Stanford’s Matt Gentzkow on Polarization

By TAP Guest Blogger

Posted on August 26, 2016


Written by Ananya Sen.


The term ‘political polarization’ is invariably associated with modern day politics across the world, be it Donald Trump vs. Hillary Clinton (vs. Bernie Sanders) or the rhetoric around Brexit.


There is a sense that different sections of the general public take a hard stance on various issues and find it extremely difficult to reach a consensus, much more so than the preceding decades. This phenomenon has been attributed, in part, to the rise of the Internet in general and social media in particular. The fact that individuals might want their own beliefs reaffirmed would lead them to seek such news online which could in turn lead to an `echo chamber’ effect. If we look at the big picture, these concerns are extremely important since it is crucial that individuals have correct beliefs about different issues so that they make informed decisions in a democratic society and are not swayed by their preconceptions, stereotypes, and echo chambers.


These are some of the issues that Matthew Gentzkow’s research addresses. The Stanford economics professor examines whether actual data supports the above widely held beliefs.


Gentzkow and Shapiro (“Ideological Segregation Online and Offline,” 2011) analyze whether people are more segregated in the way they consume their online news relative to their offline news consumption and other offline personal interactions. Using the isolation index, a standard measure used in quantifying the extent of racial segregation, they find that in absolute terms the level of segregation of online news consumption is low. Moreover, online news consumption is only slightly more segregated than offline news consumption while it is lower than offline personal interactions with family and friends. Further more they did not find any evidence of this segregation becoming more severe over time. This was a surprising result, which goes against the wisdom of the day. They highlight that this is happening despite a large number of news outlets with extreme views available online; most of the online traffic is driven by mainstream centrist news websites. They however do not analyze the role of social media (Facebook, Twitter etc.) in online news consumption and ideological segregation. This research gap is being filled slowly but surely and shows that the ideological segregation in social media is similar to offline social interactions but still much lower than conventional wisdom would suggest.


Relatedly, there are concerns regarding an increase in partisan language used by politicians which could have fuelled a more polarized political environment overall. The language used by politicians can have a direct impact on the way the general public thinks about issues (for example `death tax’ vs `estate tax’) and also indirectly because news outlets start using these phrases in their own articles (What Drives Media Slant? Evidence from U.S. Daily Newspapers,” Gentzkow and Shapiro, 2010). Gentzkow et al. (Measuring Polarization in High-Dimensional Data: Method and Application to Congressional Speech,” 2016) develop a structural model to see how the use of partisan language has changed over time in the U.S. Congress. Again, contrary to conventional wisdom, they find that partisanship of political language has increased over the past couple of decades and has reached unparalleled levels. The exact reason for this increased partisanship and how it would potentially affect political polarization in the electorate remain open questions.



This post was written by Ananya Sen. Mr. Sen is a doctoral candidate at the Toulouse School of Economics. His research interests include Economics of the Media, Digital Economy, and Applied Microeconomics.


The preceding is republished on TAP with permission by the Toulouse Network for Information Technology (TNIT). “Stanford’s Matt Gentzkow on Polarization” was originally published in TNIT’s August 2016 newsletter.


Read more from Professor Gentzkow on polarization and ideological segregation: