Truthy Research: From the Day 2 Hackathon at the Truthiness Conference

Posted on

About the Author:

Aaron Shaw

Aaron Shaw

Aaron is a Fellow at the Berkman Center for Internet & Society and a Ph.D. candidate in the Sociology Department at UC Berkeley. His research focuses on political and economic dimensions of collective action online.

As part of the Truthiness hackathon, a group of us wanted to design empirical studies to investigate how (mis)information works and how it effects people’s behavior. After some brainstorming, we decided to focus on the following three topics:

  • The dynamics of information flow in networks/communities.
  • The factors that determine how bias works and when people change their minds.
  • The effects of advocacy groups’ dissemination of (mis)information on political behavior.

We formed three sub-groups, each of which worked on developing a set of research questions and (to varying degrees) elaborating specific studies in these areas. Here they are:

Information dynamics:

How does information spread in twitter through retweets? Various studies have looked at RT behavior at various periods of time, with varying results. Has the way we have been using the system changed with time? Do the changes we observe correspond to differences in the dataset or data collection? Or do different types of spreading map to different topics, levels of credibility, etc?

TIME PERIOD    AUTHOR     TITLE     Samplesize   RTsize  usedHashtag?
1/26/09-6/13/09     boyd+       T,T,Rt                0.72M      3%          NO
6/6/09 – 6/31/09     Kwa+        What is Tt?        106M     ?             NO
1/9/10-1/17/10       MM           Obsc->Prom      0.25M    40%        Yes 
9/14/10-11/1/10     Cono+       Pol Polar          250k      ~30%       Yes
TODAY! 3/7/12      hackathon  ch hum beh      0.03M    33%       NO

 

 

We ran a small sample from yesterday, and we did a literature search of known papers.

Two (tentative) findings:

  • Users’ way of RT usage has drammatically (3% -> 33%) changed in the last two years.
  • At the time of elections, RTs are used much more often (>40%) in diffusing info.

 

How bias works and when do people change their minds:

For this topic, we discussed research approaches to explore the interaction between value-based belief systems, depth of belief, openness to new information, and the ability to identify biases and misinformtion. We came up with three research designs:

  • Data mining: identify and document trends in Twitter, Facebook, and the blogosphere for those that have either said: “I was wrong about…” or “I changed my mind about..”
  • Games: create an experimental gaming platform that allows to explore the interaction between social dynamics and opinion formation
  • Conduct surveys that evaluate the relationship between the depth of beliefs (how confident respondents are in the accuracy of their opinions), their depth and sources or information, cultural and political views, and the ability to identify biases and effectively filter out misinformation.

Advocacy Groups’ effects on political behavior:

Our third sub-group designed a study to examine the effects of advocacy organizations’ dissemination of ‘truthful’ information on political engagement and behavior. Specifically, we propose to look at the impact of union voter ID mobilization efforts – an issue where misinformation has been rife and has played a major role in shaping advocacy group interventions. We randomly assign some unions to disseminate additional truthful information to their members and then measure the effects of this information dissemination on three critical campaign outcomes (proportion of previously un-id’d voters who get ids; who register; and who vote.