I like statistics. Statistics don’t lie. Out of context, they can mislead, but they can’t lie.
I like stories. Stories create meaning. Out of context, they can mislead, but they are just as impactful.
Unfortunately, stories and statistics are very different approaches and often conflict with each other. Here are a few examples.
Baseball loves statistics. Sabermetrics is the usage of advanced statistics to analyze player performance, which led to the idea of Moneyball. By calculating Wins Above Replacement (WAR) or Batting Average on Balls In Play (BABIP), we can compare players controlling for various conditions and better quantify their performance. On the other hand, there really is something to watching a batter’s swing or seeing a clutch performance in game. Both are approaches to analyzing a prospect’s future potential or a retired player’s hall of fame candidacy.
Charity, fundraising, and non-profit organizations have to convince regular citizens and philanthropic organizations to contribute. They might tell us that there are 5.2 million Americans had Alzheimer’s in 2014. Or maybe they will play Sarah MacLachlan’s “Angel” while talking about animal cruelty. Somehow, we have to be convinced about the saliency of a problem to want to take action.
In my work on web applications, my team is always trying to learn more about our users and what they do. One way we can do it is with analytics by counting how many times users click on this link over a month, or what percentage of our users are from Europe. Another approach is with user testing by looking over a user’s shoulder as they use our application. Analytics provide a complete picture, but they don’t explain why. User testing details a user’s behavior, but it’s just one.
In all of these examples, we have quantitative and qualitative approaches of analysis. Quantitative approaches tend to rely on numbers over a broad sample to appeal to our rational nature. Sadly, we are not very rational. Qualitative approaches tend to rely on a small set of narratives to appeal to our sensitive nature. Sadly, they are empirically not particularly valid.
It’s paradoxical that humans tend not to have good statistical intuitions, largely because of our bias towards causal reasoning. A classic example of bad statistics is in guessing conditional probabilities: we aren’t good at integrating the data together. On the other hand, we tend to look for reasons and patterns behind all sorts of data. In daily life, it’s helpful, but it makes us susceptible to a good story and our desire to see things where there is just chance.
The two ways of thinking aren’t always in conflict: they can be used in tandem. FiveThirtyEight is a data journalism organization that does the work to find good numbers and present them in a digestible format. The good numbers of often statistics, and the presentation puts together a story for us to understand. In Thinking, Fast and Slow, Daniel Kahneman talks about how he uses a classic science journalism format. Each finding begins with an anecdote for the reader to attach to, then transitions into methods and results of the study. It makes the topic both gripping and valid.
This is all very troubling because I tend to see storytelling by nature as a lie to get to a deeper truth. I believe in good quantitative analysis and understanding of randomness. There’s the truth about how the world works. Stories build on top of that. Sometimes, they invent connections that don’t exist in reality. In any case, they affect us as humans deeply and can overemphasize an idea. Playing “Angel” in a commercial is intended to touch us without any regard for the relative importance or impact of ASPCA over any other issue or organization.
Of course, statistics get a bad reputation because some representations can deceive, and excluded data can present a biased perspective. As a whole, however, quantitative analysis is intended to capture representative data. Stories deliberately present limited perspectives.
To ground this entire discussion, my recent interest in storytelling has been very troubling to me because of my preference towards quantitative ways of thinking. I most recently have been biased towards them because of my studies in college: despite being in an interdisciplinary major, I leaned more heavily on engineering and social sciences rather than the humanities. The fact that I barely read fiction in college tells you what I was mostly exposed to.
So it seems like there’s something to stories. Stories are a natural way for us to communicate, whether in conversation, journalism, science fiction novels, or commercials. Although I think my skepticism is probably healthy, stories can evoke responses that even the greatest light bulb moments can’t quite replicate. Besides, I wouldn’t have much of a blog if I didn’t believe in telling stories.