Data have been a crucial part of media coverage this past election as a statistician stepped into a spotlight as bright as for any rock star.
Nate Silver, whose blog FiveThirtyEight, licensed by the New York Times, accurately predicted presidential results in all states, provided a sharp contrast to pundit predictions that proved to be more desire than fact.
He showed that numbers don't lie, as long as they are in the hands of a competent and impartial analyst.
I talked with The Bee's Phillip Reese this week about this data evolution in election coverage. Reese, a database expert, reports data-driven stories and is in charge of our Data Center at sacbee.com.
The day after the election you worked a double shift to analyze voting data. What was the overall goal?
I wanted readers to see how their neighborhood voted – not just how the nation, state or their county voted. And I wanted to find voting patterns that would give context to our follow-up stories.
Before the election, I obtained electronic files showing the boundaries of most voting precincts in the region. The morning after the election, I started matching precinct-level election results to those boundaries, creating interactive maps and posting them online.
The first maps focused on the city of Sacramento. They showed which neighborhoods supported particular candidates or ballot measures. So readers could instantly see patterns in voting – North Natomas is fine with requiring green waste containers in place of "the Claw," for example, but South Land Park is not.
Then I created similar maps encompassing Sacramento, Placer and Yolo counties showing results for the presidential race, Jerry Brown's tax proposition, the proposition to abolish the death penalty, and the congressional race between Dan Lungren and Ami Bera. Readers could type in their address and see how their neighborhood voted. (The maps are still up at www.sacbee.com/ datacenter.)
What were the most interesting things you discovered in that analysis?
Sacramento is a good microcosm of the nation as a whole. A majority of voters in the four-county region have backed the eventual national winner in the past 11 presidential elections, including this one. Among states, only Ohio is a more reliable bellwether.
At the same time, the political polarization you see across the country is also prevalent here. There's a clear dividing line that seems to become more distinct every election: The city and the inner suburbs back liberal candidates and causes; the newer suburbs back conservative ones.
Mitt Romney didn't win a single precinct in the city of Sacramento with more than 15 voters, according to election night returns. Obama didn't win a single precinct in Rocklin and Loomis. The vast majority of areas that backed Obama also backed Brown's tax measure, and vice versa.
I was also surprised that more than 2,500 local voters supported Roseanne Barr, who was running for president on the left-wing Peace & Freedom Party ticket. Some of her strongest support came from rural, fairly conservative areas like Rio Linda.
How was it different from four years ago?
More results and better results. Sacramento County and Placer County election officials have improved their election reporting online, making it easier to get precinct-level election results soon after voting. They should be commended for that. Four years ago, I remember scanning in hundreds of pages of paper printouts from Sacramento County to create similar maps – and getting nothing from Placer County for weeks.
Given the increasing availability of data, what can we look forward to in four years?
I'd expect to see more polling at the state or local level in California, which will tell us more, ahead of time, about races other than the presidential contest.
I'd hope that county elections departments would continue to take advantage of improving technology when reporting election results. There's room to do better. Ten days after this election, for example, El Dorado County residents still couldn't go online and see anything about how their city, neighborhood or precinct voted. Their county elections department had only published summaries for the entire county or entire contests.
Nate Silver and his blog became a lightning rod because of his predictions. Can you speak to whether analyzing data is simple math, as Silver points out, or whether it indeed has been used to spin reality for political gain?
During the last few weeks of the election, dozens of presidential polls were coming out each week. When you combine the results of those polls, like Silver did, you can come up with a specific prediction of how the race is going to go. That's just math.
But remember that this sort of thing lets you establish likelihood, and that sometimes things that are likely to happen, don't happen. If your car started with no problem every morning for the last year, and you just had it serviced at the shop, it will probably start fine tomorrow. But it might not!
On Election Day, Silver predicted there was a 91 percent chance that Obama would win. If Obama hadn't won, I think everyone would be castigating Silver and talking about how data analysis is inferior to old-fashioned prognostication. That would have been unfair. Probability is not certainty.