The Clustering Illusion: Misconceptions About Aviation Risk
March 25, 2025
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
The Clustering Illusion: Misconceptions About Aviation Risk
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
The Clustering Illusion: Misconceptions About Aviation Risk
PUBLISHED
March 25, 2025
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
The Clustering Illusion: Misconceptions About Aviation Risk
PUBLISHED
March 25, 2025
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
The Clustering Illusion: Misconceptions About Aviation Risk
PUBLISHED
March 25, 2025
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
The Clustering Illusion: Misconceptions About Aviation Risk
PUBLISHED
March 25, 2025
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
ARTICLE
The Clustering Illusion: Misconceptions About Aviation Risk
March 25, 2025
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
Aviation Accidents by Month, 2011–2025, number (average)
Source: Monthly Aviation Accident Data by National Transportation Safety Board
The same tendency to see patterns went beyond the randomly scattered points but also real world data too. The graph above shows the aviation accidents by month across years. Without any context, it’s easy to jump into conclusion. We could say that the accidents follow a seasonal pattern, with clear peaks in the summer month (July).
This lead to the theory that because summer is the busiest travel season, with more flights taking off and landing, naturally it leads to higher number of accidents. Which lead to people to believe that July is the month where flights are riskier/prone to accidents. However, without deeper analysis, this apparent pattern might be misleading, much like the first clusters in the first graph. So, is this graph really show seasonality?
In recent months, public concerns has grown over an increase in airplane accidents in the United States and Canada1. Notably, on January 29, 2025, a tragic mid-air crash occurred in Washington DC, involving American Airlines Flight 5342 and a U.S. Army Black Hawk helicopter. the Delta Connection Flight 4819 which overturned on the runway on February 17, 2025. These back-to-back accidents have sparked various theories among the public.
Fatal and Notable Aircraft Accidents: Key Incidents and Casualties in Early 2025
Source: Monthly Aviation Accident Data by National Transportation Safety Board
Because of the short interval between the crashes, some people even speculate that the crashes are part of the larger trend. Online forum discussions have emerged linking the American Airlines Flight 5342 crash to the presence of VIP passengers in the Black Hawk helicopter, suggesting possible foul plays.
More wild claims emerged, such as political group orchestrating the crashes to undermine Trump’s new administration, foreign nations (specifically China) sabotaging US aviation to weaken its economy, and even that such events are deliberate act by intelligence agencies for population control2. They believe that all of these crashes are not just any coincidences but also an indication for something bigger and more serious.
2"Could this situation be an unintended side effect or a deliberate act of management for future population control by the CIA?” "My initial take was a remote-controlled heli deliberately crashed into that plane to, yet again, 'send a message”
Interestingly, this rise of aircraft accidents also coincides with several real-world issues affecting aviation safety. One major factor is the ongoing shortage of air traffic controllers, a problem that worsened during Covid-19 pandemic. Once travel picked up again now, there weren’t enough trained controllers to meet demand. Therefore, all of them work overtime which lead to fatigue, an issue that raises safety concerns.
At the same time, the FAA have been experiencing leadership instability which further slowing safety initiatives. Former FAA Administrator Mike Whitaker stepped down just before Trump’s inauguration, and weeks after the DC crash, Trump’s administration fired hundreds of FAA employees citing that DEI policies are lowering hiring standards in air traffic controller.
Trump’s response to the accidents follows the same pattern of finding connection where none exist. Instead of addressing the staffing shortages, he linked the incident to the DEI policies, linking unrelated factors to fit a political narrative. Just like the conspiracy theorist, Trump framed the accidents as the proof of DEI failures, feeding it into a larger left vs right propaganda war, despite no direct evidence.
Another real-world concerns is the rising air traffic congestion. At Reagen National Airport, where the DC plane crash happened, Congress recently approved more flights despite repeated warnings from lawmakers like Senator Tim Kaine and others. This lead to further speculation that the rise of the accidents happened because lawmakers are deliberately increasing traffic and departures for their own convenience.
