Cum Hoc (Correlation as Causation)
The cum hoc fallacy assumes that because two things occur together, one causes the other. Debaters answer by offering a third factor, reverse causation or coincidence, and asking for evidence of the mechanism.
Cum Hoc (Correlation as Causation) as it appears in championship debate rounds. Each example below is a moment from a real round, with a note on why it illustrates the technique and how a debater can use it, or answer it.
Example 1: The speaker takes apart the opponents' example that anti-Semitism rose after Holocaust reparations
He labels it 'the typical example of correlation without causation' and points to a confounder, the Israeli-Palestinian conflict, that also turned people against Israel. He also attacks the warrant that 60% opposing reparations means 60% would lash out. Students can expose cum hoc reasoning this way: name the fallacy, then supply a plausible alternative cause for the trend the opponent cites.
Example 2: The speaker cites a chart showing that the SAT correlates with college performance
He then slides into saying the test is 'effective at causing good college outcomes', and he quotes shifting figures (65, 80, nearly 90 percent) without clear sourcing. Concluding causation from a correlation is the cum hoc fallacy, committed here in passing. Students should cite predictive validity as prediction only, and when facing such evidence, press opponents on the gap between correlation and causation and on inconsistent statistics.
Example 3: The third opposition speaker answers proposition's hate-crime statistics (crimes up 40-41% after Brexit and the US election)
He says they never gave structural reasons why right-wing populist speech necessarily causes hate crimes, so there is 'a certain correlation but no definite causation.' This exposes a cum hoc inference, and the adjudicator later singled the move out. Use it whenever an opponent's case rests on a statistic that rose after an event, and demand the mechanism linking the two.
Example 4: The negative answers the aff's list of former debaters who became activists
It says the aff 'has not explained a causal connection for why debate is key': those people may simply have been activists anyway. It backs this with the point that ballots track technical skill, not truth. This exposes treating correlation as causation, and students can counter anecdotal 'success story' evidence by demanding the mechanism and pointing to plausible alternative explanations.
Example 5: The first proposition speaker cites the reversal of the Flynn effect and suggests AI is why IQs are falling
The opposition interrupts with a point of information: AI has only recently become widespread, so the proposition is 'drawing a causation from a correlation.' The speaker concedes the point and falls back on a study that links AI use directly to critical thinking. Students should keep this one-line POI ready for any trend-based statistic, and when they are the ones challenged, they should have direct causal evidence prepared as a fallback, just as this speaker did.
Example 6: In crossfire, the negative claims Russia entered Syria because of sanctions, since its involvement came only after sanctions were imposed
The affirmative exposes this as correlation-as-causation with a quick analogy: scoring 100 on a test after sanctions doesn't mean sanctions caused the score. Students can use a short, absurd parallel like this to show that timing alone proves nothing, then demand the mechanism linking the two events.
Example 7: The speaker commits Cum Hoc
Because UK COVID cases fell about 20% after Boris Johnson lifted restrictions, he concludes that lifting restrictions caused the move to endemic status. He ignores vaccination rates, prior infection and variant changes as alternative causes. To counter this, name those other variables and demand evidence that isolates the policy's effect.
Example 8
The speaker says generative AI is 'directly correlated' with lower cognitive thinking and treats that as proof AI causes the decline, along with procrastination and memory loss. That is correlation presented as causation, a fallacy the speaker commits. Weaker or more procrastination-prone students may simply use AI more. A debater answering this should raise reverse causality and selection effects and demand a study with controls.
Example 9: The affirmative exposes the negative's education claim as correlation treated as causation
Wealthier people were already more likely to afford college, so the link between education and income may run the other way. The speaker then backs this up with a 64-study meta-analysis finding no meaningful association. Students facing statistical links should ask whether a third factor or reverse causation explains the pattern, then bring stronger evidence.
Example 10
The affirmative cites high reoffending rates in Ireland (62%) and the US (76%) and simply asserts that harsh, retributive punishment causes them. He ignores many other variables, such as poverty, policing and drug policy, so the speaker commits correlation-as-causation. A negative can answer by naming alternative causes and asking for evidence that isolates the effect of punishment philosophy. Students using statistics should explain the causal mechanism, not just set two numbers side by side.
Example 11: The lecturer commits a correlation-as-causation error
He argues that because all eight or nine of his Tournament of Champions elim rounds ended 2-1, his strategy of never kicking a judge must have caused those wins. He even frames it as a false dilemma: either he is 'the luckiest man alive' or it reflects strategy. He ignores other causes, such as team skill, opponent quality or panel composition. When an opponent offers a personal track record as proof of a mechanism, ask what else explains the pattern and point out that a small, non-controlled sample cannot establish causation.
