MODELING ELECTORAL FRAUDS

In the article the author tests the efficiency of the existing methods of diagnosing electoral frauds and justifies an alternative approach towards solving this task. The method offered by the author that is based on the values of U-intersections and b-coefficients of the linear regression models of how share of voices received by candidates depends on turnout allows to rather accurately determine the volume of fraud data and therefore, the real level of support for candidates. This method provides an opportunity not only to proceed with research in the sphere of electoral frauds, but also to answer a broader theoretical question: is the enormous margin by which authoritarian incumbents win over their closest rivals a consequence of the public support for the corresponding regimes or is it trivial frauds that pervert expression of voters' will?