|
Abstract:
|
"Serious allegations of fraud have been made with respect to Russia's first competitive party-based parliamentary election in December 1993 - the same election in which Russian's ostensibly ratified a new constitution for themselves. Although charges of fraud are
common in elections, these allegations are especially serious in that the argument here was that over 9 million ballots were fraudulently cast and that the turnout threshold of 50% required to render the constitutional referendum legitimate was in fact not surpassed. These are profoundly important allegations. First, they bring into question the legitimacy of Russia's new constitution and thereby offer its opponents an excuse to suspend its provisions some time in the future. Second, they naturally enough cause us to be suspicious of Russia's December 1995 parliamentary elections. Finally, to the extent that
the same methods for detecting fraud are likely to be applied to subsequent elections, if they revel significant levels of fraud there, they can provide an excuse for canceling those elections or invalidating their results. In this essay, then, we look at the two methodologies employed to detect and measure the extent of fraud in 1993. Without disputing the possibility that fraud was in fact extensive, we conclude that neither methodology as presently developed is adequate to the task at hand. The first, which assumes that
we should observe a linear relationship between the log of the rank of parties and the log of their support at the polls employs a number of ad hoc assumptions and a priori estimates that, in sum, are equivalent to assuming the conclusion. The second method, which looks at the relationship between turnout and the share of the electorate voting for
one party or position versus another, is subject to a number of methodological pitfalls, including aggregation error and the possibility that unobserved variables correlate with both turnout and support so as to render any relationship indeterminate. Nevertheless,
of the two methodologies, the second is the most promising for further development and our critique of it is intended to point the way to the requisite developments."
|