Don't Worry, Be Happy: A Sentiment Analysis of the #NGSSchat Twitter Community

• In Event: Social Opportunity Spaces: How Social Media Can Inform/Shape Educational Policy Processes

Abstract

The #NGSSchat hashtag is used frequently on Twitter, and has helped form a community of educators to discuss the Next Generation Science Standards (NGSS) education reform effort in the United States (U.S.). While the importance of informal networks such as #NGSSchat for improving policy implementations and teacher professional development has been studied (e.g., Authors, 2019; Staudt Willet, 2019), research that examines the nature of discussions in such hashtag-based communities is underdeveloped. Therefore, this study examines the sentiment and content of the conversations within #NGSSchat.
Using Natural Language Processing methods, particularly sentiment analysis (Liu, 2012), we characterize the overall sentiment of 171,048 original posts containing the #NGSSchat hashtag, sent by over 16,000 individuals over nine years (2011-2019).We find that tweets were nearly universally positive, with a sentiment ratio—the proportion of negative to positive tweets—of 0.216, indicating that for every tweet sent with a negative sentiment, nearly five positive tweets were sent. While tweets were initially (in 2011 [sentiment ration: 0.455] and to a lesser extent in 2012—2014) more negative (although still positive), relatively few tweets were sent during these years, and years 2015—2019 were associated with positive sentiment, with the lowest sentiment ratio (.163) in 2016, indicating more than six positive tweets were sent for every negative tweet (Figure 1).
We also found differences between U.S. states. Figure 2 indicates greater rates of negative tweets by individuals from Alaska (sentiment ratio: 0.50), followed by Florida, the District of Columbia, Louisiana, and Montana (sentiment ratios: 0.330-0.364). Finally, when we explored sentiment ratios based on NGSS adoption status (using data from the National Science Teachers Association, 2020), we found early adopters (2012—2014) were associated with low sentiment ratios (0.183), followed by states who adopted the NGSS in the middle of their implementation (2014—2016; 0.229) and states who adopted them later (2016—2019; 0.252). States that did not adopt the NGSS were associated with the highest sentiment ratio (0.293).
Collectively, these findings contrast to those reported for the (even) more widely-used Common Core State Standards (CCSS)-focused hashtag, which was found to be overwhelmingly negative (Wang & Fikis, 2019). Sentiment ratios of #commoncore ranged between 1—33 (depending upon the U.S. state), indicating that every positive tweet had between one and 33 negative tweets. This contrasts our primarily positive #NGSSchat findings. Moreover, #NGSSchat has become even more positive over time. Reasons may include that the CCSS were developed and implemented prior to the NGSS, providing opportunities for the NGSS developers to learn to avoid implementation-related pitfalls. Moreover, differences may—in part—be attributable to organizers of the #NGSSchat hashtag (Shelton & Ende, 2015), who worked hard to connect many science education stakeholders including teachers, parents, administrators, and researchers (Authors, 2020).
Future research can substantiate these potential explanations, highlighting the roles of individuals’ location, changes in sentiment over time, and the topics of posts. This work has implications for data science research regarding how social media-based networks may reflect trends in wider educational reform efforts.

Authors

• Joshua Michael Rosenberg, The University of Tennessee - Knoxville

• Conrad Borchers, University of Tübingen

• Benjamin Gibbons, Emory University

• Elizabeth B. Dyer, The University of Tennessee - Knoxville

• Daniel John Anderson, University of Oregon

• Christian Fischer, Eberhard Karls Universität Tübingen