This does raise an interesting, and important point. However, there are a few aspects of the data that do suggest the effect we are seeing here is that the public are being influenced by the media. First, we see that the effect of media frequency is lessened when objective information about the diseases is presented alongside the disease label. This suggests that it is not people's (stable) perceptions that drive the media coverage. Furthermore, when people were presented with the disease descriptions without the label attached, they actually thought that the low media frequency diseases were worse than the high media frequency ones.

Another thing that this study adds is that it shows how it can change people's understandings of the diseases themselves. That is, it is just simple risk perception (e.g. in estimates of prevalence), it is a decision about how serious a disease it is, and also how "disease-like."

That said, the full picture clearly has to include somewhat of a reciprocal relationship between the public and the media, and that both the public and the media are responding to expert claims about various health threats.

This raises an interesting question, I think. Experts may be giving out information about a particular threat, but it is not necessarily their job to put it into context of all other possible threats (especially those outside their own fields of expertise). Furthermore, if the media are reporting on those threats, again, there is no particularly good mechanism to put each story in the larger context. It is the information consumers who effectively must compile in their own minds all of the different sources of information, to come up with some idea of relative risks, and this is an incredibly difficult task. Given that one thing you can judge fairly easily is your relative familiarity with different diseases (from media coverage), it makes it a very compelling heuristic to use as a way of assessing relative seriousness (as well as relative risk).