Correlation and Causation: Patterns, Stories, and the Ways We Get Them Wrong

(HH Foundational Essay – Original, Part of the Foundational Series)

Introduction: The Stakes of Confusing Patterns with Causes

On a hot summer day, ice cream sales soar — and so do drowning deaths. One line goes up, the other line follows. A pattern is there, plain as day. Does one cause the other? Of course not. But it takes discipline to resist drawing a causal arrow. The human mind loves stories, and stories are made of causes. Now shift the lens: scientists once noticed that stomach ulcers tended to appear alongside a spiral-shaped bacterium. For decades, ulcers were explained away as the result of stress and lifestyle. Only when two researchers proved that Helicobacter pylori actually causes ulcers — even going so far as to swallow the bacteria themselves — did the medical community concede that the correlation was causal. Between these two examples lies a landscape of human error, discovery, delay, and surprise. Correlation and causation are not just abstract concepts from a statistics class; they shape medicine, policy, economics, and our everyday choices. Mistaking one for the other can waste billions, cost lives, and deepen mistrust. Getting them right can save both. This essay traces the distinction across fields, highlights cases where confusion changed history, and asks what our mental apparatus is really wired to do with patterns and causes.

Part I: What They Are and How They Differ

At its simplest, correlation means two variables move together. They may rise or fall in sync, or one may increase while the other decreases. Correlation says nothing about why; it only describes the shape of the dance. Causation, by contrast, claims a mechanism. It says not only that X and Y are linked, but that X makes Y happen. A spark lights kindling, which starts a fire — that’s causation. The two are related but not interchangeable. Correlation can be a clue, the first whisper of a relationship worth exploring. But correlation can also be misleading, driven by a hidden third factor (summer heat raises both ice cream sales and swimming, which in turn increases drowning). Philosophers and scientists have struggled for centuries with how to tell the difference. David Hume warned that causation cannot be directly observed — only patterns of constant conjunction. Modern science has built tools to probe beyond the pattern: randomized controlled trials, natural experiments, mechanistic explanations. But in daily life, we still leap from “these two things go together” to “this causes that” — often faster than reason can catch us. One way to see the distinction is through a thought experiment: - Every morning, a rooster crows before the sun rises. The events are correlated. - Does the crowing cause the sunrise? No. - But suppose you block the rooster’s view, and it still crows, and the sun still rises. No mechanism connects the two. The correlation remains, but causation is absent. Sometimes, though, causation is accepted even without a known mechanism. Smoking caused lung cancer long before scientists traced DNA damage. Other times, causation is resisted until a mechanism is clear, as with ulcers. And in some cases, the mechanism may be forever beyond us, as with quantum entanglement. The human mind is both equipped and limited: built to see causes, but often uneasy when they remain hidden.

Part II: Cross-Field Examples

Patterns that invite causal stories show up everywhere. From hospital wards to trading floors, from classrooms to sports fields, the temptation is the same: see a line rise and fall, then reach for a cause. Medicine Doctors once noticed that people who drank coffee seemed less likely to die from certain diseases. Was coffee protective, or were coffee drinkers different in other ways? For decades, it was unclear. Only later did research reveal partial causal mechanisms — compounds in coffee can reduce inflammation and improve vascular health — though lifestyle differences also played a role. Contrast that with insulin: inject it, and blood sugar falls. The link is direct, causal, and demonstrable in every patient. But medicine has also shown how profit can amplify or delay causal claims. Pharmaceutical companies may promote correlations that suggest benefits for a new drug, while downplaying side effects until causal mechanisms are undeniable. Economics Stock markets often rise when consumer confidence is high. Are confident consumers driving stocks upward, or are markets buoying optimism? Correlation doesn’t answer the question. By contrast, tariffs provide a clean causal line: when governments impose them, import prices go up. Education Students who do more homework tend to get higher grades. But is homework itself the cause, or do motivated students simply do both more homework and more learning? That’s a correlation problem. A clearer case comes from lead exposure: children in classrooms near highways or old paint performed worse, and when lead was removed, test scores rose. Mechanism transformed correlation into causation. Climate Science As carbon dioxide levels rise, so do global temperatures. Here, the mechanism is well established: greenhouse gases trap heat. Yet in climate debates, profit has been a powerful force against recognizing causation. Fossil fuel companies funded campaigns casting doubt, reframing causation as “just correlation.” Psychology and Sociology Stress and illness often travel together. But does stress cause illness, or do people who are already ill feel more stress? Both are true. Yet some mechanisms are well pinned down: sleep deprivation impairs memory; seatbelt laws reduce fatalities. Technology and History Faster internet speeds often correlate with GDP growth. But whether one drives the other is debated. Causation is clearer when a computer virus deletes files, or when encryption failures enable hacks. Historically, the Marshall Plan didn’t just correlate with recovery; it injected capital and reshaped institutions. Biology and Ecology Predator and prey numbers rise and fall together, but which drives which? The answer is both, in a loop. A sharper causal case is invasive species: introduce brown tree snakes to Guam, and native birds vanish. DNA sequencing has redrawn family trees, showing that many groupings once thought natural were correlations of form, not ancestry. Sports Teams with higher payrolls often win more — but do salaries buy talent, or do success and resources reinforce each other? By contrast, steroids build muscle causally, and the three-point line causally reshaped basketball play styles.