Example 12: The negative speaker goes after the affirmative's Chetty evidence linking shorter commutes to upward mobility
He points out that the study's own abstract says it does not control for causality, and offers a confounder: the people with short commutes may simply not live in concentrated poverty to begin with. This exposes a correlation-as-causation leap in the opponent's case. Students can use the move by checking whether a statistical study claims causation, then naming a plausible third variable that could explain the correlation.
Example 13
The con argues that because there is a Republican trifecta yet little legislation, Trump's executive orders must be causing congressional inaction. Two facts happening together are treated as cause and effect, with no ruling-out of other explanations such as narrow majorities, the filibuster or intra-party splits. The speaker commits a correlation-as-causation error dressed as 'thinking about this logically.' To counter it, offer alternative causes and demand the warrant that links the orders to Congress's inaction.
Example 14
The negative points out that the affirmative's Gene Watch UK evidence shows only that conviction rates fell after non-convicted people's records were retained. It gives no causal analysis linking the policy to the drop. That exposes a correlation-as-causation error. Students can use this move against any 'X was implemented, then Y changed' statistic by demanding the mechanism and naming other possible causes.
Example 15: To answer 'counterterrorism doesn't work,' the speaker uses the Byman card to argue that low terrorism deaths on U.S
soil prove that U.S. counterterrorism caused them. This is a correlation-as-causation inference that the con relies on heavily. Pro debaters should attack it by offering alternative explanations for low terrorism rates. Con debaters should make sure their evidence explains the causal mechanism, not just the coexistence of the two facts.
Example 16: The negative rebuttal answers the affirmative's examples of former debaters who became activists
It points out that this is 'purely correlation not causation' and does not show that debate caused their activism. This calls out a cum hoc fallacy in the opponent's subjectivity-shift argument. When opponents offer anecdotes of people who did X and later did Y, students should demand a causal mechanism and point to alternative causes such as school, family or friends.
Example 17: The negative attacks democratic peace theory as correlation mistaken for causation
They argue the evidence shows no causation, the causal arrow may be reversed, and other variables explain the peace better, and they cite regions without democracy that have seen little war. This is how to expose a cum hoc claim: name the missing causal mechanism, offer reverse causation and confounding factors, and give counterexamples. The affirmative's later reply about correlation strength (segments 692-694) shows that answering with correlation strength alone doesn't fully resolve the causation question.
Example 18: The Pro rebuttal says the Con's poverty evidence only shows that Mexico did worse after NAFTA
It does not show that NAFTA caused the decline: 'just because something happened after NAFTA does not mean it was because of NAFTA.' The speaker also says that comparing Mexico with all of Latin America ignores each country's different circumstances. When an opponent's case rests on before-and-after or cross-country trends, name the correlation-causation gap and ask them for a causal warrant.
Example 19
In cross-examination, Senator Nuttall points out that the sponsor's evidence of wider racial turnout gaps in states with voter ID laws only shows correlation. He presses for proof that the laws caused the drop, and the sponsor falls back on 'I guess I took it for granted' and cannot quote the studies. This exposes a cum hoc fallacy. Students can use this short line of questioning on any statistic that only compares states or time periods, and should be ready with the causal mechanism or study design when defending their own correlational evidence.
Example 20: The speaker notes that countries with rising educated classes, like China, also require physical education
From this alone he concludes that the PE requirement drives their academic performance, treating a correlation as proof of cause. Opponents should name the confounders: national wealth, education spending, exam culture and school hours. Then ask for evidence that PE, and not those factors, explains the gap.
Example 21
The speaker argues that gendered toys and clothing create a 'gender hierarchy.' She then says 'it's no wonder' that only 7% of Fortune 500 CEOs are women, that there is a pay gap, and that 130 million girls are kept out of school, and concludes the game 'has been rigged since birth.' She commits cum hoc: these outcomes coexist with gendered childhoods, but she never shows that one causes the other, and many global and economic factors are skipped. Students should challenge a 'no wonder' bridge by asking for the causal mechanism, or set up their own causal link explicitly before citing outcome statistics.
Example 22
The negative attacks the aff's Hendrick card, which ran a simple regression and found no correlation between conflict behavior and oil prices. The negative argues that a plain correlation test cannot establish causality either way. It says only its own evidence uses instrumental variables to handle reverse causality and endogeneity. This shows how to expose an opponent's misuse of correlational data: ask whether their study can separate cause from effect, and point to methodology that does.
Example 23: The opposition speaker answers the proposition's deterrence argument, which compared Southern and Northeastern murder rates
He points out that 'correlation isn't causation' and lists confounders such as population, poverty and crime levels. He then cites two academic experts who say the research on deterrence is inconclusive. Students can use this to defuse regional-comparison statistics: name the fallacy, list the uncontrolled variables, and back the point with an authoritative source.
