Skip to main content Skip to navigation

ARC WM Blog Content

Show all news items

A New Way to Measure National Happiness

Studies on well-being, happiness and satisfaction are based on direct questioning, which is subject to differences in the way words are interpreted in different languages. An alternative method has been developed by two economists, one from the University of Warwick. [1] This method is based on machine reading of books to pick up the prevailing sentiment in newspapers and other media. The results have face validity because they change according to national conditions. For example, there was a very sharp fall in happiness during the First World War. In the USA there was a sharp decline during the American Civil War in the 1860s. This research does show that happiness varies with per capita GDP. There was, for example, a gradual but persistent increase in happiness over the Victorian era. The method is not, however, perfect. It shows a rise in happiness in both Italy and Germany as the Second World War came towards a close. Likely, this was because the media was tightly controlled by the ruling party and falsely reflected the mood of the population through propaganda. In Britain, there was a marked decline in happiness over the years 1960 to 1980, which then rose rapidly to the point where the data ends in 2008. Neither effect is registered in Germany and the reunification of that country had no effect either way.

Machine learning reading of text, is a powerful method to track changes in attitude over time, and we are currently using this to examine the way slums or informal settlements have been portrayed in the news media over the last 120 years or so.

Richard Lilford, ARC WM Director


  1. Hills TT, Proto E, Sgroi D, Seresinhe CI. Historical analysis of national subjective wellbeing using millions of digitized books. Nature Human Behav. 2019; 3: 1271-5.
Fri 27 Mar 2020, 16:00 | Tags: Machine learning, Richard Lilford, Wellbeing