The Median Isn't the Message

Stephen Jay GouldThe AMA Journal of Ethic 2013 · DOI

Abstract: My life has recently intersected, in a most personal way, two of Mark Twain’s famous quips. One I shall defer to the end of this essay. The other (sometimes attributed to Disraeli) identifies three species of mendacity, each worse than the one before—lies, damned lies, and statistics.

Introduction: The Statistical Trap

In 1982, evolutionary biologist Stephen Jay Gould was diagnosed with abdominal mesothelioma, a rare and lethal form of cancer. Upon researching the medical literature, he encountered a chilling statistic: the disease carried a "median mortality of eight months." To most, this number represents a death sentence—a countdown to an inevitable end. However, Gould argues that this interpretation stems from a fundamental misunderstanding of statistics, specifically the "Platonic Fallacy," where humans view averages as fixed essences rather than measures of variation.

This paper explores how Gould applied the conceptual framework of evolutionary biology and statistical reasoning to reframe his prognosis. By shifting focus from the median to the distribution, he transformed a grim data point into a roadmap for survival and intellectual defiance.

Methodology: Statistical Reasoning and Skewness

Gould’s approach was not to ignore the data, but to analyze it with the rigor of a population biologist. He employed several key statistical concepts to evaluate his situation:

1. Distinguishing Central Tendencies

He first distinguished between the mean (arithmetic average) and the median (the 50th percentile). In survival statistics, the median is often used because it is less influenced by extreme outliers. However, the median only tells us where the middle of the population lies; it says nothing about the shape of the population on either side of that center.

2. Distribution Analysis and Right-Skewness

Gould reasoned that survival data for a terminal illness cannot follow a symmetrical "normal" distribution (a bell curve). Because a patient cannot die before they are diagnosed, the distribution is bounded by zero on the left. If the median is eight months, the left side of the curve is cramped between zero and eight. However, the right side—the "tail" of the distribution—has no such physical boundary and can extend for years or even decades. This creates a "right-skewed" distribution.

3. Individual Profiling

Gould then cross-referenced his personal variables against the characteristics of the population. He noted that he was young, his disease was detected early, and he had access to the best possible medical care. These factors suggested that he was not a candidate for the "left" side of the curve, but rather a prime candidate for the "long tail" on the right.

Key Insights: Variation as Reality

The core of Gould’s argument is a pillar of evolutionary biology: variation is the hard reality, while the "average" is a secondary abstraction. In a biological system, no two individuals respond to a disease in the same way.

The "Long Tail" of Hope

Because the distribution is right-skewed, the possibility of living far beyond the median is not a statistical anomaly but a mathematical necessity of the curve's shape. Gould realized that even if 50% of patients die within eight months, the remaining 50% inhabit a space that stretches out indefinitely. This "long tail" provides a rational, statistical basis for hope.

Knowledge as a Survival Mechanism

Gould posits that his technical literacy saved him from the "heart-mind dichotomy"—the idea that cold statistics are the enemy of a positive attitude. On the contrary, he found that a rigorous understanding of the data fueled his "sanguine personality." By finding a rational reason to believe he could survive, he maintained the mental fortitude necessary to endure grueling experimental treatments.

Results and Personal Outcome

While Gould’s essay is a narrative meta-analysis of clinical interpretation rather than a quantitative study, the results of his logic were personally profound. By reframing his diagnosis as a statistical challenge rather than a destiny, he was able to "set his house in order" for a long-term fight rather than for immediate death.

Quantitatively, the results were staggering. Despite the "eight-month median," Gould lived for 20 years following his diagnosis. He continued his prolific career as a scientist and writer, eventually passing away in 2002 from a completely unrelated form of cancer. He became the literal embodiment of the right-skewed tail he had predicted two decades earlier.

Conclusion and Implications

Gould’s experience serves as a critical lesson for the medical community and the public alike. It highlights the danger of providing raw, misunderstood numbers to patients without context.

He suggests that "humane doctors" must move beyond diplomatic silence and instead help patients understand the distribution of possibilities. When patients understand that a median is not a message of certain death, but a measure of a highly variable population, they are empowered to find hope through intellectual clarity. As Gould concludes, statistics should be used to illuminate the range of human potential, not to obscure it under the guise of a single, reductive number.

Summary updated

Abstract

My life has recently intersected, in a most personal way, two of Mark Twain’s famous quips. One I shall defer to the end of this essay. The other (sometimes attributed to Disraeli) identifies three species of mendacity, each worse than the one before—lies, damned lies, and statistics.
BibTeX
@article{gould2013median,
  author = {Stephen Jay Gould},
  title = {{The Median Isn't the Message}},
  journal = {The AMA Journal of Ethic},
  year = {2013},
  doi = {10.1001/virtualmentor.2013.15.1.mnar1-1301},
}
Figure