Surviving violence in Delhi: How control and predictability affect decision-making

  • Fall 2021 ∕  Winter 2022
Surviving violence in Delhi: How control and predictability affect decision-making

Aidan Milliff is a PhD candidate at MIT in comparative politics and international relations/security studies. His work focuses on the cognitive, emotional, and social forces that shape political violence, forced migration, and the politics of South Asia. He is a predoctoral fellow at the Institute for Security and Conflict Studies at George Washington University, and a USIP-Minerva Peace and Security Scholar. 

FALL 21/WINTER 22 :: précis Student Feature :: Aidan Milliff
Aidan Milliff
February 10, 2022

In February 2020, Northeast Delhi saw India’s deadliest Hindu-Muslim riots since 2002, and the deadliest riots in the capital since the 1980s. Rioting killed 53 and injured hundreds in perhaps under 100 hours.[1] In the aftermath, some residents erected barricades or gates at the end of their streets. Others left their homes for relief camps in Delhi, or even returned to ancestral villages across India.

The week afterward, I was in West Delhi interviewing Sikhs who had survived the city’s last cataclysmic wave of communal violence, pogroms that occurred after India’s Prime Minister Indira Gandhi was assassinated by her Sikh bodyguards on 31 October 1984.[2] In the four days after her death, over 3,000 people—almost exclusively Sikhs—died at the hands of mobs across India. Some 2,800 people died in Delhi alone.[3]

Like the survivors of 2020, the Sikh survivors of 1984 I interviewed also described a vast range of different strategies they had adopted to try and survive. Some hid in darkened flats for days, or cut their hair and beards—long hair is an important symbol of piety for many Sikhs, which also makes them very visible.[4] Some wielded daggers and knives to defend their homes or neighborhoods, while others left Delhi for good.

During political violence around the world, ordinary people respond to chaos and danger in many different ways. For everyone who participates in a riot or mob, there is someone who has barricaded themselves in a closet. For every family that seeks refuge from insurgent violence or state terror in another neighborhood or state, another family tries to adapt to the danger and stay put. In many instances, there are not really noticeable differences between the people who choose radically different strategies when they confront danger.

Take an example from 1984 in Delhi: Two women in similar neighborhoods in Southwest Delhi faced nearly identical mob violence on the same day, but responded in very different ways. One woman from the neighborhood of Palam saw her father taken from their home and set on fire (probably with phosphorous powder). Another woman, only 3km away in a neighborhood called Sagarpur, witnessed a mob take her father, brother, husband, and son from their home and beat them to death. The woman from Palam immediately left and resettled in Punjab. The woman from Sagarpur remained—she lives in Delhi today.

A new theory of civilian survival strategies

How do we make sense of these choices? In my research, I study the experiences of violence survivors, recorded in original interviews and oral history archives, to understand the decisions they made when facing violence. I use the stories of Indian Sikh survivors of 1984 to test a new theory, situation appraisal theory, that offers a new way to understand the forces that shape common people’s responses to violence across different religious groups, different types of violence, and different countries.

Even people who are very similar to each other often perceive and interpret a given violent situation in different ways. Different perceptions and interpretations in turn motivate people to respond differently to threats of violence. Two particular interpretations (I call these interpretations situational appraisals) are important for understanding people’s behavior: a person’s appraisal of how much control they have over the outcome of violence, and their appraisal of how predictable the evolution of violent threats will be in the near future.

People who assess control and predictability differently tend toward different behaviors to keep themselves safe: People who feel uncertain about the future and believe they have no agency to mitigate threats to their safety (no control) tend to flee from violence. People who feel that they have control, but aren’t confident in their ability to predict the future try to fight back against threats. People who feel the future is more predictable and understandable either try to minimize their exposure to threat by “hiding” or they try to engage constructively with the sources of danger, depending on how in-control they feel.

These appraisals guided the choices of many survivors of the 1984 violence in Delhi. In my own interviews with survivors in Delhi as well as many survivors who migrated to northern California, and also in mixed-methods analysis of an archive of over 500 video-taped oral histories of the violence from the 1984 Living History project,[5] I find that people’s assessments of control and predictability are strongly connected to the strategies they end up selecting.

