Bias and the police

Will more Muslims in India's police forces help combat communal violence?

PRAVEEN SWAMI in New Delhi

"SEGREGATED lives," wrote the scholar Ramesh Thakur in 2002, "lead to ghastly violence."

Much of the public debate on the Justice Rajendra Sachar's enquiry into the conditions of Muslims in India has so far focussed on the degrading economic backwardness of India's largest religious minority and the systematic denial of opportunity to its members. Tucked away among its findings is a comprehensive examination of communal violence: the terrorism used by Hindu fundamentalists to create the apartheid structures that perpetuate these inequalities.

Data on Muslim representation in India's police forces, along with those of members of the Scheduled Castes and Scheduled Tribes, have long been in the public domain. Published each year by the National Crime Records Bureau (NCRB) in its authoritative annual report Crime in India, the figures make for depressing reading.

In all but one State of India, including Muslim-majority Jammu and Kashmir, Muslims are under-represented in the police. And in all but one State for which data have so far become available, they constitute a larger-than-proportional share of prisoners.

The Sachar Committee's findings have added weight to what campaigners have long said: that the under-representation of Muslims in police forces across the country has contributed to institutional communalism and a persistent failure to defend the community's basic human rights.

Based on evidence that emerged from a cluster of large-scale communal riots in the late 1970s amd early 1980s, many have argued that police biases are central to the problem. More representative force structures, the argument goes, are necessary to ensure that the police defend all communities during riots rather than act as the self-appointed sword arm of sectarian interests.

Case closed? In fact, not quite.

Dispassionate analysis of the data does not, in fact, demonstrate a correlation between the under-representation of Muslims in police forces and the scale or intensity of communal violence.

Consider, for example, the case of Tamil Nadu. In 2004, the NCRB reported that just 99 of the Tamil Nadu Police's 88,524 personnel were Muslims - the lowest number of any major State. Muslims, who make up 5.56 per cent of Tamil Nadu's population, are clearly under-represented in its police force. Yet, bar the 1997 Coimbatore riots that claimed 18 lives - a modest figure by India's horrific standards - Tamil Nadu has seen no major anti-Muslim violence.

Moreover, while Muslims are significantly over-represented in the Tamil Nadu prison population, the State's record is better than several States which have better representation for the community among their police forces, such as Maharashtra, Delhi and Gujarat. Clearly, the relatively low presence of Muslims in the Tamil Nadu Police has not, in itself, made it less able to defend Muslims against potential perpetrators of communal violence, or more inclined to punish members of the community for real or imagined crimes.

Interestingly, two of the States with the best record of containing communal violence in the post-Independence period - West Bengal and Kerala - have a poor record on ensuring adequate representation of Muslims in the police. While a little over a quarter of the population of West Bengal is Muslim, the third highest figure after Jammu and Kashmir and Assam, just over 7 per cent of its police force is drawn from the Muslim community. Kerala, where almost 13 per cent of the police force is Muslim, does somewhat better - but this figure still falls well short of proportional representation. Indeed, the percentage-point gap between Muslim representation in the police force and among the general population is worse in West Bengal and Kerala than in Gujarat and Maharashtra. This hammers home the fact that more representative police forces are not necessarily less partisan.

By way of contrast, Andhra Pradesh has succeeded in ensuring more-than-adequate representation for Muslims in the police, but not in containing communal violence or bias. Of the Andhra Pradesh Police's 77,850 personnel, 10,312 are Muslims, making it the only State where the community has a greater representation in service than the population as a whole. Yet, the city of Hyderabad has seen some of the worst and most sustained urban communal violence in India, and the police in Andhra Pradesh have often faced allegations of bias. Despite the high representation of Muslims in the Andhra Pradesh Police, Muslim political organisations have sometimes charged the force with using concern over terrorism as a means to harass the community as a whole.

Similar insights can be drawn from the case of Assam. While almost a third of the State's population is Muslim, only a little over 10 per cent of its police force is drawn from the community. At first glance, this would seem to validate the representation-riots connection, since Assam has witnessed horrific communal violence and Hindu-Muslim antagonisms are at the core of the State's political life. But the data do not bear out the contention that the non-representative Assam Police acts in a partisan manner. Muslims actually form a slightly smaller share of the State's prison population than the general population, in stark contrast to the record in progressive States such as Kerala.

