Agreed that this does not represent the world, or Germany, or students in general. But it is one datapoint about social science students in Frankfurt, which was

Agreed that this does not represent the world, or Germany, or students in general. But it is one datapoint about social science students in Frankfurt, which was deliberately chosen as a "likely worst case" to detect the problem. In fact, the authors are very explicit about that point: "We are certainly not under the impression that our sample is representative of university students in general (or the wider public, for that matter). On the contrary, we purposefully consider the social science studentship at Frankfurt as a most likely case." Next, as outlined above, n=500 seems enough to reject the hypothesis that the percentage that wants to ban the books is near 0. In other words, I dispute your notion that the study has not enough power. Strong sampling bias might be a problem, but nobody has made a case for that yet.