The Partisan Politics of Electoral Manipulation: Evidence from Romania

In this paper we use a combination of constituency-level electoral data and individual-level data from an original public opinion survey (fielded in October 2012) to identify and explain electoral fraud in the presidential recall referendum of July 2012. Using a variety of diagnostics, including tests of turnout density discontinuities at “round” turnout figures and discrepancies between individual and aggregate patterns of electoral preference changes between the 2012 local elections and the referendum,  we identify a number of patterns consistent with electoral fraud. Additionally, we use a series of regression discontinuity (RD) designs to show that this fraud has clear partisan patterns, being concentrated in localities run by mayors from the governing coalition and where the main opposition party had weaker representation at the local level.