Example 24
In her summary, Daniela claims 80 percent of society went to boarding school and asks 'don't you think it's connected?' This suggests boarding school causes adults to commit FGM and rape. It is a clear correlation-as-causation fallacy, built on an unsourced and implausible statistic. Opponents should point out the missing causal mechanism and the base-rate problem: if most people attended, nearly any adult behaviour will correlate with boarding.
Example 25
The speaker relays an argument from the literature: Vietnam saw fierce anti-war protests and Iraq and Afghanistan did not, so the difference must come from the draft versus the all-volunteer force (the 'civil-military gap'). This infers causation from a single correlated difference between two eras and ignores other causes, such as casualty levels, media coverage, and the political climate, so it is a textbook cum hoc setup rather than a debate round. A student running this advantage should find warrants beyond the historical correlation, and a negative can attack it by naming alternative causes of the protest gap.
Example 26: The affirmative goes after the negative's 'biggest card', the claim that more transfer spending reduces poverty
They point out that it only says welfare 'might' help and 'doesn't actually show causality'. That exposes a correlation-as-causation weakness in the evidence. Students can use this by reading the opponent's statistics closely and asking whether the study isolates cause or just reports an association.
Example 27: In crossfire the affirmative attacks the negative's only backlash example
The negative claims Japanese American reparations caused Proposition 187, but the affirmative gets them to admit the two were six years apart and asks how that gap shows causation. They also point out the law was struck down within a month. This exposes a correlation-as-causation (post hoc) link, so students should demand the causal mechanism whenever an opponent says one event 'led to' a later one.
Example 28: The questioner exposes the Diamond evidence
It simply lists autocracies like North Korea, China and Iran alongside global threats, with no statistical relationship showing that authoritarianism causes those problems. The defender falls back on calling a 'rhetorical' card fine if it goes unanswered, which tacitly concedes that it shows only co-occurrence. Students can counter impact cards built on examples by asking for the causal mechanism or data, and they should be ready to supply that mechanism when defending their own impacts.
Example 29: The speaker argues Roberts controls the Court because he was in the majority in 52 of 53 cases and all 5-4 decisions
He treats being on the winning side as proof that his vote causes the outcome, a correlation-as-causation error that is committed but never exposed. A justice can sit in the majority because he follows the bloc rather than steers it. Students should ask whether the person actually moves the result or merely coincides with it, and should back any 'kingmaker' claim with evidence of pivotal votes.
Example 30: The speaker rebuts the claim that reparations funding will improve schools
He argues that the studies linking funding to outcomes only show correlation: well-funded schools tend to have involved parents, and once that is controlled for, extra funding does not raise outcomes. This exposes a correlation-as-causation error in the opponent's evidence. Students can use the move by naming the confounding variable and citing evidence that controls for it, rather than simply denying the link.
Example 31: The speaker cites 6.4% first-quarter GDP growth and claims that 'all of that growth actually came from' stimulus checks
Growth that came after the checks is treated as caused by them, ignoring reopening, vaccines and other spending. This is correlation presented as causation. An opponent can expose it by naming those alternative causes and asking for evidence that isolates the checks' effect.
Example 32: The first proposition speaker cites the reversal of the Flynn effect and implies AI is why IQs are falling
The opposition's point of information exposes this as correlation treated as causation, since AI only became widespread recently. The speaker concedes and retreats to a different study. Students should watch for broad trend data tied to a recent cause and challenge the timeline. When they use correlational evidence themselves, they should pair it with a direct causal study.
Example 33: The Con asks why US income inequality doubled after the country began signing trade deals
The Pro replies that this evidence is 'correlational and not causal' and names other causes, such as the pandemic and recessions. This exposes a cum hoc fallacy: the Con treats the timing of trade deals and rising inequality as proof that one caused the other. Students can use this reply whenever an opponent leans on a time-series trend, and should back the claim with their own causal study, as the Pro tries to do.
Example 34
The speaker sets 1950s classroom problems beside today's and then credits the whole change to America having 'removed God' from schools, courts and culture. He never rules out other causes, so this commits correlation-as-causation: two trends happening in the same period are treated as cause and effect. To counter it in a round, name competing causes such as economic shifts, urbanization or changes in reporting, and demand a mechanism linking the proposed cause to the effect.
Example 35: In crossfire, the Pro presses the Con on what their Italy evidence actually says
The Con admits it is just a polling chart showing Salvini's party gaining, with nothing tying the rise to the BRI. The questioner exposes this as correlation presented as causation. Students can use this line of questioning to make opponents admit their evidence shows timing, not causation. When defending, have a card that states the causal mechanism explicitly.
Sources: every example links to the YouTube video of the championship debate it comes from, at the moment described.
All techniques and fallacies