In my experience the lay people aren't getting a great idea of the state of the field. Like you say coffee good/bad studies. It isn't that simple. People test all sorts of things in different contexts. But maybe some are genuinely bad studies. You don't know though because science journalism is so crappy. They don't care about the merits of the study. They go "people drink coffee, maybe that would drive engagement."

The most interesting papers are not going to get popular press releases because they are so many steps removed from the context that lay people understand. They can understand "coffee good/bad." They can't understand anything about the stories we are actually telling at the bleeding edge of a field, because even our undergrads working in our labs on these projects can scarcely understand them. Second year grad students struggle to understand them. How can a science journalist who only has a bs from communications department, or the lay public, possibly understand?

So, they don't reach for those papers when they seek to write articles for engagement. And the lay public doesn't learn the state of the art, and assumes the worst of the field from what they do read about.