Human beings have an innate tendency to recognize patterns, a trait that has likely provided evolutionary benefits throughout our history. However, in today’s information-saturated world, this instinct can lead us to erroneous conclusions. Let me illustrate this with a personal experience and explore its implications for data analysis.
On June 18th, while wandering the west side of Vancouver, I stumbled upon a street named “Trutch.” It struck me as an unusual name. Shortly thereafter, Tyler Cowen shared an article from the Vancouver newspaper discussing the recent renaming of this street:
Vancouver’s Trutch Street is now šxʷməθkʷəy̓əmasəm Street. Not everyone is happy
Dan Fumano: Many residents of the street, formerly known as Trutch, expressed support for the name change but voiced concerns regarding the practicality of a street name exclusively in a non-English language.
Author of the article: ByPublished Jun 17, 2025, Last updated Jun 18, 2025
Now that’s an even stranger name!
Interestingly, the timing of the name change coincided precisely with my observation of the street. What a remarkable coincidence! But wait, there’s more. On the following Monday, I spent a night in a Calgary hotel before heading back home. The next morning, I checked Marginal Revolution, and the first post that greeted me was titled:
Calgary is resuming with fluoride, and Quebec fact of the day
This seemed like an even more improbable coincidence. It almost felt as if Tyler, unbeknownst to me, had crafted a post to align with my travel itinerary. Of course, that’s absurd; he had no idea I was on vacation. Yet, it’s easy to see how someone with a superstitious bent might attribute significance to such coincidences. What are the odds?
You might argue that in a world where billions of events transpire daily, a coincidence lacks substantial meaning. However, much of scientific and social scientific research operates under the assumption that coincidences carry great significance. In the realm of physics, for instance, scientists often demand coincidences of extraordinary rarity, termed “5 sigma events,” which are more than five standard deviations from the expected value. In contrast, many other disciplines accept a much lower threshold for significance, often just two standard deviations from the null hypothesis. This means that random coincidences with a mere 20-to-1 odds against are considered highly meaningful.
In a recent post on EconLog, Kevin Corcoran discussed the topic:
In 2007, Eliezer Yudkowsky penned an intriguing article advocating for what he termed “defying the data.” The concept is straightforward: suppose you have a theory explaining how the world operates, and then a new study emerges with data that your theoretical framework cannot accommodate. How should you respond?
One option is to discard your theory in favor of the new data. Alternatively, you might retain your theory and, as Yudkowsky suggests, “attack the experiment” — accuse the researchers of dishonesty, flawed design, or conflicts of interest. A third possibility is simply to defy the data. . . .
If a theory has been rigorously established and supported by multiple studies, then a striking appearance of contrary data shouldn’t carry much weight.
At first glance, this may seem unscientific. Yet, in practice, it often turns out that evidence that purportedly refutes a given theory is merely a commonplace coincidence—something that occurs on a daily basis.
Even the most astute analysts (including myself) can fall prey to the allure of coincidences. A particularly glaring example emerged during the discourse surrounding the origins of Covid. Historically, pandemics frequently initiate in major cities in southern China, where densely populated communities coexist with wild animal markets. This was precisely how the initial SARS outbreak began in November 2002. The Covid pandemic (SARS-2) also appears to have originated under similar circumstances, emerging from a wild animal market in one of southern China’s largest cities.
Despite this, numerous commentators have clung to the baseless theory that Covid originated from a lab leak, primarily because it first manifested in a city housing a significant virus research institute. This is one of the weakest coincidences I’ve encountered, yet many perceive it as compelling evidence for the lab leak theory. In stark contrast, the animal market theory is rooted in a coincidence that is statistically far less likely to occur by chance.
There are millions of streets globally. The fact that I noticed a specific street in Vancouver right before Tyler posted about it is indeed a striking coincidence. And for Tyler’s fluoride article to appear just 10 days later, right as I was in Calgary, is an even more remarkable coincidence. By comparison, the entire lab leak theory hinges on little more than a flimsy coincidence, akin to rolling the same number twice in succession on a standard six-sided die.
