Catherine Tucker on Google Trends and NSA Surveillance

Catherine Tucker, the Mark Hyman Jr. Career Development Professor at MIT's Sloan School of Management, writes in with the following thoughts on measuring public reaction to the NSA revelations using Google Trends search data:

When I saw Ben’s recent post recommending using Google Trends data to examine the effects of surveillance revelations, I realized I had been doing a terrible job of highlighting a new academic paper where I do just that. There are many possible reasons to think that the surveillance revelations would not affect Google users’ search behavior. As Ben suggests, the revelations could simply be of limited interest to people outside the Beltway, and therefore would have a negligible impact on overall US or international search behavior. Google users in general could be apathetic or ill-informed enough about surveillance that they did not change their behavior; alternatively, they could be well enough informed that they had already factored in at least the possibility of government surveillance of what they were searching for. My coauthor, who runs the advocacy group Digital Fourth, thought there might be an impact; I suspected there wouldn’t be;    but as it turns out, the impact was significant and measurable. Using lists of possibly troublesome search terms from the Department of Homeland Security’s Social Media Monitoring Unit and elsewhere, we identified 282 search terms that were then independently evaluated for whether raters thought they would get you in trouble with the government or with a friend if it became known you had used this search term. We employed coders to download weekly data from Google Trends for each of these search terms for the U. S. and for its top ten trading partners, covering the whole of 2013. We found that U. S.-based search traffic falls by around 5% in the Google Trends index for government-sensitive terms after the PRISM revelations. This is the first academic empirical evidence of a chilling effect on users’ willingness to enter search terms that raters thought would get you into trouble with the U. S. government. When we look outside the U. S., at the effects on its top ten trading partners, we find that Google users in these countries on average searched less not only on  government-sensitive search terms such as “anthrax” but also on personally-sensitive terms  like “eating disorder.” So we know that there is that much of an effect.  What we can’t tell from our data is what other steps international users may have taken in reaction to the surveillance revelations. Maybe they will also switch to non-U. S. search engines; maybe there will be broader knock-on effects on sales of Google products such as Android phones. If international users are reacting strongly, it may damage the exports of U. S. tech companies. We view this paper as being an initial attempt to try and understand how consumer surveillance affects search behavior both in the U. S. and in the U. S.’s major trading partners. It suggests that there may be tradeoffs between government surveillance and the profits of U. S. data-enriched technology firms.  The full paper, for those with an interest in statistics or econometrics, is available here.