The Affect Heuristic
The affect heuristic is letting how you feel about something decide what you believe about it. You do not weigh a technology's risks against its benefits and arrive at an attitude; you arrive at the attitude first, and the risks and benefits are filled in to match. The name and the research programme are Paul Slovic's, and Daniel Kahneman is the popular vehicle for it rather than its author. The phenomenon looks real and general. Almost none of it is quantified in the sources this vault holds, which is why the page is marked mixed.
What it is
Structurally this is substitution with feeling as the easy question. Instead of asking what the evidence says about a risk, the mind answers how it feels about the thing, and translates that answer onto the risk scale without noticing the swap.
The signature finding is a correlation that should not exist. Slovic asked people to rate a list of technologies on benefit and on risk separately: water fluoridation, chemical plants, food preservatives, cars and others. In the world, benefit and risk are largely independent, and a high-benefit technology is as likely to be dangerous as a safe one. In people's ratings the two are strongly negatively correlated. Whatever is judged beneficial is judged safe; whatever is judged harmful is judged useless. The correlation is high enough that it cannot be a summary of any real relationship in the world. It is a summary of an attitude.
The book this vault holds does not print the correlation value, and this page does not invent one.
Two results sharpen the picture. Time pressure tightens the link: rushed, the correspondence between rated benefit and rated risk gets closer, which is the signature of a fast feeling-based process rather than a slow analytic one. And experts do it too. Members of the British Toxicology Society show the same pattern, which is the result that stops this being a story about an ignorant public. Domain expertise does not dissolve it. No sample size, year or publication is given for that study in the source held here.