Showing posts with label social media biases. Show all posts
Showing posts with label social media biases. Show all posts

Saturday, May 31, 2014

31/5/2014: Twitter: Promoting Isolation, Ideological Segregation and All Things Good to Your Political Engagement


A very interesting study looking at comparatives of media and news use via twitter (social media) and traditional media (print, radio and TV). The paper, titled "Are Social Media more Social than Media? Measuring Ideological Homophily and Segregation on Twitter" (May 2014_ by YOSH HALBERSTAM and BRIAN KNIGHT is available here: http://bfi.uchicago.edu/sites/default/files/research/Twitter_may232014.pdf

Some highlights:

Per authors, "Social media represent a rapidly growing source of information for citizens around the world. In this paper, we measure the degree of ideological homophily and segregation on social media."

The reason this is salient is that there has been a "tremendous rise in social media during the past decade, with 60 percent of American adults and over 20 percent of worldwide population currently using social networking sites (Rainie et al., 2012)…. Indeed, this phenomenal growth in social media engagement in the U.S. and around the world has transformed the nature of political discourse. Two thirds of American social media users—or 39 percent of all American adults—have engaged in some form of civic or political activity using social media, and 22 percent of registered U.S. voters used social media to let others know how they voted in the 2012 elections."

Per authors, "Three key features of social media distinguish it from other forms of media and social interactions." These are:

  • "…social media allow users to not only consume information but also to produce information." It is worth noting that social media can also reproduce information produced on social media, as well as that produced by traditional media.
  • "…the information to which users are exposed depends upon self-chosen links among users." In other words, social media produced and distributed information can be self-selection biased. The extent of this selection is more limited in the case of traditional media, where individual biases of consumers can be reinforced by selecting specific programmes/channels/publications, but beyond that, the content received by consumers is the one selected for them by someone else - journalists, editors.
  • "…information on social media travels more rapidly and broadly than in other forms of social interactions. …[social media network model] leads to a substantially broader reach and more rapid spread of information than other forms of social interactions."

As authors put it: "Given these three distinguishing features, the rapid growth of social media has the potential to effect a structural change in the way individuals engage with one another and the degree to which such communications are segregated along ideological lines."


To examine this possibility, the authors construct "a network of links between politically-engaged Twitter users. For this purpose, we selected Twitter users who followed at least one Twitter account associated with a candidate for the U.S. House during the 2012 election period. Among this population of over 2.2 million users, we identify roughly 90 million links, which form the network." Based on political party followed, users were assigned ideological identifiers.

Two key findings of the paper are as follows:

  1. "…we find that the network we constructed shares important features with face-to-face interactions. Most importantly, both settings tend to exhibit a significant degree of homophily, with links more likely to develop between individuals with similar ideological preferences." In other words, we do show strong selection biases in networks we form. Doh!..
  2. "…when computing the degree of ideological segregation and comparing it to ideological segregation in other settings, we find that Twitter is much more segregated than traditional media, such as television and radio, and is more in-line with ideological segregation in face-to-face interactions, such as among friends and co-workers." Worse: we not only form biased networks, we also create selection-biased interactions and generate selection-biased chains and flows of content. Doh! Redux...

Conclusion: "Taken together, our results suggest that social media may be a force for increasing isolation and ideological segregation in society."

Wait… so we act on the social media base to create networks that are closer to friends networks… and this leads to… isolation?.. Well, my eye, I would have thought this would be the opposite…

But top conclusion makes sense:  "The issue of ideological segregation is important when providing such information. Exposure to diverse viewpoints in a society helps to ensure that information is disseminated with little friction across a large number of people. When a community is polarized and is divided into factions, by contrast, information may spread unevenly and may miss intended targets. Our results suggest that social media are highly segregated along ideological lines and thus emphasize these potential problems associated with the flow of information in segregated networks."

The problem, of course is: Can the selection bias be ameliorated? Can people be 'incentivised' to engage with ideological opposites? In my view - yes. This can be achieved most likely by educating people about systems of thought, logic, structures of knowledge, information. The thing is: in social networks, such education is both more feasible (volume of information delivered and speed are both higher) and probably more productive (because there is inherent trust in one's own network that is stronger than in detached media networks. Peers generate stronger bonds than preachers...

The paper has some fascinating data illustration of media biases, though - worth looking at in the appendix.