Cherry-picking data to fit a preconceived pattern.

Explanation

The Texas Sharpshooter Fallacy arises when observers examine data or events after the fact, isolate a cluster of similarities or apparent patterns, and then treat that cluster as evidence of deliberate design, causation, or predictive power while downplaying or ignoring the broader distribution of outcomes and the high likelihood that such clusters emerge from random variation alone. This process mirrors the anecdote of a marksman firing randomly at a barn wall and afterward drawing a bull’s-eye around the densest grouping of holes to claim exceptional accuracy; the target is created to fit the shots rather than the shots being aimed at a pre-existing target. Psychologically, the fallacy stems from apophenia—the human tendency to perceive meaningful connections in unrelated or random information—and the closely related clustering illusion, in which people systematically underestimate how often tight groupings appear in purely chance distributions. These tendencies are amplified by confirmation processes that direct attention toward elements aligning with an emerging narrative while filtering out contradictory ones.Neuroscience research links such pattern-seeking to the brain’s predictive coding architecture, in which sensory cortices and association areas generate expectations that are updated by incoming data; when random fluctuations produce locally coherent features, reward-related dopamine signaling in frontostriatal circuits reinforces the perception of order, making it subjectively compelling even when statistical scrutiny reveals it to be illusory. The result is a post-designation error: significance is assigned after the data have spoken rather than through hypotheses formulated and tests specified in advance. As psychologist Thomas Gilovich and colleagues demonstrated in controlled studies of perceived streaks in random sequences, this mechanism leads individuals to overinterpret local regularities while neglecting base-rate probabilities across the entire sample space.

Examples

• Percival Lowell’s mapping of Martian canals (1890s, Flagstaff, Arizona): In 1894 astronomer Percival Lowell used his personal fortune to establish an observatory on a high plateau near Flagstaff, Arizona, equipped with a 24-inch refracting telescope. During successive oppositions of Mars he and his assistants recorded faint, intermittent linear markings across the planetary disk and connected them into an extensive network of straight “canals.” Lowell published detailed maps and arguments in his 1895 volume Mars and expanded them in the 1906 book Mars and Its Canals, claiming the features formed a single, geodetically precise system. He wrote that “the lines run straight throughout their course… Nothing on Earth of natural origin on such a scale bears them analogue” and that their “geodetic precision… instantly stamps them to consciousness as artificial.” Later spacecraft imagery from Mariner 4 in 1965 revealed only craters and natural albedo variations, demonstrating that atmospheric turbulence and perceptual completion had produced the apparent regularity. Over-reliance on the selectively highlighted linear cluster as evidence of intelligent engineering created vulnerability to a civilization-scale misinterpretation; balanced investment in pre-specified observational protocols, statistical analysis of all markings rather than chosen subsets, and higher-resolution instrumentation could have prevented the premature conclusion and preserved scientific credibility until genuine data arrived.
• Popular compilations of Lincoln–Kennedy assassination parallels (post-1963, United States): After the November 1963 assassination of President John F. Kennedy, lists enumerating supposed parallels with the 1865 assassination of Abraham Lincoln circulated widely in American newspapers and political newsletters. Compilers selected items such as both presidents being elected in a year ending in 60, both being shot from behind in the head on a Friday, and both being succeeded by a Southern Democrat named Johnson. Writer Martin Gardner later examined these lists in Scientific American and his 1985 book The Magic Numbers of Dr. Matrix, documenting that many claimed facts were inaccurate or stretched while countless differences between the two events were omitted. One typical entry read: “Lincoln was elected President in 1860; Kennedy was elected President in 1960,” alongside assertions about assassins born roughly a century apart. The selective emphasis on a curated cluster of surface similarities ignored the vast number of possible comparisons across any two historical figures and the role of chance in producing occasional matches. Over-reliance on this post-hoc pattern as proof of deeper historical design created vulnerability to conspiratorial or mystical interpretations; balanced investment in exhaustive comparison of all attributes or statistical evaluation of coincidence rates across random pairs of events could have revealed the lists as artifacts of selective attention rather than meaningful signals.
• 1993 Swedish study of proximity to high-voltage power lines and childhood leukemia: Swedish epidemiologists examined health records of children living within 300 meters of high-voltage transmission lines over a 25-year period and analyzed associations with numerous medical conditions. The study highlighted a fourfold elevation in childhood leukemia incidence among the exposed group, prompting public concern and policy discussion. Subsequent independent analyses and pooled international data failed to confirm a consistent causal link, illustrating how examining hundreds of potential health outcomes increases the probability that at least one statistically significant association will appear by chance alone—the classic multiple-comparisons problem. The researchers’ emphasis on the leukemia finding while the broader dataset contained many null results exemplified the drawing of a target around one cluster after the data were collected. Over-reliance on the selectively reported cluster as a preventive warning signal against an environmental hazard created vulnerability to regulatory overreaction and public anxiety unsupported by replication; balanced investment in pre-registering primary outcomes, applying corrections for multiple testing, and requiring independent replication before policy action could have produced more reliable public-health guidance.
• Michael Drosnin’s Bible Code predictions (1997, United States): Journalist Michael Drosnin published The Bible Code in 1997, asserting that equidistant letter sequences (ELS) hidden in the Hebrew text of the Torah encoded future events, including the 1995 assassination of Israeli Prime Minister Yitzhak Rabin. Drosnin described meeting Rabin in 1994 and warning him on the basis of a code sequence placing “assassinated” near Rabin’s name; the book presented additional “predictions” for other historical and contemporary occurrences. Statistical critics demonstrated that in a text of roughly 300,000 letters, the enormous number of possible skip distances generates countless coincidental letter clusters that can be retrofitted to almost any event after it occurs. Drosnin’s method selected and publicized only those sequences that matched known outcomes while ignoring the vastly larger set of non-matches. Over-reliance on these post-facto clusters as encoded foreknowledge created vulnerability to pseudoscientific claims that influenced popular belief and distracted from rigorous textual scholarship; balanced investment in blinded, pre-specified search protocols and comparison against control texts of similar length could have shown the “codes” to be expected products of random letter arrangements rather than deliberate messages.

