There has been much public debate about a recent research paper by Ashoka University’s Sabyasachi Das which claimed to have found ‘objective evidence’ of ‘democratic backsliding in the world’s largest democracy’. The paper emphatically states this happened ‘in the form of electoral manipulation’ in India by the BJP and claims ‘results consistent with fraud’ while studying the BJP’s vote-share density.
These are serious claims that deserve to be looked at closely. Predictably, the paper has already ignited a vociferous debate between the echo chambers of the Left and critics of the ruling party, who see in it a vindication of long-held beliefs on BJP’s electoral ascendance, and the Right, which sees a conspiracy to besmirch the legitimacy of PM Narendra Modi and the BJP’s poll triumphs. As the debate blew up, Ashoka University distanced itself, saying it was ‘dismayed by the speculation and debate’; that it ‘encourages research but does not direct or approve specific research projects by individual faculty members’ and ‘social media activity or public activism by Ashoka faculty, students or staff in their individual capacity does not reflect the stand of the University’.
Das presents several datasets and dozens of charts as ‘evidence’ of what he calls ‘significant irregularities’ and ‘electoral fraud’. These are big claims. How robust are they? And does the evidence stack up?
The simple answer is No. Here are four reasons why:
REASON 1
To begin with, the crux of Das’s argument boils down to the basic assertion that the BJP, in 2019, won more seats in closely contested contests that the historical pattern would otherwise have shown.
Das applies what is known as the McCrary test and uses the regression discontinuity design (RDD) method to conclude that the BJP won ‘disproportionate’ wins in such close contests where the margin of victory was low. That these wins were out of whack with the historical pattern, on this measure, becomes the starting point of the whole subsequent argument on whether they were acquired fair or square or via fraudulent means.
Shorn of all the statistical sophistry, the paper’s own findings show the number of such “excess” wins by the BJP in parliamentary constituencies, where the win margin was less than 5% of votes, as 11 seats. Fifty-nine Lok Sabha seats in 2019 were won by a margin of less than 5% vote-share. The BJP won 41 of these contests. Calculated on a 50% benchmark chance of winning, statistically, BJP won 11 more than the historical pattern would have predicted.
Put simply, the whole basis of the argument is predicted on the pattern shift in these ‘extra’ 11 seats (the range is 9 if the threshold is kept at 3% or 18 seats if the threshold is 7%).
Now, just because something has happened for the first time in these seats, does not mean that it happened because of fraud. Even if, for the sake of argument, if we put a question mark on these constituencies for a second, in an election in which the BJP won 303 seats (in a house where the halfway mark is 272), the BJP still won an emphatic victory on the back of a Modi wave that broke the normal rules of politics at multiple levels. For example, the BJP won 31 of 47 reserved ST seats in 2019.
REASON 2
Of 41 seats that BJP won among the 59 seats won with less than 5% vote-share that the author lists, about half were won in non-NDA ruled states.
Twenty-two BJP wins came from 29 such close contests in BJP-ruled states, while 19 of the party’s victories came from of 30 such low-margin seats in non-BJP ruled states.
This is important to point out since the paper specifically co-relates poll result outliers to data on deployment of observers from BJP-ruled states in terms of allegations of manipulations in counting. Das asserts that in ‘PCs [Parliamentary constituencies] won by BJP, the fraction of counting observers who are SCS [State Civil Service] and come from BJP rules states positively predicts the extent of turnout data discrepancy in the PC; in PCs that BJP lost, no such relationship holds’.
The charge is that officers from BJP-ruled states had something to do with electoral manipulation in polling booths they were deployed to. Yet, the fact is among opposition-ruled states at the time, among the 59 close contests that Das lists, the BJP won:
- 6 of 10 in West Bengal
- 6 of 10 in Odisha
- 2 of 4 in Chhattisgarh
- 4 of in Karnataka.
- Overall, BJP won 63.3% of these 59 close contests in non-BJP states and 75.8% in BJP ruled states. (After correcting for Andaman and Nicobar and Dadra and NH, which were erroneously listed in the paper as non-BJP ruled states, and are in fact centrally administered union territories. This has been pointed out by @saiarav)
To believe that biased poll officers from other states could have manipulated local results in Mamata Bannerjee’s West Bengal or Navin Patnaik’s Odisha defies logic and common sense.
REASON 3
The paper argues that manipulations occurred in voter registrations. As evidence, it says that statistically, the growth rate in the electorate vis-à-vis 2014 in seats BJP won narrowly was 5% lower than the growth rate in ones it narrowly lost. The allegation is that Muslim voter names were deleted, and hence BJP won close contests.
Yet, as has been pointed out, by at least one analysis, there is nothing abnormal in the growth rate of the electorate in seats BJP won narrowly among the 59 close contests listed in the paper.
At the national level, the electorate grew by 8.7% between 2014 and 2019. In contrast, among the seats in question, it grew by 8.8% in seats BJP won narrowly, and 11.6% in the seats it lost. Beyond the statistical modelling, the author does not offer any new systemic ground evidence from any constituency or polling booth of systemic fraud.
REASON 4
Just winning ‘excess’ seats does not necessarily imply fraud, as Das himself writes. The wins could also be because of several other factors, including better campaigning (precise control). Yet, Das chooses to privilege ‘fraud’ and ‘manipulation’ as the better explanation without any concrete ground evidence. The only basis he uses for this measure is one – just one question – from the National Election Survey 2019, which asked voters ‘Did a candidate/party worker of the following parties come to your house to ask for your vote in the last one month?’ Basically, he finds that because BJP home visits, calculated in this survey, were not significantly larger than those by other parties therefore campaigning couldn’t have been the reason for more wins.
The fact that only this reason is used to rule out ‘better campaigning’ beggars belief. Better election results can come from a multitude of factors- candidate selection, caste matrixes, social media, the nature of political messaging, trust in leaders etc. None of these is factored in.
Big claims like ‘electoral fraud’ and ‘manipulation’ must be substantiated with proof. This paper, based on statistical modelling, does not provide a shred of any new ground evidence. To draw such sweeping conclusions on ‘democratic backsliding’ without robust proof is clickbait headline-hunting or ideological pamphleteering rather than robust academic research. The Election Commission of India has a robust reputation and to so far as to even indirectly impinge on its efficiency and integrity needs not only more by way of evidence but better reasons.
As the psephologist Jai Mrug, CEO of Voters Mood Research, says, ‘This is more like a project data exercise to create a hypothesis which you want to research, not the outcome of an experiment conducted with rigour. There is no direct corroboration with regard to voters of anything on the ground. In fact, the results reflect the vagaries of the first-past-the-post system’.
In other words, the headlines, keywords and claims in the articles are misleading and simply do not hold. Applying advanced statistical methods applied from economic may be a great advantage in wider social science research but however well the modelling is done, the interpretations will always be misleading if not co-related with ground understanding of politics, common sense and logic. And if they are made to fit a pre-set conclusion.
Nalin Mehta, an author and academic, is the Dean of School of Modern Media at UPES University in Dehradun, a Non-Resident Senior Fellow, Institute of South Asian Studies at the National University Singapore, and Group Consulting Editor, Network 18. He is the author of The New BJP: Modi and the Making of the World’s Largest Political Party. Views expressed in the above piece are personal and solely that of the author. They do not necessarily reflect News18’s views.
Comments
0 comment