The voices of Urban Institute's researchers and staff
June 25, 2014

Guns, policy, and big data: what Internet search data say about public interest and behavior

June 25, 2014

What can search engine data tell us about offline behavior? Probably more than you’d expect. Researchers have used Google data to predict everything from influenza patterns to stock market movements. Though big data can’t provide the level of specificity available through traditional “small data” research, it can supplement insights garnered from surveys, experiments, and conventional statistical reporting. It may also help bridge gaps where reliable data don’t exist.

Sitting at the center of a perfect storm of partisan spin and limited data, few relationships in social science remain as elusive as that between Americans, their guns, and the current events that shape gun-buying decisions. There are no publically available counts of how many guns are bought or sold in America—such a registry is illegal at the federal level, and states don’t have any consistent data on gun ownership.

Soft measurements, like media analysis, are equally problematic because of political interest in portraying gun ownership as moving in one direction or another. Depending on the source, record numbers of Americans are either arming themselves or rejecting gun ownership entirely.

Because of challenges like these, gun ownership research in the United States is done through proxies, like the percent of suicides committed with a firearm. Online search data offer a new source of information we can use to investigate how public interest in gun ownership reflects response to big events, and how interest translates into action. Here’s how.

Interest in guns tracks big events

We tracked the frequency of gun purchase-related search terms over the past 10 years: “buy gun,” “buy guns,” and “buy a gun.” We then normalized the search score between 0 and 100 so that  100 represents the highest levels of search engine activity and all the other data points are relative to this point. The data are also adjusted for the total number of all searches on Google.

The resulting graph reveals how effectively search data can track public interest. Media outlets reported increased interest in gun laws and ownership after both the election of Barack Obama in November 2008 and the Sandy Hook school shooting in December 2012. Search engine data offer new support for these reports, showing spikes in related search-engine activity around both events.

From interest to action

Linking Google search data to other sources gives us the next piece of the story: how public interest in buying a gun relates to Americans’ gun-buying activity. The National Instant Criminal Background Check System (NICS) conducts a check whenever a licensed firearms dealer makes one or more sales to ensure buyers are legally allowed to purchase firearms. (Although NICS has several limitations as an indicator of gun buying, it’s a common measure of US firearm commerce we can use to get a sense of gun purchases.)

NICS and Google search data measures are closely correlated and show a spike in activity around both the election of Barack Obama and the Sandy Hook school shooting. Americans’ interest in buying guns and the background checks that facilitate these purchases follow each other closely.

What’s next?

There’s much more work to do to verify the relationship between Google data and gun purchasing, but this example demonstrates the immense opportunities presented by search engine data. Both the granularity and accessibility of search engine data are powerful supplements to conventional research tools. By drawing on trends in search activity across states and even cities, researchers can develop new insights into how changes in public opinion and public policy affect one another, and policymakers can be more responsive to the needs of the communities they serve.  

SHARE THIS PAGE

As an organization, the Urban Institute does not take positions on issues. Experts are independent and empowered to share their evidence-based views and recommendations shaped by research.

Comments

I am disappointed that you don't disucuss the big biases in the NICS data. First, many states use it to regulate their conceal carry permits. States like Kentucky will run annual, or even quarterly NICS checks that get reported with the total NICS numbers. So its impossible to tell which of this checks are for purchases or from regulation agencies doing routine checks.

The NICS data is also limited in that it represents only a subset of the states (i.e. those that rely on the FBI for checks of the NICS).