Can numbers tell us if any of the parties engaged in ballot stuffing?
In the 2014 Romanian presidential elections, about 10 percent of those who went to the polls within Romania did so “on supplementary lists”, i.e., at another precinct than the one they were registered in. 
Voting on the supplementary list happens not only in large urban areas or university towns, but also in places where there seems to be no reason for out-of-towners to appear in great numbers. The size and geographical diversity of this phenomenon draws suspicion, and voter fraud has been the subject of much news and speculation during these elections, as well as previous ones.
There have been many press and citizen reports alleging a form of ballot stuffing that in Romania is referred to as „electoral tourism” — the practice of the same people voting repeatedly at multiple locations. It is, of course, primarily the mission of election committees and law enforcement authorities to prevent, detect and sanction fraud. The claim that voter fraud could go on undetected and potentially influence election results suggests that these actors do not do their job.
Can analysts and researchers help shed some light on this?
Some have already tried to use electoral statistics to point out what they see as „evident” signs of large-scale fraud. However, the available data can easily lend itself to rash interpretations and conclusions, in the heat of the campaign or the immediate aftermath of the elections. Anecdotal evidence and arbitrarily selected examples cannot be trusted to show how frequent ballot stuffing is and which parties engage most actively in it.
With that in mind, we set out to investigate the following questions:
- Armed with all the publicly available data and various statistical tools, can we identify suspicious patterns with regard to supplementary list votes that can only be plausibly explained by ballot stuffing?
- If so, where, and what political camp benefits more from these suspicious supplementary list votes?
Part one of our investigation lays out what we know and what we do not know about the supplementary list voters. We start with two simple tests to see if there’s any suspicious relationship between turnout and how many votes the main candidates obtained.
In the second part, we delve more deeply into electoral statistics to see how voters changed their behavior from the first round of the election to the second round, and whether the significant number of persons who voted on the supplementary list made a big difference in the final results.
In part three, we examine some precincts that our tools highlight to have had unusual results. One of these places will be Colonesti, perhaps the most famous polling station in Romania.
In part four, we engage in a simple, cautionary exercise to show how the data, if you push it hard enough, can confess to almost anything.
If you prefer to skip all this and fast forward to our main findings, you can find them, as well as some suggestions on how to prevent fraud, in the final section.