"TRENDS IN GLOBAL HAPPINESS
• Nordic countries lead the happiness rankings once again. Finland is still in a group of one at the top, followed by a group of three: Iceland, Denmark, and Costa Rica. Sweden and Norway complete the top six, followed by the Netherlands, Israel, Luxembourg, and Switzerland to round out the top ten. Costa Rica’s rise to 4th marks the highest ever ranking for a Latin American country.
• Looking at changes in happiness from the 2006–2010 base period to 2023–2025, nearly twice as many countries have had significant gains (79) than significant losses (41), among the 136 countries ranked.
• Within that period, most of the 21 countries that have gained a point or more on the 0–10 life evaluation scale are in Central and Eastern Europe, reflecting the convergence in European happiness levels that has been clear for more than a decade. Most of the eight countries with drops of more than one point are in or near zones of major conflict.
• In general, most Western industrial countries are now less happy than they were between 2005 and 2010. Fifteen of them have had significant drops, compared to four with significant increases.
• In a ranking of happiness changes for under-25s, the United States, Canada, Australia, and New Zealand (the NANZ region) rank between 122 and 133 in the list of 136 countries.
• In eight of the ten global regions shown in Figure 2.2 (covering roughly 90% of the world’s population), those in the youngest age group have higher life evaluations now than in 2006–2010, either in absolute terms or relative to those over 25. Youth wellbeing has only fallen in the NANZ countries and Western Europe, both absolutely and relative to adults.
• Negative emotions are becoming more common in all global regions. They are less frequent for the young, except for NANZ, where the increases, relative to other regions, were in sadness. Worry rose more broadly for the young, while the frequency of anger fell everywhere, for young and old alike.
• Positive emotions continue to be twice as frequent as negative emotions. In all regions, they are higher for those born recently than for their elders, except for NANZ, where that gap has closed in recent years.
HAPPINESS AND SOCIAL MEDIA
• Life satisfaction is highest at low rates of social media use and lower at higher rates of use, according to data from the Programme for International Student Assessment (PISA) covering seven internet activities for 15-year-old students in 47 countries (but not the NANZ countries, unfortunately).
• Internet activities fall into two groups. Communications, news, learning, and content creation are associated with higher life satisfaction. Social media, gaming, and browsing for fun are associated with lower life evaluations.
• All internet activities are associated with lower life satisfaction at very high rates of use, especially for girls and for those in the UK and Ireland, the two English-speaking countries in PISA.
• Data from Latin America reveal that the type of platform is crucial. Platforms designed to facilitate social connections show a clear positive association with happiness, whereas those driven by algorithmically curated content tend to demonstrate a negative association at high rates of use.
• When combined, the PISA and Gallup data show significantly higher youth wellbeing in countries that spend more time using the internet for communication, and insignificantly lower youth wellbeing in countries with higher average hours of social media use. Latin American countries combine high levels 6 World Happiness Report 2026 of social media use with high youth wellbeing, while English-speaking countries show lower youth wellbeing than their fairly typical patterns of internet use can explain.
• The prevalence and use patterns for social media are similar in all countries, but their links to life evaluations are stronger in English- speaking countries and Western Europe. These differences do not account for the large drops in youth wellbeing in these countries, however. Much, no doubt, depends on how social media is used and the broader social and economic context.
• When school belonging goes from low (10th percentile) to high (90th percentile), the life satisfaction gains for girls in the UK and Ireland are four times greater than social media use going from high (90th percentile) to low (10th percentile). In PISA’s 47-country global sample, the belonging effect is six times larger. This should be relevant when policy options are being considered.
SOCIAL MEDIA IS HARMING ADOLESCENTS AT A SCALE LARGE ENOUGH TO CAUSE CHANGES AT THE POPULATION LEVEL
• Is social media use reasonably safe for children and adolescents? We call this the “product safety question”, and we present seven lines of evidence showing that the answer is no.
• The evidence of harm is found in: 1) surveys of young people; 2) surveys of parents, teachers, and clinicians; 3) contents from corporate documents; 4) findings from cross-sectional studies; 5) findings from longitudinal studies; 6) findings from social media reduction experiments; and 7) findings from natural experiments.
• We show there is now overwhelming evidence of severe and widespread direct harms (such as cyberbullying and sextortion), and compelling evidence of troubling indirect harms (such as depression). Furthermore, we show that the harms and risks to individual users are so diverse and vast in scope that they justify the view that social media is causing harm at a population level.
• We further argue that when these lines of evidence are considered alongside the timing, scope, and cross-national trends in adolescent wellbeing and mental health, they can help answer a second question: was the rapid adoption of always-available social media by adolescents in the early 2010s a substantial contributor to the population-level increases in mental illness that emerged by the mid 2010s in many Western nations? We call this the “historical trends question”. We draw on our findings about the vast scale of harm uncovered while answering the product safety question to argue that the answer to the historical trends question is “yes”. (Executive summary, pages 6-7)