"People may endorse a conspiracy theory to the same extent," Javier Granados Samayoa told the Annenberg Public Policy Center, "but that does not necessarily mean that it will translate into behavior to the same extent."
That sentence should have ended a lot of arguments. It didn't, because it contradicts the premise that funds the entire misinformation-correction industry.
Here is a claim debunkers rarely say out loud, though the evidence pack does not establish it: that decades of fact-checking rest on the assumption that false belief drives bad behavior, and that the data has never really supported that assumption.
The link between conspiracy belief and what people actually do is, in the researchers' own word, "weak and heterogeneous." Weak enough that for years the field treated the problem as a measurement failure—we just needed sharper facts, better corrections, cleaner delivery.
This week the target is more specific: the variable that separates a belief someone merely holds from a belief that moves their hands and feet. It isn't exposure. It's conviction.
A new study in the British Journal of Social Psychology, from a team at Penn's Annenberg center and Texas Christian University, ran three studies totaling 6,082 U.S. participants. The design was built to catch behavior in the wild, not just in the survey—including one arm that tracked people prospectively over three months.
The domains were deliberately combustible. The "great replacement" theory and immigration attitudes. COVID-19 vaccine conspiracies and actual vaccination status. A "deep state undermining Trump" narrative tracked across donating, volunteering, sharing political information, and attending rallies.
Across all three, the headline finding is the same, and it is quietly devastating to the standard playbook: conspiracy belief on its own was only "modestly" associated with behavior.
Endorsement, by itself, is close to inert. Two people can score identically on believing the election was stolen or the vaccine is poison, and one goes to a rally while the other goes back to scrolling. The belief content doesn't tell you which.
Two people can score identically on believing the vaccine is poison, and one goes to a rally while the other goes back to scrolling. The belief content doesn't tell you which.
What tells you is how the belief is held.
The researchers isolated a chain. Certainty—how sure you are the claim is true—strengthened the pathway from belief to attitude to action. So did personal importance—how much the belief matters to who you are.
According to the study, those two effects were in turn amplified by a third quality: perceived actionability, whether the person treats the belief as a useful guide for deciding what to do.
Actionability, in one report's framing, reportedly amplified the certainty and importance effects—described as "the hinge." When people saw a conspiracy belief as a decision tool, the certainty and importance effects intensified. When they didn't, the same belief sat there like furniture.
This is the reframe. A conspiracy theory is not dangerous because it is false. Plenty of false beliefs are held with total certainty and never do anything. It becomes dangerous when the believer converts it into an instruction.
Which means content-debunking is aiming at the wrong variable. You can win the argument about whether the claim is true and lose entirely on whether the person uses it—because those are different psychological objects.
Consider what correction actually targets. A good fact-check attacks certainty: here is why the claim is false, here is the evidence. Arguably it nudges one of three variables, and perhaps not the load-bearing one, though the evidence pack does not evaluate this.
It arguably rarely touches importance, which may be bound up with identity and grievance rather than evidence; and it may do little to actionability, since whether a belief feels like a usable decision rule could be a separate judgment from whether it feels true—though the evidence pack does not establish these points.
You could, by way of illustration, believe the moon landing was faked with rock-solid certainty and never once treat that belief as a reason to do anything. Certainty without actionability is a curio. It's the conversion to a decision tool that should worry us.
Senior author Dolores Albarracín put the consequence plainly: "Simply correcting misinformation is not enough. If we want to reduce the harmful effects of conspiracy beliefs, we need to understand the psychological qualities that give those beliefs their power to motivate behavior."
So where is the leverage, if not the claim?
At the hinge. If actionability is the step that converts conviction into behavior, the highest-value intervention interrupts that specific conversion—not the belief, not even the certainty, but the automatic slide from "I believe this" to "therefore I act on this."
There's a clean demonstration of what interrupting that slide looks like, in a different but adjacent domain: sharing.
Research summarized by the American Psychological Association found that most people share information first and evaluate it later. The urge is automatic; the judgment, if it comes, arrives after the click.
But when people are prompted to actively reflect on the ethics of sharing, they become less likely to spread fake news—"even when it aligns with their own beliefs." The reflection creates, in the APA's phrasing, "a small barrier."
Read that carefully, because the crucial detail is the italicized part. Per the APA, the intervention reportedly still has an effect even when the headline matches what the person already believes.
That is the whole point. It is not changing the belief. The belief is unchanged, still there, still endorsed. What changes is the automatic path from believing to acting on it. Forcing a pause weakens the reflexive urge to convert conviction into behavior.
That "thinking twice" pause is not persuasion. It doesn't debunk anything, doesn't make anyone feel heard, doesn't de-escalate a fight. It does one narrow thing: it inserts an evaluative moment between conviction and use.
It moves the decision point. Normally people evaluate before they believe—which is exactly the step the fact-check tries to influence, and exactly the step that's already lost by the time someone is certain and invested. The pause instead makes them evaluate before they act.
Those are different targets. The first tries to stop the belief from forming. The second lets the belief exist and stops it from becoming an instruction.
If the study is right that actionability is the hinge, the second target is the tractable one. You are not asking someone to abandon a conviction bound up with their identity. You are asking them to pause before treating that conviction as a decision rule—a far smaller request, aimed at a far more consequential joint.
This also explains why so much counter-messaging feels like shouting into a well. It fires at belief content and certainty while the variable that, per the study, amplifies the pathway to behavior—perceived actionability—goes largely untouched. You can move the readings on the wrong dials all day.
The authors note the logic likely generalizes beyond conspiracy theories. Any belief system—political, medical, financial—becomes behavior-shaping under the same three conditions: high certainty, high importance, and perceived action value.
Which reframes the entire project. The question was never whether people believe wrong things. People believe wrong things constantly, harmlessly, permanently.
The question is whether a given belief has been armed—held with certainty, charged with importance, and above all loaded with the sense that it tells you what to do. Debunk the claim and you've disputed one reading on the least important gauge.
Interrupt the hinge and you've reached the mechanism that actually converts a conviction into an act. That's the difference between arguing with a belief and defusing one.