People’s assessments of control and predictability are part of an effort to make sense of chaos when facing political violence. And in the haze of conflict, not everyone makes sense of chaos in the same way. The women from Palam and Sagarpur, for example, saw their situations differently. The woman from Sagarpur believed she understood how the mobs worked. Before her father died, he described a theory of the violence: mobs were targeting those who were visibly Sikh (mostly men), and were especially brutal toward people who tried to fight back. This set of “rules” made the violence more predictable—though no less traumatic—for the woman from Sagarpur. Her perception of predictability led her to stay put in Delhi, where she still lives. The woman from Palam felt no such sense of predictability. Members of her family were basically tricked: her father was killed by people who had promised him safety. As soon as possible, the woman from Palam left the neighborhood, and Delhi all together. Even these two similar-seeming women, facing nearly identical violence on the same day, interpret their circumstances differently, and follow those interpretations to decide what they need to do to survive.

Informing security policy

What are the political implications of these deeply personal choices about survival during political violence? And how can a theoretical framework that focuses on variation in individual perception help inform policy to protect civilians during violence? This research has two lessons for policy-making in the future. First, once individual choices to fight, flee, or hide aggregate up, they can become massive policy challenges like ethnic cleansing,[6] refugee flows, or deepening cycles of violence. I trace the roots of these phenomena back to individual perceptions, suggesting that policies to reduce escalation or encourage resilience-in-place ought to focus on making civilians’ experiences of conflict less uncertain, and ought to encourage civilian agency.

Second, my work suggests a new answer to a puzzling question in US security policy: why have efforts to improve the lives of people living in conflict affected states not led to more stability? The US has spent $145 billion on reconstruction in Afghanistan,[7] $60 billion in Iraq,[8] plus billions in other states, and even more billions helping people who are already displaced. My research suggests that money, which mostly focuses on improving material conditions and physical security, might be mis-directed. I show that whether a person ends up displaced, involved in violence, or trying to adapt depends heavily on their perceptions about how much uncertainty they feel during violence —perceptions are often more important than relative exposure to violence, or how much development aid they get.

Predictability and control are important determinants of people’s choices, so it is important for policy makers to target those perceptions directly if they want to shape civilian behavior. Ultimately there might be circumstances where helping people feel more certain about the future or more agency in their own lives could be an appropriate—and efficient—way to increase resilience and adaptation during violence.


The Delhi riots in February 2020 never spilled across the Yamuna River into the center of the city, but rumors and fear spread all the way to Punjabi and Sikh neighborhoods in the city’s West, like Dwarka, Janakpuri, and Tilak Nagar. In my interviews only a week after those February riots, people described the scenes of 1984 returning to their minds when they heard rumors that the 2020 riots were coming to West Delhi. Some prepared to correct “mistakes” they made in 1984 like not having access to weapons. Some cancelled weekend plans and kept their children home, fearful of having a son or daughter stranded across the city during a riot. Others talked about fleeing: what relative could take them in if the violence started? As in 1984, everyone saw something slightly different in the facts they could gather and made different plans accordingly.

This article is adapted from an essay that appeared in the Hindustan Times on 30 September 2021:  “From 1984 to 2020, Delhi’s Response.”


[1]Northeast Delhi Riots, One Year Later: Wrong Place, Wrong Time,” The Indian Express, March 2, 2021.

[2] Virginia Van Dyke, “The Anti-Sikh Riots of 1984 in Delhi: Politicians, Criminals, and the Discourse of Communalism,” in Riots and Pogroms, ed. Paul Brass (New York: Springer, 1996), pp. 201-220.

[3] Report: Justice Nanivati Commission of Inquiry (1984 Anti-Sikh Riots), (New Delhi: Government of India Ministry of Home Affairs, 2000). Sikh activists maintain that these numbers vastly under-count the true death toll.

[4] Bhupinder Singh, “The Five Symbols of Sikhism,” Sikh Formations 10, no. 1 (2014): 105-172.

[6] Chaim Kaufmann, “When All Else Fails: Ethnic Population Transfers and Partitions in the Twentieth Century,” International Security 23, no. 2 (1998): 120–56.

[7] John F. Sopko. What We Need to Learn: Lessons from Twenty Years of Afghanistan Reconstruction, (Arlington, VA: Office of the Special Inspector General for Afghanistan Reconstruction, 2021).

[8] Stuart Bowen. Learning from Iraq: A Final Report from the Special Inspector General for Iraq Reconstruction, (Arlington, VA: Office of the Special Inspector General for Iraq Reconstruction, 2013).