Photo Credit: Scott Meyer
By Sarah K. MacFarland, Sonya Sachdeva, Spencer A. Wood, and Joshua J. Lawler.
When managing public lands, it’s important to understand what people value about nature and why. People’s values influence their beliefs and behaviors towards the environment and which qualities of natural environments they prefer. Consequently, decision-makers often face trade-offs among competing values, with value differences often leading to conflicts. Each year, land managers receive millions of written comments from the public during formal planning processes. These comments contain rich information about how people perceive and value nature. Although researchers have studied values using other data, such as surveys and social media, they have rarely analyzed public comments.
In this paper, we explored how to better understand what people value about public forest land by analyzing the comments they submit to the U.S. Forest Service. We developed new tools using natural language processing (NLP) and machine learning to analyze over 400,000 public comments. We looked for four types of values people express about forests: economic values (usefulness for human purposes), life support values (like providing clean air and water), aesthetic values (natural beauty and scenic qualities), and moral values (including spiritual, personal, and cultural connections).
We found that our NLP tools could successfully identify these different values in forest-related comments and was most successful at identifying aesthetic values. Our analysis revealed several patterns. First, projects that received more public comments tended to involve more discussion of aesthetic and moral values in those comments. Comments for recreation-related projects tended to include more aesthetic and moral values, but were less likely to mention economic values. Road management projects received comments that were less likely to mention moral, life support, and economic values.
Our study demonstrates that public comments are a valuable source of information about how people values forests and the natural world. We discuss how to carefully develop and test computational tools to detect complex topics like human values. We conclude that, when developed and evaluated carefully, these NLP tools are powerful enough to identify a diverse range of values people hold towards nature. We encourage researchers to explore whether these or similar methods can identify values in other large text datasets, such as public comments from other federal or state agencies.
Relational Thinking 