What can we learn from our community about happiness?

What we've found

  • People who do 30+ minutes of exercise per week report higher happiness levels.*
  • Spending as little as 30 mins outside each day is strongly correlated with higher levels of happiness.*
  • On average, having a healthy BMI is associated with being happier.*
  • A sense of community is important: in particular, even a small amount of volunteering correlates to being happier.*
  • People who monitor at least one aspect of their health in the Evergreen Life app report higher levels of happiness.*
Path
Shape
Path
These insights are based on correlations which are statistical links between two sets of data. While we focus only on more plausible and interesting links, they should not be seen to imply cause or effect.

How happy are we?

How to “get happy” has never been more important. Statistics show that this year, perhaps understandably, average anxiety in the UK jumped to its highest level since records began. And that trend is reflected in our community.

The graph above shows how anxiety levels in our community have changed in 2020 alongside external events that may have contributed to changes in anxiety. Interestingly people’s anxiety seemed to fall after the first lockdown back in March.

What makes us happier?

Research suggests that when you feel in control and able to influence your health for the better, you are happier, healthier and use health services less. We were interested to see if there is a relationship between control and happiness within our community and found that those who monitor at least one aspect of their health report higher levels of happiness.

While many things aren’t within our control, studies also show that we are capable of modifying up to 40% of our happiness by our own efforts. So, what can we all learn from our happier users?  

In the graph above, the green bars indicate higher levels of happiness. So, for example, you can see that having a regular sleep schedule is strongly associated with reporting higher levels of happiness. However we can't say that one causes the other.

BMI and Happiness

Our BMI is one area of our life that we can control, so we looked at BMI and happiness to see what we could learn. Looking at the graph below, it tells us that happiness levels in our community tend to steadily increase for users in the healthy BMI bracket but then begin to steadily decline as BMI rises to overweight and obese levels.§

Statistical Disclaimers:

*Correlation ≠ Causality:

All statistical associations mentioned on this page represent statistically significant (95% level) correlations only. We do not make any claim of causality running in either direction between any two variables discussed, and have controlled for variation in age, sex and the Index of Multiple Deprivation (IMD) decile only.

Anxiety graph:

These graphs use information provided by over 53,000 Evergreen Life users. Each data point uses the latest response for each user that has responded within the 2 weeks prior to each day. The reported percentage is taken as the expected value of the Beta Distribution (https://bit.ly/3kcGWfq) corresponding to the counts for symptomatic, asymptomatic, staying at home, and not staying at home (using a=b=1 as shape parameters), respectively. CIs are taken as the frequencies that correspond to the regularised incomplete beta function taking a value of 2.5% and 97.5%.

Correlates graph:

The displayed correlations are the coefficients from 6 linear probability models of happiness, one for each item, which also control for age, sex, and IMD decile. The happiness outcome variable takes a value of 1 if a user has reported a score of 7 or higher, and zero otherwise. Smoking behaviour pertains to the previous 30 days and exercise is taken to be either moderate or vigorous exercise. All coefficients are statistically significant at the 95% confidence level.

§BMI and Happiness:

This graph uses data from over 18,000 Evergreen Users who have both answered a question about happiness and a recorded value for their BMI. A locally weighted scatterplot smoothing method is used (see https://bit.ly/3pcmdMd) to generate the curve and its confidence intervals (95% Confidence level). All BMI values under 60 are used to estimate the curve, but BMI values over 50 and under 18 are truncated due to small sample size in these ranges, resulting in imprecise estimates that are uninformative.