I've encountered a paper before that argued the opposite, that Hitler's rallies didn't appear to actually have much impact on voter preferences, but now I can't

I've encountered a paper before that argued the opposite, that Hitler's rallies didn't appear to actually have much impact on voter preferences, but now I can't find it. Too bad. Firstly, where did you get that abstract? It appears to be selectively quoted and also phrased much more strongly than the copy of the paper I found. I'll give you the benefit of the doubt and assume you got it from a different source to the paper itself: https://poseidon01.ssrn.com/delivery.php?ID=612085086071106084126025107007023078006040020061071033076093083097072088066069082109101003043040006010034084025002117079065094061087094034004090103067083004123003030012051097123066028072092024012027124024103092111073112073002126064123084014094065024&EXT=pdf https://poseidon01.ssrn.com/delivery.php?ID=6120850860711060... The final two sentences which you missed out are rather important in any discussion of the effectiveness of propaganda: "The effect of anti-Semitic propaganda varied depending on the listeners’ predispositions toward the message. Nazi radio was most effective in places where anti-Semitism was historically high and had a negative effect in places with historically low anti-Semitism." Later the paper authors draw the obvious conclusion: "This result highlights potential pitfalls of propaganda: it can backfire, if listeners are unlikely to believe its message." It's very unfortunate that this part of the abstract went missing - one might say it was censored to strengthen a political message ;) I also don't think this sort of pseudo-scientific historical analysis is much use overall, to be honest. No criticism for raising it though, as Adena et al are of course welcome to contribute their views to the debate. But consider the replication and statistical failures in fields like psychology and biology, where experimentation is possible. Now consider that this paper in the field of history, where experimentation is not possible, makes claims like this: "In the absence of radio during the campaign for the September 1930 election, the Nazis would have got additional 4.1 percentage points, i.e., 22.3% instead of 18.2% of the total vote." Two problems with this. 1) The version of abstract you quoted claims "significant negative effect", but the paper itself says the opposite. It says the modelled effects on the outcome were "modest", "not big" and "such a small difference". 1) They specify precise numbers but can't possibly know this: it's wild conjecture made to look scientific by the use of extremely high precision. Modern opinion polling in contemporary elections routinely yields numbers far off the final result - we can't even accurately measure or model things happening right now, but here they're making claims to 3 significant figures about a parallel universe that never happened, on the results of an election that took place nearly a century ago. It makes intuitive sense that campaigning and speeches have some sort of impact on voters, if only to inform them of what politicians believe. But it's not clear that trying to censor such speeches works in an environment with many other ways for people to find out what candidates stand for, like word of mouth.