It was maybe a month or two ago that showed that studies like these - all usually done as surveys - are highly susceptible to changes in input from users, just by word choice. In their example, they showed that if you use terms like ‘mental health’ while surveying people, a discrepancy emerges between non-conservatives and conservatives. This discrepancy disappears when you instead ask other questions, such as ones about sleep, worries , and fears.

So I don’t really know if I overall trust the data on happiness overall.