https://phys.org/news/2018-08-hitler-election-campaigner-limited.html > Peter Selb and Simon Munzert base their findings on election statistics from 1,000 counties and districts as well as from 3,864 municipalities. For their analyses, they include information on Hitler's campaign locations, attendance figures at individual events, and NSDAP party membership numbers. They also take into account campaign event appearances of Joseph Goebbels—the second most important speaker for the Nazi movement. By using the statistical method of "difference of differences" (DID), the researchers compare election trends in areas where Hitler gave public speeches with the electoral trends from similar areas where he did not appear. > "We are surprised about how marginal the effect of Hitler's election appearances was, especially when one considers contemporary witnesses and historians who have confirmed his exceptional rhetorical abilities," explain Selb and Munzert. Overall, their data assessment demonstrates only minor electoral effects, which were very limited in terms of geography and time. According to the researchers, Hitler's personal involvement only had a positive effect in the run-off election for the office of German president in 1932. This election took place after an unusually short, intensive and one-sided election campaign—competitor Hindenburg did not make any public appearances—leading Hitler to garner an estimated additional one to two percentage points in terms of vote share in counties and districts in the immediate vicinity of locations where he made a public appearance before the election.