Sidebar: Rethinking Mental Disorders — From Correlation to Causation

Psychiatry has long wrestled with the problem of correlation. Do symptoms that appear together reveal a common cause, or are they simply co-travelers? In recent years, researchers have reframed the question through three complementary approaches: Dimensional Models: Traits measured along a continuum rather than treated as all-or-nothing categories. Example: Autism spectrum. Network Models: Symptoms are treated as nodes that can activate each other. Example: Depression cycles of insomnia → fatigue → hopelessness → further insomnia. Transdiagnostic Frameworks: Mechanisms are studied across traditional diagnostic boundaries. Example: Threat reactivity across anxiety, PTSD, and depression.

Sidebar: Taxonomy’s Lessons — When Names Mislead

The history of classification is also a story about correlation and causation. Linnaean ranks, created before evolutionary theory, grouped organisms by outward traits — a kind of pattern recognition. Cladistics reframed the task: names should track actual ancestry. Missteps fall into two traps: Paraphyly (incomplete history): Example: Reptiles once excluded birds, leaving out descendants. Polyphyly (false story): Example: “Algae” grouped unrelated lineages based on looks. These redrawings remind us that correlations based on surface traits can mislead. Mechanism — here, the actual history of descent revealed by DNA — is the deeper truth.

Part III: Case Studies of Reversals

Ulcers and Helicobacter pylori: For decades, ulcers were blamed on stress. Profitable treatments reinforced the view. In 1982, Marshall and Warren proved bacteria caused ulcers. The correlation became causation. Smoking and Lung Cancer: By the 1950s, smoking and cancer were strongly correlated. Tobacco companies insisted correlation was not causation, funding doubt for decades. Mechanisms confirmed causation, but profit delayed action. Millions died. Hormone Replacement Therapy: Observational studies linked HRT with lower heart disease. In trials, it increased risk. Correlation, not causation. Profit amplified error, exposing millions. Lesson: when correlation is downplayed or promoted for profit, science blurs into manipulation.

Part IV: The Human Apparatus

Humans are wired to mistake correlation for causation. Pattern-seeking: rustle in grass = lion. Safer to assume cause than risk death. Narrative minds: we demand causal glue in stories. Culture/language: some emphasize agents, others patterns. - Science disciplines instincts with trials and experiments, though even scientists slip. - Sometimes mechanisms are unknown yet causation is accepted (smoking). Sometimes resisted (ulcers). Sometimes unknowable (quantum entanglement).

Part V: Historical Lens — 25 Years of Shifting Ground

1990s: Tobacco on Trial. Smoking’s causation undeniable, after profit-driven delay. 2000s: HRT reversal. Apparent causation collapsed into correlation. 2010s–2020s: Big data floods us with purported correlations, many spurious, sold as insights. Present: Vaccines. Anti-vaccine groups misused correlation (autism diagnoses vs vaccine schedule). Massive studies confirm no causation; vaccines prevent disease at far higher odds than side effects. The dilemma: how to weigh odds rationally under uncertainty?

Part VI: Unexpected Consequences

Dismissed as correlation: smoking, lead, asbestos — delays harmed millions. Mistaken as causation: HRT, vaccine-autism myth — harm or mistrust followed. Profit often drives debate. Tobacco, fossil fuels, pharma resisted or promoted stories to suit them. Money amplifies uncertainty and delays action. Consequences ripple outward, shaping lives and trust.

Conclusion: Patterns, Causes, and the Courage to Act

Correlation is a whisper; causation is a story. Both matter, but confusing them costs dearly. Examples across fields show the range; reversals reveal entrenched resistance; profit often clouds judgment. Vaccines highlight the human dilemma: we must act under uncertainty. Science disciplines our storytelling, but cannot eliminate our instinct to see causes. Our task: old both together, test the story, act wisely while uncertainty looms.

Classroom Prompts

- Think of a time you assumed A caused B. Did it hold? - Why is “correlation does not imply causation” repeated so often? Is it always true? - How can citizens guard against profit-driven misuse of correlation? - How would you explain correlation vs causation to a younger sibling?