I was recently teaching an undergraduate class on complexity theory. After talking about the impossibility of forecasting future outcomes, a student asked: “What about statistics? Don’t big aggregate numbers, percentages, and trends give us the ability to make justifiable inferences and plans for the future?”
My response was another question to the entire class: “How many of you would predict that I will not find myself homeless one day?” Most of the hands went up. I asked them why they felt confident in their predictions:
Economics and social status: since you are teaching at a university, the odds are exceedingly high that you enjoy extra levels of financial stability and fallbacks based on your class position.
Fair assessment, to be sure. In fact, to bolster their case, I disclosed a few more pieces of personal information: I told them that I did not suffer abuse or neglect when I was young, which strongly works in my favour (as adverse childhood experiences are strongly correlated with poverty). I grew up in a profoundly loving and caring family. I tend to possess average to good mental health. To top it off, I confessed that I don’t struggle with significant addictions to anything outside of vices generally approved by society – like caffeine and Twitter.
Polled again on the question, the class was feeling even more confident in their prediction that their instructor would not land in a state of desperate poverty.
“I’m glad you are so certain about my future,” I said. “I am not so confident.”
We began discussing hypothetical scenarios: what if I got hit by a bus when I leave class and suffer a debilitating injury? What if my spouse or children passes away tragically and I don’t recover from the grief? What if I am in an accident and survive with an acquired brain injury? It does not take a lot of imagination to cook up scenarios – all of which are real and possible – that could send my life into a spiral of depression, escapism, substance use, gambling, and into any number of narratives we heuristically associate with homelessness.
The point of the exercise is to make a simple point: while patterns are powerful and statistically significant, they are useless when applied to individuals.