Complaining about the weather is as synonymous to Britain as Wimbledon or Glastonbury. Indeed the weather often takes centre stage at both events. As we pass the half way point of summer and start to count the number of good days that might still be left headed into the autumnal months, it occurred to us that our happiness might depend on meteorological factors such as the temperature. Clearly the weather affects many aspects of our daily lives but does it change our state of mind?
Google the question “does the sun make us happier” and there are a plethora of articles – each less decisive than the last. The BBC asked “Does sunshine make us happier?” and concluded that though the British public is convinced living in a warmer, more sunny climate would make it happier, it is likely not the case. The Guardian then ran an article playfully titled “Sun + People = Happiness” to investigate the same conundrum. It cited one study that showed depression in the Netherlands is highest during summer and autumn months and another which worryingly reasoned that aggression levels are elevated.
Then there are the increasingly common instances of Seasonal Affective Disorder – which the NHS describes as ‘a type of depression that has a seasonal pattern. The episodes of depression tend to occur at the same time each year, usually during the winter’. It is estimated to affect up to 2 million people in the UK but little is understood about the causes of SAD. The NHS suggests changes in production levels of melatonin and serotonin, and the body’s internal clock may be causes.
The national mood with regards to weather is equally difficult to pin down. It is often anecdotally said that the British yearn for the sun when it is rainy and the rain when it is sunny. To shed some light on this we will use the wealth of public social media data available to investigate whether the general sentiment of conversation improves with the weather.
Are people happier in the summer months?
To determine whether people are happier in the summer months we took a large sample of UK mentions from across the last 12 months (starting August 14 – July 15) and ran them through a sentiment analysis code which scores mentions between -1 (most negative sentiment) to +1 (most positive sentiment). The sample was comprised of 60,000 randomly selected Twitter mentions; 5000 for each month.
Fig 1. Sentiment score versus average monthly sunshine hours
The sentiment analysis here is inconclusive. Although sentiment is highest in June and July – two of the months with the most sun in the UK – August is the month with the lowest score. Overlaying the UK average hours of sun (Met Office, 1981-2010 averages) reveals sentiment does not neatly follow weather on a macro scale. Overlaying average high temperatures in the place of average sunshine hours reveals no significant difference since the two tend to be correlated.
Fig 2. Sentiment score versus average monthly temperatures (°C)
Are people happier when weather beats their expectations?
Perhaps we have expectations that are built in. For instance, we may think a 10°C day during the height of summer to be worse than a 10°C day during winter due to pre-existing expectations of what the weather should be like. It might be that we are happier when weather exceeds our expectations – for instance, a day in March that hits 20°C. Conversely we might expect to see a significant dip in sentiment on days that are notably colder, windier or with higher rainfall than expected.
To analyse whether days that beat or miss our expectations by a noticeable amount sees a change in sentiment we selected the hottest and most pleasant days of the year, along with those with the worst weather highlighting stand out days throughout the last 12 months and which were broadly the same weather across the country to allow us to use a UK data set throughout.
Fig 3. Sentiment of days with above average temperature versus average sentiment of corresponding month
Fig 4. Sentiment of days with below average temperature versus average sentiment of corresponding month
The two charts show the sentiment score for outlier days - where weather was better or worse than is typical for that time of year – plotted against the average sentiment score of the month the outlier day was. It appears that especially cold, unpleasant days correlate with reduced sentiment versus the average, while the results for the above average days were relatively inconclusive.
Despite the sample being relatively small – we purposefully restricted the data to a small number of dates in order to show only the extremes in what is a relatively temperate and consistent UK climate – a pattern has emerged and further investigation may be warranted to confirm this.
Our investigations set out to investigate whether our happiness is in a large way determined by the weather we live in. Irrespective of other factors, does weather impact the national psyche enough to make the population noticeably cheerier?
Taking in a wealth of previous studies that have argued for and against this causal link, this piece looked at, for the first time, whether there is a demonstrable correlation between the weather and our happiness on social media, as measured by a sentiment score. Over the last year the mentions collected and analysed revealed there to be no positive correlation between average monthly temperature and sentiment. Furthermore, the monthly average hours of sunshine did not appear to be linked either.
Recognising a warm day in March might drive positive sentiment relatively more than a warm day in June because our weather expectations adjust throughout the year based on seasonal patterns, we assessed whether unexpectedly good or bad days drew an associated change in sentiment levels. The sample was inconclusive for typically pleasant days but did demonstrate a considerable reduction in sentiment on those days tagged with bad weather.