People take these real-world concerns and reframe them as evidence of a bigger scheme. However, the reality is more complex. Some of these factors may be correlated and can be combined as the external factors for the accidents, but the more people search for a pattern, the more they connect data points that don’t actually link together.
Note: Diagram airplane accidents in US. Source: Monthly Aviation Accident Data by National Transportation Safety Board. Chart: Instructura
Are the theories about rising aircraft accidents simply a clustering illusion?
While the possible causal factors may be valid to a certain degree, statistically, these airplane crashes are often a random clustering events. Aviation experts, Jason Matzus, argue that the recent Delta Air incidents, specifically, is a “freak accident”3 and that it is statistically random.
3"It’s not unexpected to have two major incidents in a row take place on small aircraft because one-third of commercial planes in the U.S. are smaller jets like the Delta CRJ. It’s unlikely that aircraft issues are to blame”
Our tendency to seek patterns makes us assume that back-to-back incidents indicate a worsening problem, when in reality, they are just expected statistical variation. The aircraft accidents cluster randomly and they create the illusion of an increasing trend even when no systemic issues exists.
In 2014, when airplane accidents seemed to rise, MIT statistics professor Arnold Barnett explained the back-to-back incidents through the Poisson Distribution: shorter intervals between accidents are actually expected from random chance alone, without implying any deeper cause.
“If the average fatal accident rate is one per year, the probability of a crash on any given day is 1/365. The probability of the next crash occurring exactly two days later is (364/365) × (1/365), illustrating how random intervals follow an exponential distribution”
Distribution of Days Between Consecutive Fatal Aviation Accidents, probability density visualization
Source: MIT Statistic Professor Arnold Barnett
The Role of Media Amplification
Media coverage makes the clustering illusion even stronger. When dramatic accidents happened, it’s all over the news, making rare events feel like they are happening all the time. Sensational reporting about the airplane crashes fueling fear about aviation safety. In February 2025, Google searches for “Is it safe to fly” hit the reached levels second only to those during Covid-194.4Google Trend Data
Just as people today believe airplane crashes follow a deliberate plan, Londoners assume bombings during WWII were strategic. Newspapers during that time, published bombing maps without further context, which made people see patterns that really weren’t there. The same thing happened in today’s airplane accidents, news outlet focus on clusters of accidents without showing the statistical pictures, making random events seems connected.
However, media framing can be improved by reducing the sensationalization of isolated events. In 2023, media coverage of road traffic crashes in Colombia shifted from treating accidents as isolated incidents to providing deeper context about risks and contributing factors. This change helped reduce misinterpretations and align public perception with reality rather than illusion. If U.S. aviation accidents were reported the same way, discussions might shift from panic and conspiracy theories to real concerns like safety improvements, FAA staffing shortages, and air traffic congestion.
Fear of flying is often heightened by sensational news coverage, even though data shows that plane crashes are far rarer than road accidents. Photo by Markus Winkler
Clustering Illusion in Financial Data
Just like with aviation accidents, people also see patterns in financial markets where none exist. Investors often mistake random market movements for meaningful trends, falling for clustering illusion. This leads to overconfidence, excessive trading, and speculation based on perceived patterns rather than real economic fundamentals.
Psychological factors such as recency bias, the tendency to give more weights to recent events, usually influence investor’s interpretation. If a stock price surges for a few weeks, investors assume it will keep rising, even if there’s no fundamental reason. Herding behavior, when people see others acting on a pattern, they follow those crowds, also reinforce the illusion.
Similar to the aviation accidents, where random clusters of accidents spark conspiracy theories, random clusters of stock price movements lead investors and even governments to see trends that aren’t really there. And just like in aviation accidents, recognizing these illusions can help avoid costly mistakes.
Why do people assume intentional causation?
Whether it is airplane crashes or stock market trends, people tend to assume patterns have a clear clause. They instinctively look for explanations even when randomness is at play, leading to illusory causality. In finance, this can be costly, as investors ignore other factors like interest rates, inflation or economic conditions. For example, if a stock rises after five straight earnings reports, an investor might assume “good earnings cause stock increases,” when in reality, macroeconomic trends, investor sentiment, and algorithmic trading might play a bigger role.
