Designing the participatory experience
Interface design of online participatory systems can subtly (and not so subtly) influence whether and how members of the public will engage with policymaking's deliberative process. This project developed design elements that promote more informed, more broad-based, and deeper participation on Regulation Room, SmartParticipation and other online deliberative platforms. One study looked at how signaling community norms and expectations through action prompts may enhance participation. Conducted through Amazon Mechanical Turk, this study tested three conditions: generic prompt, community norms specific prompt, and content specific prompt.
Drafting Room Experiment
The Drafting Room experiment tested the boundaries of effective online civic engagement. Is it possible to go beyond soliciting feedback from individual members of the public and help them move towards collaboratively producing effective policy inputs? Specifically, we focused on three main questions:
- What psychological and experiential factors predict different levels of civic engagement in online deliberation of policy?
- What effect does engagement in online deliberation of public policy have on people’s perceptions of the decision-making processes and institutions?
- What effect do different facilitative interventions have on co-production of policy inputs in an online environment?
We investigated those questions by using a large-scale controlled experiment focused on deliberation of an actual campus policy change. The practical goal of the project was to test platform features and facilitation procedures for collaborative drafting of policy inputs by the members of the public.
Framing for participation
One of the main challenges with online civic engagement in policymaking is breaching the wall of skepticism and distrust, which has become common in developed democracies. There is a significant body of literature addressing this issue within the context of traditional, mass media, but there is still a lot to unpack in the context of personalized, new media environments. Tackling this challenge, this study looked into the framing of calls for engagement on social media studies. We conducted this experiment using Facebook's advertising engine. Controlling for gender and political orientation, we presented Facebook users with either thematically or episodically framed calls to action crafted with less than 90 characters.
Predictors of effective online civic engagement
Policymaking bodies have limited financial, human, computational, and temporal resources for recruiting members of the public to participate in online deliberations surrounding rulemaking processes. Thus in order to make the most efficient and effective use of these resources, this project was an effort to improve outreach strategies by identifying people who are likely to be highly motivated and capable contributors. Its aim was to develop natural language processing techniques that could analyze text online to detect cognitive and experiential characteristics that are positively or negatively associated with a person's willingness and ability to participate effectively.
Experiments concentrated on recruiting people from the social media platform Twitter by analyzing the text that Twitter users posted. An initial experiment in Spring 2012 examined whether text similarity between rulemaking concepts and a Twitter user's bio, tweets, or some combination was correlated with that person's willingness to participate during an open comment period on CeRI's Regulation Room.
Additional experiments continued to explore predictors of an individual's readiness for engagement. In particular, the focus was on developing methods for:
- identifying topical expertise and interest according to online behavior and content
- determining linguistic markers of psychological characteristics known to motivate engagement such as self-efficacy and certain personality traits.
Additionally, we investigated whether outreach messaging could be crafted to amplify and appeal to these interests and characteristics in order to be more persuasive, achieve better response rates, and elicit higher quality comments.
Regulation Room was designed and operated by the Cornell e-Rulemaking Initiative (CeRI) and hosted by the Legal Information Institute (LII). The site was a pilot project that provided an online environment for people and groups to learn about, discuss, and react to selected rules(regulations) proposed by federal agencies. It expanded the types of public input available to agencies in the rulemaking process, while serving as a teaching and research platform. Learn more about the Regulation Room here.
Situated knowledge in policymaking
Technology-enabled civic participation in policymaking has become one of the most important e-government topics. As barriers to participation are lowered and more citizens are willing to engage directly with decision makers, sifting the flow of commentary to identify the most essential and useful information remains a significant challenge. This project was focused on developing Natural Language Processing (NLP) based solutions for extracting situated knowledge from public commentary on policy; it also aimed to create tools for exploring various aspects of that knowledge. This work aimed to make broad civic participation more effective in complex public policymaking at the federal, state, and municipal levels.
Started in 2012, our work focused primarily on conceptualizing the value of situated knowledge in policymaking activities and building an annotated corpus for NLP analysis. This work resulted in a number of publications and a corpus that allowed for initial NLP experiments.We continued to expand the annotated corpus and build on the preliminary results of our NLP experiments, exploring the best ways to visualize situated knowledge for the purposes of more efficient and effective management of online public consultations. This work was funded by the Jacobs Technion-Cornell Innovation Institute Research Project Award and the NSF.
Unsubstantiated Claim Detection
With advancements in information technology, we experienced an explosion of user participation in the web environment. In order to efficiently manage the growing amount of information, this project aimed to automatically evaluate the quality of user generated texts, such as reviews and comments, by means of determining whether each claim is accompanied by substantiation. A working assumption here was that user generated texts that consisted of substantiated claims are of better quality than those that contained unsubstantiated claims.