What else is like...

...the effect of the economy on election results. That is, there is a pretty broad consensus in the political science literature that the "fundamentals" determine electoral outcomes, at least in the US where our long election periods allow for campaign effects to wash out, leaving variables such as economic growth, unemployment, and party identification as the principal determinants of election results. But rarely are subsequently perceived consequential elections described in terms of the  economic growth of the previous administration (although there are exceptions, such as 1932).

What we get is the structure/agency disconnect, in which prominent and even dominant narratives (history trumps political science in discourse)  stress agency, but statistical analysis suggests that agency is much less important than the structural determinants. Popular perception, amongst practitioners and non-practitioners of the dark arts of politics alike, are in serious tension with the statistical evidence.

Question of the day: in what other domains does this general trend hold, in which there is a pretty strong consensus by people who study the field that everyone who actually does this for a living (i.e., the Obama advisors who believe a "choice" option rather than a "referendum" option will save the democrats come September)  is mistaken?