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

HSE Researchers Learn to Determine the Level of Happiness of Russians by Posts in Social Networks

Research to gain ‘likes’

ISTOCK

Researchers at the HSE Graduate School of Business have created a model for calculating the indicator of subjective well-being, based on the analysis of 7.2 million posts on the Odnoklassniki social network. They found that the lowest level of observed subjective well-being can be registered in the morning, and the highest can be found in the late evening. The results of the study were published in the journal ‘Mathematics’.

Subjective well-being is one of the most important sociological indicators, allowing politicians and researchers to assess people's reaction to social changes and better understand people’s needs. To measure this in recent years, researchers have started to apply new approaches, using digital footprints as the main source of information. According to scientists, such methods have the potential to overcome the shortcomings of traditional surveys.

In this new work, the researchers decided to analyze the observed subjective well-being, that is, how people themselves assess their standard of living and satisfaction with their conditions. Text posts by Odnoklassniki users were used as empirical material, since this social network is quite representative of the entire Russian Internet audience.

All posts were openly published by users on their own pages. The messages were randomly selected and were completely anonymized. However, the date of birth and gender of the author, the time of publication, the country of posting and time zone, and the text and language used in the message were taken into account. To analyze the text tonality, several machine learning models for natural language processing, trained on the RuSentiment dataset, were used. The use of demographic information and post-stratification methods made it possible to make a data sample representing the entire population of the country according to the selected demographic characteristics.

An analysis of 7.2 million messages showed that users of the social network have certain mood patterns. Thus, certain general daily fluctuations are clearly visible: the lowest level of happiness was recorded in the morning, and the highest could be found in the late evening. The authors took into account multiple time zones and analyzed the local time for each time zone.

Weekly patterns were also clearly observed. As expected, weekends turned out to be happier than weekdays, with the lowest level of happiness falling on the first three weekdays. Starting on Thursday, the mood level starts to improve and reaches its peak at weekends.

Russians feel the happiest during the night from Saturday to Sunday.

These weekly patterns can be expected intuitively, as they are usually associated with cultural differences between weekdays and weekends in modern societies that regulate social practices and behavior. Similar results were obtained for other countries in the framework of both traditional sociological research and research based on digital footprints.

The study also shows that growing up makes people unhappy. The older a person becomes, the less happiness they tend to experience. This trend is typical for both men and women, although in general, women feel happier compared to men. However, the authors note that this may be due to the fact that men and women are differently inclined to show positive emotions online.

The results of the observed subjective well-being in general turned out to be similar to the happiness index calculated by the Russian Public Opinion Research Center, which confirms the validity of the approach. In the future, we will be able to conduct a more detailed analysis of emotions. For example, instead of the classical categories of ‘positive’ and ‘negative’, we could consider happiness, sadness, fear, disgust, anger and surprise.

Sergey Smetanin
Author of the research, Doctoral Student at the Department of Business Informatics

IQ