Conclusion

The Texas Sharpshooter Fallacy distorts judgment across domains by converting the brain’s adaptive capacity for detecting regularities into a source of systematic error once data volume and complexity exceed ancestral environments. Neurobiologically, the same frontostriatal and cortical circuits that generate reward for coherent percepts continue to operate when those percepts arise from sampling artifacts or exhaustive post-hoc search, producing subjective certainty detached from objective probability. For individuals the bias encourages overconfidence in personal narratives and investment decisions; for society it fuels pseudoscience, selective historical memory, and policy based on unrepresentative clusters. In scientific fields it undermines reproducibility by rewarding data dredging over hypothesis-driven inquiry. Mitigation requires institutional practices such as hypothesis pre-registration, correction for multiple comparisons, and insistence on replication with fresh data—practices that shift emphasis from pattern discovery after the fact to pattern testing before interpretation. The enduring lesson is that meaning is not conferred by the tightness of a cluster but by the rigor with which the entire distribution is examined; without that discipline, even the most elegant target remains a decoration painted around chance.

Quick Reference

→ Synonyms: clustering illusion; post-designation fallacy; data dredging (related form)
→ Antonyms: pre-registered hypothesis testing; base-rate reasoning; multiple-comparison correction
→ Related Biases: confirmation bias; illusory correlation; availability heuristic; p-hacking

Citations & Further Reading

• Drosnin, M. (1997). The Bible code. Simon & Schuster.
• Evers, J. L. H. (2017). The Texas sharpshooter fallacy. Human Reproduction, 32(7), 1363–1364.
• Gardner, M. (1985). The magic numbers of Dr. Matrix. Prometheus Books. (Reprint of earlier Scientific American analysis.)
• Gilovich, T., Vallone, R., & Tversky, A. (1985). The hot hand in basketball: On the misperception of random sequences. Cognitive Psychology, 17(3), 295–314.
• Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
• Lowell, P. (1906). Mars and its canals. Houghton, Mifflin and Company.
• Sheehan, W. (1988). Planets and perception: Telescopic views and interpretations, 1609–1909. University of Arizona Press. (Contextual analysis of Lowell’s observations.)