Using Social Media Data to Predict Communication Styles


A Q&A with a UCIPT Postdoctoral Researcher

Based on what you see on social media, how is the 2016 U.S. presidential election affecting the stress levels of college students and college applicants? Is there anxiety about joining the political process?  

That’s an interesting question. I’m not sure how the 2016 Presidential election is affecting the stress levels of college students and college applicants. However, I think we might be able to find out using social media data from platforms such as Twitter. We would need to collect tweets from college students and college applicants for a period of time. We could then apply topic modeling and sentiment analyses to find out how the students feel about the election.

Is there a difference in stress between students who are primarily communicating on mobile devices versus laptops?

We are not aware of any study that compares differences in stress level for those who primarily communicate on mobile devices versus laptops. However, based on the latest report from the Pew Research Center, “stress is not associated with the frequency of people’s technology use, or even how many friends users have on social media platforms. But there is one way that people’s use of digital technology can be linked to stress: Those users who feel more stress are those whose use of digital tech is tied to higher levels of awareness of stressful events in others’ lives.” Therefore, it may be possible that the type of device (e.g., mobile vs. laptop) used may not affect stress level.

How do foreign students in the United States express their stress on social media vs. native-English speakers? Is there data on this subject?  

In order to answer this question, we may need to look at social media platforms beyond Twitter because foreign students may express their stress on a social media platform that is popular in their home country. There have been several studies showing that emotions are expressed differently in different cultures. Therefore, I think that we will find a difference in social media use as well. This will be an interesting area of future research.

Is there a way to demarcate adolescence, puberty, and adulthood using Twitter? How about education level? 

I think there is a difference in how people communicate on Twitter based on their age. People usually will tweet about events that are happening in their lives or things that they care about. In one of our recent studies of tweets made by college students, we found that the type of content posted reflected the events going on in the students’ lives. For example, we found that during the exam period compared with the rest of the semester, academic-related tweets increased, while tweets related to social activities decreased. Therefore, I think we might be able to demarcate age and education level based on the content of tweets.

Does the government collect social media data en masse that would be of value to your research? Aside from Twitter, are there tools that the Freedom of Information Act makes available to you as a researcher?  

There are many public organizations that collect social media data. Twitter also allows researchers to download user data. We have been working with our collaborators at the San Diego Supercomputer Center to collect Twitter data that can be shared with other researchers in the field. The Freedom of Information Act (FOIA) provides researchers with the right to request access to records from any federal agency. This can be extremely helpful because researchers can combine other data sources (e.g., STD incidence) with Twitter data to predict certain outcomes, such as in our HIV and Twitter study. You can review the type of information that is available on FOIA.gov.

Leave a comment

Your email address will not be published. Required fields are marked *