Saturday, December 02, 2017 by Ethan Huff
Does the type of car you drive indicate your political affiliation? A computer algorithm developed by researchers from Stanford University seems to suggest so, having found that neighborhoods dominated by sedans tend towards liberalism, while those containing mostly pickup trucks lean more towards the right.
The model used by the Stanford team evaluated vehicle types identified in 50 million images contained in Google’s Street View database. A total of 200 American cities were included as part of the research, which paid close attention to the most prevalent types of vehicles driven in each neighborhood in order to compile data that was then used in a comparative analysis of voter trends.
What was discovered was that in neighborhoods where sedans outnumbered pickup trucks, there was an overwhelming 88 percent chance that these areas voted primarily Democratic. On the other hand, neighborhoods where pickup trucks outnumbered sedans were 82 percent more likely to vote Republican, suggesting that conservative Americans prefer bigger and more powerful vehicles.
“We show that it is possible to determine socioeconomic statistics and political preferences in the U.S. population by combining publicly available data with machine-learning methods,” the team wrote in their paper, which was published in Proceedings of the National Academy of Sciences (PNAS) journal.
Part of the research also involved honing down specific makes, models, and years of vehicles to perform an even deeper analysis of how car preference correlates with how people vote. The Stanford team was able to identify all of this information in every area they looked at for every single car sold in the United States since 1990.
They also looked at race, income, and education levels in conjunction with this data to see if there was anything more to learn. What they found is that white neighborhoods tend to have more foreign-made vehicles like Volkswagens and Aston Martins, while black neighborhoods tend to have more domestic cars like Chryslers, Buicks, and Oldsmobiles.
To verify this further, the researchers cross-checked their observations with both data from the U.S. Census Bureau and voting results in each precinct. Doing so ultimately confirmed the team’s original findings, leading them to suggest that such an analysis tool might one day be used to compile demographic data as opposed to the Census Bureau’s standard door-to-door approach.
However, some people worry that using Google’s Street View data for such purposes, specifically as it pertains to how people vote in elections, might be misused for nefarious purposes. It would be rather simple, say people like Scott Cleland, a tech blogger who has been notoriously critical of Google’s privacy protections, for political parties to reconfigure entire neighborhoods to gerrymander the election process, which could end up leading to widespread voter fraud.
“Once you start sharing and deducing people’s private stuff in a group setting for group purposes, it doesn’t take a genius to see that this could end badly,” he says. “You can imagine the manipulations of a neighborhood.”
But others insist that gaining access to data like this with relative ease will actually produce the opposite. Social scientists will be able to better assess how people think and vote without having to pour millions or even billions of dollars into the census models of old just to procure basic demographic information.
“The more you know about someone, the better you can engage with them and the more relevant you can make the communications that you send to them,” says Alexander Nix, the head of Cambridge Analytica, the company that Donald Trump’s presidential campaign utilized to target voters during the 2016 presidential election. “Our job is to use data to understand audiences.”
Sources for this article include: