{
  "id": "post-hoc-fallacy",
  "title": "Post Hoc Fallacy",
  "lang": "en",
  "source_lang": "en",
  "summary": "The post hoc fallacy infers that one event caused another merely because it occurred earlier, mistaking temporal sequence for sufficient evidence of causation.",
  "category": "philosophy",
  "subcategory": "metaphysics-epistemology",
  "type": "concept",
  "aliases": [
    "Post hoc ergo propter hoc",
    "After this, therefore because of this",
    "Post hoc reasoning"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/post-hoc-fallacy/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/post-hoc-fallacy/",
      "json": "https://aiwiki.page/zh/post-hoc-fallacy.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The post hoc fallacy is an informal fallacy in which an event is taken to have caused a later event merely because it preceded it. Its name abbreviates the Latin expression post hoc ergo propter hoc, meaning “after this, therefore because of this.” Temporal order can be relevant to establishing causation, but it does not, by itself, establish a causal connection. The error lies in treating succession as sufficient evidence of causation, not in considering earlier events as possible causes."
    },
    {
      "heading": "Logical structure",
      "text": "The basic pattern is:\n\n- Event \\(A\\) occurred.\n- Event \\(B\\) occurred after \\(A\\).\n- Therefore, \\(A\\) caused \\(B\\).\n\nThe premises establish a temporal relationship, while the conclusion asserts a causal relationship. The missing support concerns whether \\(A\\) actually made a difference to the occurrence of \\(B\\).\n\nFor example:\n\nA person wore a particular scarf to a match, and their team won. Therefore, wearing the scarf caused the victory.\n\nNothing in the stated premises connects the scarf to the result. The conclusion might reflect a sincere belief, a joke, or a superstition; it exemplifies the fallacy when offered as a causal inference supported only by the sequence.\n\nThe fallacy is usually classified as informal because evaluating it requires attention to the argument’s content, evidence, and context, rather than merely its grammatical or symbolic form. As a piece of deductive reasoning, the bare inference is invalid: its premises could be true while its conclusion is false. As inductive reasoning, temporal succession alone generally provides inadequate support for the asserted causal conclusion.\n\nA fallacious argument can nevertheless have a true conclusion. Identifying a post hoc inference does not prove that \\(A\\) did not cause \\(B\\); it shows that the argument has not adequately established that claim."
    },
    {
      "heading": "Why temporal succession is insufficient",
      "text": "In ordinary causal explanations, causes precede their effects, or at least do not occur later than them. But this requirement does not make every preceding event a cause. Several alternative explanations can account for an observed sequence.\n\nCoincidence\n\nTwo events may occur in succession without a relevant causal relationship. Because innumerable events occur each day, some striking sequences are to be expected even when the events are independent. An isolated coincidence provides little basis for distinguishing causation from chance.\n\nCommon causes and confounding\n\nAn earlier event and a later event may both result from some third factor. In research, confounding occurs when the relationship between a supposed cause and an outcome is mixed with the influence of other factors.\n\nFor example, ice-cream sales may rise before swimming-related incidents increase during hot weather. The ordering does not establish that ice-cream purchases cause those incidents. Weather and patterns of outdoor activity may help explain both.\n\nBackground trends\n\nAn outcome may already have been changing before the supposed cause occurred. A shop might introduce new signs shortly before sales rise, although sales had been increasing because of seasonal demand. A simple before-and-after comparison cannot, on its own, separate the effect of the signs from the existing trend.\n\nRegression toward the mean\n\nRegression toward the mean occurs when unusually extreme observations tend to be followed by less extreme observations in situations involving imperfectly correlated measurements. If an intervention is introduced immediately after an exceptionally poor result, the next result may improve even if the intervention has no effect.\n\nFor example, a manager might change a routine after an unusually low-output week. A subsequent return to more typical output does not by itself establish that the new routine caused the improvement."
    },
    {
      "heading": "Relation to other causal errors",
      "text": "The post hoc fallacy belongs to the broader family of false-cause fallacies, in which an argument asserts a causal relationship without adequate justification.\n\nIt is closely related to confusing correlation with causation. The distinction is one of emphasis:\n\n- Post hoc reasoning emphasizes temporal order: \\(A\\) happened before \\(B\\), so \\(A\\) caused \\(B\\).\n- Correlation-based causal error emphasizes association: \\(A\\) and \\(B\\) vary together, so one causes the other.\n\nThese errors can overlap. Repeatedly observing one event before another may establish a pattern, but a pattern can still result from common causes, selection effects, or other processes.\n\nPost hoc reasoning also differs from affirming the consequent. That formal fallacy has the structure “If \\(A\\), then \\(B\\); \\(B\\); therefore \\(A\\).” A post hoc argument need not contain such a conditional premise. Its distinctive unsupported step is the move from temporal succession to causal attribution."
    },
    {
      "heading": "Temporal evidence and legitimate causal inference",
      "text": "Temporal order remains important evidence. If a proposed cause occurred only after an effect, that ordering ordinarily counts against the proposed explanation. Establishing that a candidate cause came first, however, generally leaves further questions unanswered.\n\nA stronger causal argument may draw on:\n\n- Comparison: whether the outcome differs under otherwise comparable conditions with and without the proposed cause.\n- Control of alternative explanations: whether other changes could account for the outcome.\n- A plausible mechanism: how the proposed cause could produce the effect.\n- Consistency across observations: whether the relationship appears under relevant repeated or varied conditions.\n- Independent evidence: whether different methods support the same explanation.\n\nAn experiment can help distinguish causation from mere sequence by deliberately varying a factor and comparing outcomes. A randomized controlled trial uses random assignment to reduce systematic differences between comparison groups. Such methods do not automatically eliminate every source of error, but they address questions that an uncontrolled before-and-after observation leaves unresolved.\n\nCausal investigation need not always involve experiments. Historical events, geological processes, and many social phenomena cannot be freely manipulated. In these cases, investigators may combine observational comparisons, chronology, mechanisms, and evidence bearing on competing explanations.\n\nThe counterfactual question—what would have happened without the proposed cause?—helps express what a simple sequence cannot answer. For a single observed event, that alternative outcome is not directly available, so causal inference requires assumptions and supporting evidence."
    },
    {
      "heading": "Philosophical context",
      "text": "The distinction between succession and causation is central to philosophical discussions of causal knowledge. David Hume examined how experience of repeated conjunction leads people to expect one kind of event after another, while questioning whether a necessary connection is directly perceived in the events themselves.\n\nHume’s analysis is broader than the post hoc fallacy. It concerns the basis of causal expectation and the problem of induction, not merely mistakes made from a single sequence. Nevertheless, it clarifies an important distinction: observing that events follow one another is not identical to establishing why they occur.\n\nIn the philosophy of science, causal claims are consequently assessed through more than chronology alone. Their justification depends on the quality of the available evidence, the adequacy of comparisons, and the treatment of rival explanations. A single sequence can suggest a useful hypothesis; it becomes post hoc reasoning when that sequence is treated as sufficient grounds for accepting the causal conclusion."
    }
  ],
  "links": [
    "fallacy",
    "latin",
    "causation",
    "deductive-reasoning",
    "inductive-reasoning",
    "confounding",
    "regression-toward-the-mean",
    "affirming-the-consequent",
    "experiment",
    "randomized-controlled-trial",
    "counterfactual",
    "david-hume",
    "problem-of-induction",
    "philosophy-of-science",
    "evidence",
    "hypothesis"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}