We waded into the murky waters of electoral statistics in order to figure out whether we can prove anything suspect is going on with the vote on the supplementary list. We wanted to find out the following:
- Is there any evidence of a systematic, atypical pattern, with regard to supplementary list voting?
- Can it be proved that either of the main candidates benefitted from supplementary list votes via fraud?
Our investigation did not identify any evidence to support the theory that temporary list voters behave any differently from permanent list voters or that the voting results are massively compromised.
- In part 1 we discovered that, in the first round of the elections, Ponta seemed to do no better or worse in precincts with a greater proportion of supplementary list voters. In the second round, he did worse in such precincts. Iohannis, in contrast, tended to perform somewhat better in the polling stations with a greater share of temporary voters — in both rounds.
- The regression we did in part 2 gave us a better picture of what happened in the second round. On the whole, Iohannis had more to gain from any new voters showing up for the November 16 runoff, although extra voters helped Ponta to some extent too.
- Our regression showed almost no difference between the permanent list vote and the temporary list vote in terms of how the two split between the candidates.No evidence of a widespread and suspicious pattern there either.
- In part 3, we looked at some extremely deviant precincts where our model-based estimate of what may have happened there in the second round given the votes in the first round was way off the mark in predicting the results. While the data from these precincts can be taken as sign of fraud, there remain other plausible explanations for them too.
- In part 4, we demonstrated how one can play with the numbers in order to cast the suspicion of fraud on either side of the political divide by selectively focusing on some cases and ignoring others.
No doubt, the fact that about 10 percent of Romanians who went to the polls did so at other precincts than the ones they are assigned to will continue to arouse suspicions. As we have shown, the results from the over 18,553 polling stations are diverse enough that, if you play with the numbers a little bit, you can cast doubts about the validity of the vote.
So where do we go from here? Can anything be done to ascertain what is happening with all the supplementary list voters and to secure the vote more, so that there is less potential for fraud and more trust in election results? We have a couple of suggestions.
Probably the most modern way in which we can keep tabs on those who vote in other towns than the ones they officially live in is by establishing an electronic database, one that could be accessed in real time by polling station officials. Instead of or in addition to having paper-based „supplementary” lists, which voters have to sign but can only be cross-referenced much later, authorities might set up a common electronic supplementary list, where, once one’s voter’s information was introduced at one polling station, the system would alert the officials if he/she attempted to vote again at another precinct. As a matter of fact, a pilot project based on this very same idea was already implemented in Romania in 2010. Representatives of the Electoral Authority claim that so far there has been no political will from any government to expand the program.
Since an online-based record has somewhat high implementation costs, we also thought of another, simpler manner of curbing repeated voting: using election ink for stamping the fist or a particular finger of each voter to mark that they already voted once. You may have seen images from various countries where, after voting, a person’s finger is dipped in or stamped with ink that cannot be washed off but will naturally fade away soon after election night. Provided it is done at every precinct and using the right materials, this could be a fairly cost-effective way of discouraging „electoral tourism”.
Our methodology vs. other investigative methods
In the recent academic literature on detecting election fraud via statistical analyses of election outcomes, there is a general agreement that no statistical evidence can conclusively identify or prove fraud on its own (see e.g. Mebane 2008). Rather, computations can highlight unusual patterns that may have been caused by fraud or something else, and a closer inspection of other evidence about the precincts marked as potentially suspect by statistics should then be undertaken to see if fraud may have occurred there at all.
Three key methods have been used for such analyses lately. The most famous one examines if different numerals (from 0 to 9) appear in precinct-level results with the kind of frequency expected by Benford’s Law. This method is considered to be the least convincing. Leading specialists in detecting election fraud in fact consider it completely useless because there is no good reason why that law would apply to election results or why fraudsters would violate it (Deckert et al. 2011). In any case, we did carry out the so-called second-significant digit Benford tests on all significant candidates’ votes, both for the country as a whole and separately for each county. We found no deviations from Benford’s Law in the precinct-level results for either the first or the second round of the 2014 Romanian elections.
The second method looks for irregular patterns in the relationship between electoral participation and the vote for different candidates across precincts (Klimeka et al. 2012). The most obvious irregularities – conspicuous reports of (nearly) 100 percent turnout and (nearly) 100 percent support of a particular candidate – frequently occur in places like Russia. They are completely absent from Romanian elections. Unfortunately, a search for less spectacular irregularities in the relationship between turnout and, say, support for Ponta or Iohannis is made impossible by the fact that proper turnout figures cannot be estimated here for the precincts. Why? Because there is a large number of temporary list voters that inflate turnout, sometimes even above the 100 percent mark, for many Romanian precincts. Looking merely at the relationship between temporary list votes and candidate support is, as we have seen, entirely inconclusive about whether and where fraud may have occurred.
Finally, the third method examines if we can notice irregular precincts or regions that depart, in ways that might be caused by fraud, from the kind of relationships that one would expect between voting support for certain candidates and various other variables (see e.g. Myagkov et al. 2009; Wand et al. 2001). This is the path that we took with our regression, which also yielded no suspicious widespread patterns.
Obviously, none of this means that there is absolutely no ballot-stuffing in Romanian elections. We just do not find convincing evidence in electoral statistics that this may be widespread, and asymmetrically favor one political party or candidate over another.
Deckert, Joseph, Mikhail Myagkov, and Peter C. Ordeshook. 2011. „Benford’s Law and the Detection of Election Fraud.” Political Analysis 19 (3): 245-268.
Klimeka, Peter, Yuri Yegorovb, Rudolf Hanela, and Stefan Thurner. 2012. „Statistical Detection of Systematic Election Irregularities.” PNAS 109 (41): 16469–16473
Mebane, Walter R., Jr. 2008. „Election Forensics: The Second Digit Benford’s Law Test and Recent American Presidential Elections.” In Election Fraud: Detecting and Deterring Electoral Manipulation The Art and Science of Studying Election Fraud: Detection, Prevention, and Consequences, edited by R. Michael Alvarez, Thad E. Hall and Susan D. Hyde. Washington, DC: Brookings Institution, pp. 162-181.
Myagkov, Misha, Peter Ordeshook, and Dmitry Shakin. 2009. The Forensics of Election Fraud: Russia and Ukraine. Cambridge: Cambridge University Press.
Wand, Jonathan, Kenneth W. Shotts, Jasjeet Singh Sekhon, Walter R. Jr. Mebane, Michael C. Herron, and Henry E. Brady. 2001. „The Butterfly Did It: The Aberrant Vote for Buchanan in Palm Beach County, Florida.” American Political Science Review 95 (3): 793-810.