Detecting Trends in Community Engagement: At Wesleyan and Beyond

When it comes to activism and community service, Wesleyan has always tried to stay ahead of the curve. But this can be difficult, as the concerns and trends of community engagement are constantly shifting. Often, new topics will seemingly erupt out of nowhere, and it will take a while for word to spread. There are so many existing concerns that it can be difficult for new voices to be heard and for old voices to catch on to the changes. It might seem as though the trends in community engagement are shifting constantly, without any pattern. But can technology detect one?

Wesleyan’s Text Mining class was assigned the task of investigating this dilemma. They were asked to analyze the relationship between approaches to community engagement in the past and what people desire from it in the future. For past data, they collected the text of old Argus articles tagged “community engagement.” These articles were meant to illuminate what kinds of activities were most popular. Present data was collected through focus groups that were asked about the current state of community engagement at Wesleyan and how it could be improved. From this data, class groups hoped to discover how much the current activities overlap with the desires of the focus groups, as well as identify which community engagement topics are popular and which ones are new.

The biggest challenge for the class was figuring out how to pin down the major topics that were discussed in the Argus articles and focus groups. One group predicted the topic of each article from looking at the title and the first paragraph. From this, they decided that the eight most frequently used words were a good representation. This group did the same thing for the focus group summaries, and pulled the ten most frequent used words. They then compared these two lists of topics and found that, generally, the pre-existing topics seemed to be foundational aspects of community engagement, while the new topics were describing new ways to intervene and improve. Below is an example of one group’s results:

  • Frequently discussed Topics: volunteerism, inclusivity, health, construction, appreciation
  • New topics: family, social media, Christmas, substance abuse prevention, privacy

However, many groups came up with different topic names. This is because, while the analysis functions could tally up the frequency of use for each word, they could not give an overall label to a pattern. Once their model had returned a list of most frequently used words, groups had to infer which topic a word was connected to. Some groups did this by looking at the words that came before and after a frequently used word in order to detect associations.

This brings forward a dilemma in this kind of meta-analysis: Currently, the technology can’t do everything for us. We have to make inferences, and this relies on background knowledge and circumstances that will differ from person to person. For example, groups had to make other decisions, such as: How do we separate old and new periods? One group decided to split the data into two periods from 2003-2014 and 2015-2017. How do we discern which popular words are meaningful and which aren’t? Often most used words were ones like “Wesleyan,” “Community,” and “Students,” which don’t reveal much about the topic because they are part of the base rhetoric for community engagement at Wesleyan. Why are new topics coming forward? One group made the interesting point that the Argus articles were showing how community engagement is presented by the media, which is not necessarily going to reflect the desires of the focus groups.

There is a disconnect here between what we see and what we are able to gather from background information. Standards of community engagement can either be really broad and really narrow, depending on the outlook. In the end, the students decided that the Argus was presenting themes more catered to students, while the focus groups were thinking about the community. However, the focus groups were also thinking about what a community would ideally want, which is not always the same as what we can accomplish now. When reflecting on this assignment, many students said they didn’t know of all the opportunities that are currently out there until they poured through the data. This suggests that a broad definition of community engagement, one that includes basic opportunities and community goals, could aid in better communicating what is out there. The definition of community engagement and what it looks like is changing. Wesleyan needs to communicate that this change has happened, otherwise we will be sitting in the gap between too many opportunities and not enough participation.