States with a high percentage of Mac or Chrome users are also highly likely to cast their electoral votes for President Barack Obama, while states with a high percentage of Pinterest or Kindle Fire users are likely to lean toward Mitt Romney.
Using state-by-state election forecasting data from the New York Times’ FiveThirtyEight blog and Monetate’s own data from 332 million online shopping sessions from July-September 2012, we’ve found some surprising correlations between states’ red/blue political leanings and shoppers’ preferred devices, operating systems, browsers, and websites.
Below, we’ve graphed some of the stronger correlations.
The horizontal axis of each scatter plot shows the likelihood that Obama will win each state, based on the New York Times’ data. States further to the left (in the graph, not ideologically) are less likely to lean toward Obama; states further to the right are more likely to lean toward Obama.
The vertical axis shows the market share of a particular device, operating system, or browser, or the referral rate of a particular website, based on data from Monetate’s soon-to-be-released E-commerce Quarterly Report for Q3 2012, which analyzed online shopping sessions at major retailers from July to September. States in which the market share is relatively low are plotted toward the bottom, and states in which the market share is relatively high are plotted toward the top.
Among blue states, the correlation with Mac use was the strongest of the metrics we tested against. Among red states, the correlation with Pinterest referrals was the strongest.
We noticed an interesting trend across the four versions of Internet Explorer that we analyzed (6 to 9): The higher the version, the stronger the negative correlation was for voting Obama.
Perhaps just as interesting as the data that show a correlation with red/blue leanings are the data that show no apparent correlation.
For instance, the scatter plots for Facebook users and those who browse on Apple’s iOS devices (like the iPhone) look a lot more, well, scattered, with no significant trend either way.
Want to view more analysis of red-versus-blue trends? View our new infographic, “Your Ecommerce Guide to Election Day.”
For data-science nerds only
Monetate backend engineer Jeff Patti collected state-by-state data on projected voter share from the New York Times’ FiveThirtyEight blog.
He then plotted it against data from the upcoming Monetate EQ Report, which includes a breakdown of traffic segments from 332 million online shopping sessions (geo-located to one of the 50 states) at leading online retailers.
The EQ Report also includes data from the previous four quarters, for a total of 2.2 billion sessions analyzed. That’s a lot of data to parse. We put the raw data on Amazon S3 and used Apache’s Hive and Hadoop to query and crunch the numbers. The process took about 350 computer hours on Amazon EC2, with about 50 machines running concurrently. Total cost to compute: less than $150.
Give us your armchair analysis
We’d like to hear what you make of these results. For instance, what do you think explains the increase in steepness of the best-fit line as Internet Explorer version increases? Are you surprised that any particular device/OS/browser/site is correlated with red or blue states? Post your thoughts in the comments section.