Just as people falsely linked airplane crashes to political conspiracies, many investors saw rising and falling stock prices as a sign of economic strength and ignoring deeper market forces. These misjudgments drive poor financial decisions, just as clustering illusions in aviation accidents can fuel unnecessary panic. When people are in fear, they interpret causal links between random events projecting negative outcomes and interpretation.
How to avoid the clustering illusion
Whether in airplane crashes or financial trends, separating real trends from random patterns is the key to make sound decisions. In the financial world, analysts use statistical tools like hypothesis testing, Monte Carlo simulations, and volatility models (ARCH/GARCH) to differentiate meaningful trends from short-term trends.
Beyond statistics, critical thinking and bias awareness also play a crucial role. Recognizing biases like recency bias, confirmation bias, and overconfidence helps investors and policymakers question whether patterns are meaningful or just random. When people understand how randomness works, they’re less likely to assume patterns must have a cause.
And the same logic applies to the aviation accidents. Just because the crashes cluster in a short period doesn’t mean there’s an underlying trends. Before jumping to conclusion, we need to step back and considered other external factors.
So, is the first graph show seasonality then?
To answer this questions, we used the aviation accidents records in the US from 2011 to 2025. The data also includes all reported aviation accidents, without filtering severity or aircraft types. We used time series decomposition to break the data into trend, seasonality, and residual components. This helped separate long-term changes from short-term fluctuations and recurring seasonal patterns. We analyzed the seasonality component to check for consistent monthly trends across years.
Decomposition of Seasonality, Trends, and Residual of Aviation Accident Trends, 2010–2025, index
Source: Monthly Aviation Accident Data by National Transportation Safety Board
Seasonal Pattern of Aviation Accidents,average effect by month
Source: Monthly Aviation Accident Data by National Transportation Safety Board
The data shows a clear seasonal pattern, with aviation accidents peaking in July and reaching their lowest in January. This pattern repeats across multiple years, confirming strong seasonality. The trend component shows fluctuations, such as a drop in 2020, likely due to reduced flights. The residual component shows minor random variations but does not disrupt the overall seasonal trend. While accidents are higher in summer, this does not necessarily mean flying is riskier in those months—higher flight volume, weather, may be the main factor.
Seasonal Pattern of Aviation Accidents,average effect by month
Source: Monthly Aviation Accident Data by National Transportation Safety Board, US Commercial Air Traffic and Fare Report: July 2024
We analyzed, then, how those flight volumes affect accident trends using correlation analysis, In the result, we found a moderate positive correlation (0.60) between monthly passengers and aviation accidents. More passengers usually means more flights, which can lead to more accidents. However, this relationship is not perfect as passenger volume alone cannot determine accident rates.
Importantly, a rise in reported accidents doesn’t always mean more severe crashes. If the data includes minor incidents like runway skids or technical issues, the number might increase without real rise in serious accidents.
These patterns can create the illusion that certain months or higher passenger volumes make flying riskier, but correlation doesn’t imply causation. The clustering of accidents may be driven by external factors like higher flight volumes or reporting differences. That’s why it’s important to look beyond numbers and understand the bigger picture instead of jumping into conclusions.
ARTICLE
The Clustering Illusion: Misconceptions About Aviation Risk
March 25, 2025
People naturally look for pattern in randomness. That’s just how our brain works, so when we see clusters of events or data points, we instinctively try to connect them, assuming there’s an underlying reason for the grouping. This tendency came from cognitive bias that can lead to: Clustering Illusion.
Source: Monthly Aviation Accident Data by National Transportation Safety Board
For example, the graph above shows randomly generated points. But at first glance, certain areas seems to have more clustering than others. Our brain wants to believe there’s a pattern somewhere in this sets of data and theorize that there’s something that caused these points to grouped together. But it’s only a random data.
.png)
.png)
.png)
.png)
