In an earlier post, I described how Twitter can be used to measure happiness and predict the future. The Hedonometer is a tool that’s used for this exact purpose by analysing 10% of Twitter’s user data. With it comes a long list of implications for public wellbeing. This process is called sentiment analysis.
With that said, I’d like to dive into some of the findings that came from the Hedonometer team. They might not be applicable to improving public wellbeing, but they still offer interesting insight into how accurate Twitter reflects our emotions.
All languages are happy
In a paper published by the minds behind the Hedonometer, it says, “The most commonly used words of 24 corpora across 10 diverse human languages exhibit a clear positive bias.” In laymen’s terms, humans created more happy words than sad words.
Could this point to humans being naturally inclined to express happiness? I’d like to think so.
Still, some languages are “happier” than others, Spanish being the happiest while Chinese is the most negative. This doesn’t surprise me one bit. I’ve been to many Spanish speaking countries, and recently spent a summer in Shanghai. While I’m not fluent in either of the languages, I still noticed a significant difference in the tone of voice between Chinese and Spanish.
My experience isn’t exactly scientific evidence, but it’s worth noting that we can now test our presumptions with real data.
Twitter = approval ratings (pretty much)
Twitter proved to be a GREAT source for gauging approval rating for Barack Obama. In a another paper by the Hedonometer team, they explain how unsolicited public opinion polls correlate very well with their Twitter sentiment analysis. In fact, their program was able to predict President Obama’s job approval three months in advance.
Imagine how useful this approach would have been when measuring approval ratings in the 2016 election. The mere surprise people faced when Trump was elected was huge contributor to the bedlam that arose (and continues to arise). Given it’s past success in predicting outcomes, sentiment analysis could have given the public a better idea of Trump’s chances of winning.
People send mixed messages
This part is particularly interesting. A while ago, economists performed a study in order to see how happy people were compared to other age groups. It worked like this: once enough people agreed to participate in the study, they were asked to report on how happy they were – not just once, but periodically as time moved forward.
Statisticians attempted to conduct the same research, but with an entirely different method. This time, it was sentiment analysis conducted on Twitter that lead to identifying happiness by age. The results showed an opposite trend of the economists’ findings. See below:
It appears that when we’re asked to intentionally report on how we are doing emotionally, our answers conflict with our projected emotions online. This could be for a variety of reasons.
First of all, the sheer interruption that takes place in order to get a survey answer out of someone could be enough to put that person in a worse mood. Second of all, our preferences and priorities can be time sensitive.
So if I’m currently falling behind an important deadline and I’m all of a sudden asked to report on how I’m doing, my answer might carry the angst I have towards the project. If I’m asked the next morning once I’m rested and relieved, I may have a different outlook.
Saturdays are the happiest
No surprise there. Sentiment analysis through Twitter shows that we produce the most positive words and ideas on Saturdays. Meanwhile, Tuesdays account for most people’s dismay.
I always knew Tuesday were cursed.
‘Make America Great Again’ was genius
What does this chart tell you?
For starters, it shows that we love our holidays and we hate our massacres – shocker. What’s most important here is that our satisfaction goes up and down like a seesaw, continuously.
Let’s break down one of the most popular sentences of the past couple years: make America great again. Donald Trump won a completely unexpected election off of the promise that he was going to make America great again. But should his victory really have been so unexpected?
The phrase is basically saying that life used to be happy, we’re currently unhappy, and we’ll soon be happy again. It would be hard to argue against the idea that Trump’s slogan perfectly caters to the way we naturally think and feel. What’s ironic is that Trump announced his campaign in June of 2015 which, as you can see from the chart, was during a peak period of happiness.
Considering all of this, we’re left to ponder the notion that we might not always feel the way we think we feel. Or, maybe do. The beauty is that there are so many kinds of data point out there, making it easier for us to learn about our own behavior. Twitter is just one source.