person's status as a Hispanic American, an Asian American, or an African American becomes as noteworthy as German heritage.


The source code for downloading and parsing the population projections, in Python, is available on my Github page. Got questions? Email me at cewilson@yahoo-inc.com.