How social media big data helps us better understand social dynamics

If tweets are measured in characters and a picture is worth a thousand words, what do you get when you combine and examine thousands or even millions of social media posts at once? The answer is a lot of data, and researchers at the University of New Mexico are using it to study social dynamics and human behavior.

Dr. Xi Gong, assistant professor in the Department of Geography & Environmental Studies and founder of the Spatially Integrated Social Science (SISS) Lab at UNM, uses big data from social media to do just that. His lab has studied topics ranging from sports fan behavior to crisis communication.

“The beauty of social media is that it provides a more cost-effective way to observe and analyze a large sample of human activities and behaviors than traditional measures such as surveying or interviewing a population,” Gong said. “What initially drew our team to social media research was its benefits and that it fits very well with our research on social dynamics.”

It all starts with a post. You watch your favorite basketball team win or lose a game in real time and vent your thoughts on social media. If your team won the game, you can post, “We won!” on social media? but if the team was losing, it was definitely caused by bad calls from the refs, at least that’s your interpretation.

These behaviors –– using your team’s victory as a win or deflection of the opposing team, fans and referees–– known as Basking in Reflected Glory and Blasting, respectively, are just two tactics people use to manage social picture them. Gong wanted to investigate attitudes using tweets from the 2019 NBA Finals.

More than 11 million tweets were collected for the study, “Exploring the dynamics of sports fan behavior using big social network data,” published in Applied Geography, and more than 240,000 of them included geotags. All tweets and related data, such as the time they were posted and how much engagement they received, are all taken.

“With this large dataset of these geo-tagged social media posts, we can really explore the spatial and temporal changes in human attitudes and behavior, what they say, or the ups and downs of emotion related to the process and outcome of the game “, he said.

In studying the NBA Finals, Gong and his research partner, Yong Wang, were able to observe and support a number of existing social theories about how fans perceive themselves as part of their favorite team and will blame the external forces when faced with negative information about the group.

In 2021, Gong and Dr. Xinyue Ye tweeted “Governors Fighting Crisis: Responses to the COVID-19 Pandemic into the United States” The Professional Geographer. The pair used Twitter data to examine how US state governors used the platform for crisis communication during the COVID-19 pandemic.

It used spatiotemporal analysis, network analysis, and text mining to see if state governments were communicating effectively with the public, which governments cooperated, how tweet frequency aligned with the severity of the pandemic, and how dynamics differed between political parties, etc. .

The study provided suggestions for how government agencies and officials can leverage social media in developing future crisis communication plans. Tips included following more public accounts and consistently listening to their concerns, organizing crisis communications with hashtags, tweeting more about policies and updates, and following and updating governor peer accounts more often.

“I’m interested in the spatiotemporal dynamics of human society, but social media is not the only data we’re looking at,” Gong said. “We also combine social media data with data from other sources.”

More recently Gong used Twitter georeferencing data from the US from 2018 and 2019 to better understand people’s attitudes towards fracking. A multiscale geographically weighted regression (MGWR) was used to examine county-level relationships between rates of negative tweets and factors such as demographics, economic development, environmental impacts, and others. Millions of tweets were used in the study, “Understanding Public Perspectives on Fracking in the United States Using Big Social Media Data,” published in GPS timelines.

Researchers found that factors shaping people’s attitudes toward fracking differed across different regions and scales in the U.S. People living in eastern and central U.S. counties with higher unemployment rates, counties east of the Great Plains with fewer fracking sites nearby, and area Western and Gulf Coast counties with higher health insurance enrollment are more likely to express less negative views of fracking activities.

Findings from studies using social media can facilitate understanding of public perception on certain issues and help policy makers make decisions about policy adjustments.

Recognized Gong social media studies have advantages and disadvantages. While it can be a great way to reduce research costs, there are some considerations that need to be made.

“It has limitations. Different social media platforms have different age groups of users, a single platform may not fully represent the entire population. Therefore, combining information across social media platforms to create a more holistic view of social dynamics is required in future studies,” Gong said.

“Traditional methods could also have potential bias in samples — surveys, interviews. Compared to that, social media data is available at a much lower cost, faster, and in a larger amount. It’s also open source, because people share volunteer the data to the public,” he said.

The opportunities for using social media big data extend beyond what has been explored so far.

“I welcome any possible collaboration with other disciplines. Social media data is not just for geographic information scientists or computer science researchers,” Gong said.

“Large-volume, georeferenced, and open-source social media data, as a type of emerging spatial big data, provides an unprecedented opportunity to reveal the spatiotemporal patterns of social dynamics at large scales, which can be used to explore many things in different branches”.