Overview
This project seeks to develop a tool to allow internet users to monitor their media consumption and better understand its health in two dimensions: truthfulness and partisanship. The Media Health Tool would also allow for experimental interventions to help users adjust their media consumption towards more accurate or less partisan content.
In the first phase of the project, a web browser extension will be developed which, installed in a user’s browser, would monitor their media consumption using lists of media sources (and accompanying partisanship scores) developed in previous literature [1,2]. In addition to recording this information, the tool will allow for behavioral interventions along two axes: information accuracy and source partisanship. In the case of the former, interventions when a user is viewing content from misleading or false sources might include pop-ups flagging the content, suggestions instead read articles from other sources about the same topic, or outright blocking content from those sources. In the latter case, interventions against partisan content might include showing the user a distribution of partisanship in their typical media consumption to prompt reflection, suggestions for other articles or other sources “across the aisle” to read, a leaderboard competition encouraging the consumption of more balanced news sources, among others. This framework for a browser-based tool for behavioral interventions has shown promise in prior work [3].
The second phase of this project will involve testing the various interventions for efficacy. Paid participants will be recruited for a month-long experiment in which participants use the Media Health Tool; in the first week their media consumption will be passively monitored to gather training data, and in the subsequent three weeks different interventions will be deployed and evaluated. A control group will receive no interventions. This experimental phase will allow testing along several dimensions: evaluating the relative efficacy of intervening along the two axes (accuracy and partisanship), as well as testing which interventions work best and identifying differences in participant receptiveness based on demographic characteristics such as age, political beliefs, or education level.
In the last phase of the project, insights about effective interventions will be leveraged to package and release the Media Health Tool for general use. This will allow any interested web user to download the extension and choose what insights and interventions they wish to receive for learning about and adjusting their media consumption habits.
[1] Faris, Robert, Hal Roberts, Bruce Etling, Nikki Bourassa, Ethan Zuckerman, and Yochai Benkler. “Partisanship, propaganda, and disinformation: Online media and the 2016 US presidential election.” (2017).
[2] Ronald E. Robertson, David Lazer, and Christo Wilson. 2018. Auditing the Personalization and Composition of Politically-Related Search Engine Results Pages. In Proceedings of the 2018 World Wide Web Conference (WWW ’18). International World Wide Web Conferences Steering Committee, Republic and Canton of Geneva, Switzerland, 955-965. DOI: https://doi.org/10.1145/3178876.3186143
[2] Ronald E. Robertson, David Lazer, and Christo Wilson. 2018. Auditing the Personalization and Composition of Politically-Related Search Engine Results Pages. In Proceedings of the 2018 World Wide Web Conference (WWW ’18). International World Wide Web Conferences Steering Committee, Republic and Canton of Geneva, Switzerland, 955-965. DOI: https://doi.org/10.1145/3178876.3186143
[3] HabitLab: https://habitlab.stanford.edu