Participants could agree, disagree, or pass. Pol.is would then serve up the next statement. At any point, participants could also put statements of their own into the queue.
Over 11 days, 2,026 people participated. 403 of them submitted statements on issues ranging from traffic to education to immigration. As the moderators, we rejected or rephrased roughly a third of the statements—usually in order to weed out duplicates, sometimes to improve clarity, and only very occasionally to eliminate inappropriate or joke statements.
In the end, the Pol.is included 609 statements. On average, participants voted on 111 statements each, resulting in a total of 225,460 votes!
Throughout this process, our goal was to maximize participation. We spread word about the virtual town hall via an email campaign, media outreach, flyers, and by contacting government officials, directors of nonprofit organizations, educational leadership, clergy, and community groups. In order to increase participant comfort with sharing opinions, we made the process open and anonymous. Anyone could vote or comment. We did not collect personal or demographic information. We did, however, identify the zip codes associated with the IP addresses of participants in order to measure participation from outside the area. As a result, we can say with certainty that participation was overwhelmingly local to Bowling Green and the surrounding area.
This process had some important limitations. First, although we publicized the event through a wide range of local networks, we made no attempt to ensure a randomized participant pool. As a result, we cannot characterize the margin of error for particular statements or readily extrapolate them to the larger population.
Second, not all participants saw the same statements. Pol.is serves statements via an algorithm that favors both the most recently-submitted statements and those that already demonstrate high levels of engagement (i.e., few ‘passes’). In practice, however, statements submitted early in the process were more likely to win this sorting process and be seen by the most people. With over 600 statements in the queue, most people saw only a small portion of the statements.
In the end, a substantial part of the local population participated in the event—some 2,000 participants in a county of around 123,000 people over 11 days (Bowling Green, where most of the participants were located, has 65,000 residents). Participation was still growing when we closed the poll.
Out of this process, we identified areas of relative agreement and disagreement that emerged as participants voted on each other’s statements. We then created opportunities for participants and local leadership to engage with and test the results against their own understandings of the issues and challenges facing the area. To this end, we held a Town Hall and community workshop in late February.
This report is part of that process, too.
Consensus and Groups
Like any survey tool, Pol.is can show the vote totals for submitted statements. But it can also identify, in real time, voting patterns that present a clearer picture of how agreement and disagreement are organized in a community. In the case of Bowling Green, voting clustered into two groups (Group A and Group B), which tended to line up, respectively, with liberal and conservative perspectives on a number of the more divisive issues.
Here is one of the ways that Pol.is represents this clustering (the dotted lines define the boundaries of two groups).
Obviously, the clustering process is complex, and we won’t do a deep dive here. The most important thing to note is that every statement is numbered on the graph (001, 002, etc.) and that statements that received higher levels of agreement are closer to the middle of the bullseye. By the same token, the more contentious statements drift out to the edges. The blob in the middle indicates that participants generally agreed on most of the statements. Differences between the groups are formed by the smaller number of low-consensus statements toward the edges.
Another way to look at the distribution of agreement and disagreement is via a ‘beehive diagram.’ The Pol.is technical report has a nice animated version that provides mouse-over detail about the statements. The version below is just an image.
For our purposes, the key point is again that the overwhelming majority of statements are ‘high consensus’ statements on which nearly everyone agreed—without major differences across the two groups. Nearly all of these deal with ‘quality of life’ issues: traffic, zoning, services and amenities, and so on. The divisive comments relate primarily to a handful of social issues, including immigration and refugees, adoption of a local fairness ordinance (LGBTQ rights), and drug policy. What’s the most divisive statement in this context?
Overall Group A Group B
How Should I Read These Bar Charts?
The first thing to keep in mind is that not every participant sees every statement. Pol.is distributes the attention of participants across a spectrum of statements, favoring the most recent and the most highly engaged statements. Although it’s possible to respond to them all—and a few people came close to doing so—most people stopped before completing all 609.
In the case of the sanctuary city statement, among the 1578 people who voted at least 7 times, 1138 saw the statement. Among these, 31 percent overall agreed, 41 percent disagreed, and 27 percent passed.
The white section on the right represents the number of people who didn’t see the statement.
The Group A and B breakdowns follow the same logic but are based on participants who voted at least seven times (seven is the mathematical threshold that allows Pol.is to identify groups). 1,578 participants did so. Groups A and B, accordingly, are subgroups of the 1,578 and of slightly different sizes: (liberal) Group A has 751 members; (conservative) Group B has 827.
Given the multiple sources of uncertainty in this process, it’s important not to read too much into the exact breakdowns of the group votes. The main takeaway is that the sanctuary city issue helps define a larger structure of disagreement in the community. It is perhaps not very surprising that the major disagreements play into familiar categories of left and right. What may be more surprising is the range of issues on which people substantially agree.
Because the Pol.is survey is generated by its participants, it has the potential to go in any direction. There are no fixed questions or topics. There are, however, loose forms of path dependency as participants build up an impression of the process based on the statements they initially encounter. For our part, we provided a framing question designed to encourage respondents to think locally. We asked: “What do you believe should change in Bowling Green/Warren County in order to make it a better place to work, live, and spend time.” With the help of the Bowling Green Daily News, we also added a number of starter statements in order to prime the pump.
By all appearances, the framing statement had a powerful effect. Most of the statements referenced very local community issues—traffic, development, zoning, school and municipal governance, and others. Some were local inflections of issues with wider state and national profiles—a debate over passing a fairness ordinance in the city and drug sentencing, for example. Only a few referenced state, federal, or international decision making or policy areas. The subsequent Town Hall and workshop retained this focus. Overall, the Civic Assembly dealt with issues that participants thought could be addressed locally.
The rest of this report presents selections of statements from the survey that speak to some of the major lines of participant concern, with a focus on statements that received relatively high levels of participant engagement. We highlight areas of agreement first, and then move into areas of disagreement.
This is not a complete or exhaustive account. If you’d like to dig deeper into the full results, see the Pol.is technical report.
High Consensus Topics
On the great majority of issues related to quality of life, governance, services, local economic issues, and development, there is a very high level of agreement among residents of Bowling Green. Most of these statements can be grouped into a few categories.