Do they really <i>believe</i> that? Measuring salient conspiracy endorsement

Surveys frequently report widespread belief in conspiracy theories, prompting concerns about their democratic consequences. Yet, standard survey measures often implicitly treat agreement as equivalent to politically consequential belief, even though agreement can reflect a range of engagement—from momentary reactions to durable worldviews. This paper argues that an important dimension of belief is often insufficiently captured in existing approaches: salience. I introduce a salience-based measure that incorporates certainty and prior familiarity to distinguish more tentative or situational endorsement from internalized, action-relevant belief. Using original survey data, I show that this measure correlates more strongly with psychological traits associated with conspiracism and better predicts self-reported engagement: including discussing, posting about, and researching conspiracy theories. These results suggest that traditional measures may overstate the prevalence of politically meaningful conspiratorial belief and obscure substantial heterogeneity among those who agree with conspiracy claims. By refining how belief is measured, this paper offers a tool to more accurately identify which survey endorsements are likely to reflect consequential belief.