• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

What Students’ Membership in Social Media Groups Can Tell about Their Academic Outcomes

A recent study by IOE expert Ivan Smirnov shows that students’ academic achievement can be predicted from their ‘digital traces.’

study carried out by Ivan Smirnov, Head of the IOE Laboratory for Educational Data Science, has found that high-schoolers’ membership in particular social media groups allows judging about their academic achievement. A computational model that analyzes students’ PISA scores as well as their ‘digital traces,’ i.e., what sort of communities they are subscribed to on networking sites, is able to accurately differentiate between straight C-graders and high-performers.   

Social Media Data and Students’ Learning Outcomes

The study argues that how teenagers perform at school can be reliably determined by looking at their digital traces. Ivan has shown that upper-graders with higher academic achievement are more likely to follow public pages about science and art, while lower-performing teenager will prefer social media subscriptions related to humor and horoscopes.   

How learning outcomes are differentiated depending on the sort of subscription preferences that students have for social media communities has been analyzed on the basis of the Russia-wide longitudinal study, Trajectories in Education and Careers, that tracks over 4,000 PISA participants and also provides data about their activity on VKontakte, Russia’s most popular social networking site. About 4,500 public pages on VKontakte have been selected for analysis; the average number of community subscriptions per sample user was 54. 

OECD’s PISA Programme is a large-scale study that tests the skills and knowledge in mathematics, reading and science of 15-year-old students. While students’ membership in social media communities gives a good idea about what kind of preferences they have in real life, this data may not be considered as exhaustive to judge about the full range of interests that schoolers have. Moreover, social media activity often leaves no digital traces at all as it is possible to follow certain public pages without being subscribed to them. Nevertheless, students’ accounts still enable obtaining a meaningful understanding about their academic and life contexts, and this domain holds a huge potential for educational research which is yet to be unlocked.   

Mapping out Student Profiles through Digital Traces

How schoolers behave in the digital realm, including the kind of comments and photos they post, etc., indeed represents a deep well of personality-specific information that allows researchers to gauge one’s way of life and multiple psychological features.

For example, analyzing a person’s tweets, profile data, visual feeds (photos of the neighborhood, etc.) and the language of posts makes it possible to plausibly determine their socio-demographic attributes (ethnicity, gender, income level, etc.).   

Similarly, much can be understood about one’s personality qualities and intelligence by tracking the individual’s behaviors in the networking space. Specifically, social media ‘traces’ have proved to exhibit a connecting thread with academic outcomes. “We have built a simple model that predicts PISA scores based on students’ subscriptions to various public pages on the social network,” the study author Ivan Smirnov explains. As a result, a clear correlation was established between the students’ preferences for social media subscriptions and how they perform at school.  

Analyzing how students’ actual PISA results dovetail with the score estimates obtained from their ‘digital traces’ on social media has shown high precision of the developed computation model. Thus, the model’s ability to accurately discriminate between high- and low-performing students is 90% for mathematics, 92% for science, and 94% for reading.  

The Signal Function

While various types of social media communities share basically the same principles of how their content is structured and presented, there is a strong division in terms of which audiences choose to follow which topics and thematic areas that particular public pages address.    

Ivan’s study has shown that the schooler cohort is strongly segregated by the types of social network communities that students are likely to join, which in turn is linked to their academic achievement.   

Thus, low-performers have been found to prefer social media communities that are related to horoscopes and humor. By contrast, those demonstrating strong academic results are likely to follow public pages about science, technology, arts, and cinema. 

It might be conjectured that high-performers’ media preferences should add up to their intellectual development and achievement at school, however, no evidence has yet been obtained to warranty this assumption. “So far, we are unaware of whether students’ subscriptions in any way impact their formal achievement, but I tend to believe there is unlikely to be any material connection found. There are factors way more powerful which directly determine how one is doing at school. These include, for instance, a person’s inborn abilities, the family resource, the quality of education at a given school, etc.,” Ivan Smirnov comments.  

There are two specific points that Ivan makes in connection with the study findings. First, the author stresses that those who show better academic results are also likely to use the internet as an important source for self-development. By contrast, children with lower performance, who could potentially achieve academic improvements by resorting to adequate social media content, will typically confine their subscriptions to horoscopes and junk fun. Secondly, a closer consideration of various media communities shows that virtually none of them can be identified to feature learning content proper. This may be suggesting that how students are divided by their subscriptions on social media is more likely to be attributable to the so-called ‘signal function’ reflecting how identity formation is taking place and the fact that the subject subscriptions are public.      

Total Segregation

The study indicates that the schooler sample is segregated at least in terms of the following three facets:

  • The two student groups exhibit starkly different interests
  • There is a clear division by academic performance
  • Schoolers tend to perceive and use the internet in differing ways

The above findings dovetail with what multiple other studies in digital inequality have reported. In particular, it has been shown that people with less education will primarily use digital resources for entertainment, whereas their more educated counterparts tend to make the most of the information & knowledge potential that the World Wide Web affords.     

According to Ivan Smirnov, the difference in knowledge and skills between the two groups of students is equivalent to up to two years of formal schooling. If we were to apply the OECD methodology, where 40 points on the PISA framework translate into one year of schooling, then the educational attainment gap between the students who follow world art & culture groups and those preferring love horoscopes could be measured as two grades of schooling (the difference in PISA scores between the two groups is 79–88 points).