There is much evidence, not only from India, of the link between political authority and its capacity to arouse ethnic tensions, sometimes leading to violence

An Ahmedabad metropolitan court recently rejected a petition challenging the Special Investigation Team’s (SIT) 2011 conclusion that the post-Godhra violence in Gujarat in 2002 was an impulsive reaction of Hindus enraged by the Godhra killings and not a political conspiracy fronted by Chief Minister Narendra Modi.

On the other hand, some in the media, academia, police, and advocacy groups have maintained that the violence was a well-orchestrated anti-Muslim “pogrom,” not a spontaneous “riot.” The Sangh Parivar had diligently planned the violence to help a wilting Bharatiya Janata Party (BJP) in Gujarat, they charged, and disgruntled members among Scheduled Castes (Dalits) and Scheduled Tribes (Adivasis) were instrumentally mobilised to execute the plan. It was suggested that “spontaneity” was a ruse, following ploys adopted by earlier governments, notably the Congress in 1984 and the BJP-RSS in 1992-93.

The first question about the post-Godhra violence that comes to mind, and has indeed been noted by scholars like Achyut Yagnik and Lancy Lobo is: why did it occur so unevenly across the State? Although violence affected many towns, and pervaded rural areas with unprecedented frenzy, many places in Gujarat remained peaceful. By implication, if the post-Godhra killings were an outcome of anger and revenge, why were perpetrators selectively angry or vengeful? If different places experienced different levels of violence, did these spatial variations occur at random or did they indicate a pattern?

These are crucial questions to be addressed for any event of large-scale violence. To do so, I teamed up with Dr. Michael Biggs, a sociologist at Oxford University, to conduct a systematic analysis taking account of all possible factors that might have caused the violence. We meticulously gathered information on the population of Muslims, Scheduled Castes and Tribes on socioeconomic conditions such as unemployment and migration. We measured the strength of the BJP by its votes in the 1998 Assembly election and by the presence of a BJP MLA. Crucially, we compared peaceful with violent places in order to identify specific risk factors associated with violence. Explaining the factors leading to violence by examining only those cases where violence occurred can lead to spurious conclusions. This methodology follows from the exemplary work on ethnic violence in India by sociologists and political scientists such as Paul Brass, Steven Wilkinson, Henrik Urdal and Ashutosh Varshney.

We found persuasive evidence that the violence was not spontaneous. Had it been spontaneous, it would be correct to expect the most outraged people — and, subsequently, the worst violence — in places where the BJP was strong. Even if the party had not taken a leading role, it was its supporters who would be most likely to lash out against Muslims. Instead, after taking into account other economic and social factors, we found that lethal violence was considerably less likely where the BJP was strong (for example, Junagadh, Navsari). Endorsing this pattern, we found that violence was less likely to happen in places with a sitting BJP MLA.

Killings were less likely where the BJP was very weak (for example, Narmada, Dangs). It was in places where the BJP faced the greatest electoral competition, having gained about 35-40 per cent of the vote in 1998, that lethal violence was the worst (for example, Anand, Kheda). Here, the party will face the greatest competition for votes in the coming election.

These findings take into account the social and economic factors that could have led to “spontaneous” violence, and so identify the specific effect of the BJP. This cannot be judged by taking one or two well-known instances of violence.

Violence delivered votes

Violence, therefore, was greatest in places where the BJP faced the greatest competition from other parties. By inflaming anti-Muslim sentiment, it was possible that people who had previously voted for the Congress or other parties would switch to the BJP the next time. Remember that it was later in 2002 — nearly a year after the most lethal anti-Muslim attacks occurred and close to the next election — that Mr. Modi delivered the now infamous speech in Mehsana with obvious references to “Alis, Malis and Jamalis” and their “child producing factories.”

So, did this electoral tactic work? We measured how the BJP’s vote went up or down between 1998 and the Assembly election at the end of 2002. We found a strong pattern: where more Muslims were killed, the greater the boost to the BJP’s vote. For example, the BJP’s vote increased substantially in districts where the violence was severe, as in Panchmahal. In districts without violence, by contrast, the vote declined, as in Surendranagar.

These findings flag a crucial aspect about the post-Godhra violence: that it had an implicit political logic. There is much evidence, not only from India, of the link between political authority and its capacity to arouse ethnic tensions, sometimes leading to violence. Depending on the will of the government controlling local law and order, the resulting violence is either allowed to continue or is stopped. This is particularly the case in multi-ethnic societies because they provide wider scope to change the salience of ethnic issues to suit political elites.

The foil of “spontaneity” in violence has worked admirably in favour of the political elites in India. Rajiv Gandhi used it to justify the “spontaneous” reaction of angry Hindus at the assassination of Mrs Gandhi in 1984; Mr. Modi continues to use it with eclectic references to Newton and puppies. His latest blog post following the metropolitan court verdict, where he expresses “absolute emptiness” at the violence again signals the government’s purported helplessness at the spectacle of perpetrators letting off steam.

The methodology

It is crucial to compare peaceful places with violent places in order to identify specific risk factors associated with violence. Attempting to understand causes of violence by examining only those cases where violence occurred can lead to spurious conclusions. In India, comparative methods have been pioneered by sociologists and political scientists such as Paul Brass, Steven Wilkinson, Henrik Urdal and Ashutosh Varshney. Following in Wilkinson’s footsteps, we systematically analysed the number of killings in the 191 towns and 25 rural areas of Gujarat. We chose to measure violence by the number of deaths because we suspect that newspapers did not report some non-lethal riots. This is not to undermine the significance of non-lethal violence, which often continued longer and resulted in enormous losses to property.

We checked data on the volume of arson, and found a strong correlation with killings. This substantiates our original findings.

(Raheel Dhattiwala is a Public Policy Scholar at The Hindu Centre for Politics and Public Policy. She successfully defended her doctoral thesis at the University of Oxford earlier this month. Findings reported here are from her co-authored paper published in the American journal, Politics & Society.)

More In: Lead | Opinion