Zum Seitenende Übersicht Artikel Home & Impressum
First the link to this week’s complete list as HTML and as PDF.
Mehl et al. is just another example that you must not listen to what people say but look at what they actually do if you need objective and reliably information. The advertising industry, where serious money is involved, has understood this long ago but large parts of ethnography and the social sciences still just don’t get it.
Romano et al. is the first article I’ve seen that doesn’t show any results whatsoever, not even as tabulated regression coefficients. Yes, they do talk about results in their discussion, but they don’t show any at all.
Large scale grids in industrialized countries need and can accommodate large generating plants allowing intricate and efficient exhaust gas cleaning systems and making the clean use of coal feasible. Wasting precious high quality natural gas in these applications amounts to criminal waste. As Laursen points out, these clean and easy to use fuels are desperately needed in developing countries. Stealing their resources just to avoid coal and nuclear for purely ideological reasons is immoral and ultimately detrimental for climate and environment in spite of any first order local improvements they may yield.
Looking at their figure 1b, the whole of Kohler et al.’s conclusion rests on two, possibly even just one, data points – one (Kahun) an extreme outlier (as figure 1a shows) and the other (Pompeii) the probable artifact of exceptional preservation. That said, the comment by Elliott is still worth reading, even if resting, as it does, on extremely shaky foundations.
Jayles et al. is another very dangerous study about how best to manipulate public opinion through state (or at least oligarchy) controlled mass media. One might glean a little hope from the fact that figure 1a only shows a noticeable effect for suggested errors of more than two orders of magnitude. But in their setup the median values of unbiased estimates and of peer group influences are identical. The interesting question is what would happen if an artificial offset was introduced between the two, a procedure explicitly recommended by the authors.
There is one unexplained error though. By definition the initial estimates before peer group influence can’t be effected by any parameters that influence may take. So the blue model predictions in figure 4 as well as the experimental results within error bounds have to result in exactly horizontal lines. They don’t.
Interestingly the two extreme positions discussed here, contradicting and overreacting, are the only rational ones to take when, as here, you can’t observe the peer group and its distribution and only see an average. You either presume the group to be better informed on average than you are. In that case it will still contain individuals as ill informed as yourself, drawing the true value toward your estimate. So you overcompensate a bit. Or you know yourself to be better informed about the issue in question than the general public. Unless perfectly informed that still leaves you somewhat susceptible to preconceptions wrongly influencing the majority. Now knowing the direction of that influence you may go a little way in the opposite direction, i.e. contradict.
Zum Anfang Übersicht Artikel Home & Impressum