What lessons ought to be drawn from these data? Perhaps the most important one is that some of the discourse on the role of the police in communal violence suffers from the same biases it sets out to critique. There is no evidence to suggest that police forces necessarily advocate the interests of their co-religionists. The largely Sikh Punjab Police ferociously put down Khalistan terrorist groups in Punjab and, contrary to popular myth, the Muslim-majority Jammu and Kashmir Police has long been at the cutting-edge of counter-terrorist operations targeting Pakistan-backed Islamist groups. Similarly, Hindu-majority police forces have often demonstrated their skill at preventing or rapidly terminating communal violence, witness the case of Kerala or West Bengal.

If nothing else, the questions thrown up by these data point to a black hole in the research on communal violence: our knowledge of precisely how police forces operate at times of communal crisis and the decision-making processes that abet or terminate riots and pogroms. Much of the debate has rested on the facile assumption that attracting more Muslims into the police would help create a force more resistant to biases of the kind that are thought to underpin the persistence and scale of communal violence.

Underpinning these ideas is the durable notion that communal biases determine the character of police responses to riots. Writing in 1999, Vibhuti Narain Rai, a police officer, attributed biased policing to "the same predetermined beliefs and misconceptions that influence the mind of an average Hindu". "Not unlike his average co-religionist, an average Hindu policeman too believes that Muslims by nature are generally cruel and violent," Rai argued.

Rai found the police "held the view that apart from being cruel and violent, Muslims were untrustworthy, anti-national, easily influenced by a fanatical leadership, and capable of rioting at the slightest provocation. Further, most policemen believed that riots are initiated by the Muslims. Even when confronted with evidence that it was not in the interest of Muslims to start a riot, the arguments rarely changed."

When they first appeared in the 1990s, Rai's arguments appeared to help explain why police forces had failed to contain decades of anti-Muslim violence. According to material distributed to students of the premier Lal Bahadur Shastri National Academy of Administration, which trains Indian Administrative Services Officers, 1,598 Muslims were killed in communal violence between 1968 and 1980, against 530 Hindus. Recent scholarly work - notably Mumbai-based activist and researcher Asghar Ali Engineer's meticulous monitoring of riots - has shown that the same patterns persisted until at least 2005.

However, post-riot police action was always overwhelmingly directed against Muslims. Judicial investigations of riots as diverse as those of 1982 in Meerut, 1978 in Aligarh or 1970 in Bhiwandi showed systematic anti-Muslim biases in everything from the use of lethal force, patterns of arrests and the treatment of prisoners.

As the Lal Bahadur Shastri Academy's notes point out, the murder of Hindus provokes considerably more police interest than the killings of Muslims. During 1980, 89 Hindus and 275 Muslims died in communal violence. The police arrested 5,457 Hindus and 5,743 Muslims for their alleged role in the rioting. "This shows," the notes pithily point out, "that for each Hindu who was killed in the riots, 5743/89 = 64 Muslims were arrested, whereas for each Muslim casualty [sic; fatality] 5457/275 = 20 Hindus were arrested."

To attribute these actions only to police bias, though, fails to explain why some States have done so much better than others in containing communal violence. There is, after all, no a priori reason to believe that an "average Hindu" constable in Delhi is less hostage to communal biases than his "average Hindu" counterparts in Ahmedabad or Mumbai - and yet, India's capital has seen no anti-Muslim pogrom of significance since 1947.

Similarly, there is no evidence that the police forces of Bihar and Uttar Pradesh have been subject to ideological transformation since the rise of Laloo Prasad Yadav or Mulayam Singh Yadav. Although both States have seen episodic communal violence in recent years, the police have succeeded in ensuring that the clashes did not escalate into the generalised pogroms that these States witnessed regularly two decades ago.

Political action, not police attitudes, then, could prove the key to explaining what determines police responses to communal violence - and to policy interventions that will help to ensure that the coercive resources of the State are used without bias.

Ashutosh Varshney, Professor of Political Science and Director of the Centre for South Asian Studies, University of Michigan, is the author of several books on ethnic conflict, including Ethnic Conflict and Civic Life: Hindus and Muslims in India. He has demonstrated that communal riots are not the outcome of free-floating, generalised Hindu-Muslim antagonism; they are, in fact, highly localised responses to specific political and economic circumstances. Using data for riots from 1950 to 1995, Varshney showed that the violence was concentrated in four States, and, moreover, that 96 per cent of fatalities took place in urban areas.

Violence of this kind is linked inextricably to political power and administrative authority - not just the composition of police forces. Despite the large representation of Sikhs in the Delhi Police, for example, there is no evidence to show that their presence tempered the force's criminal conduct in any way during the pogrom of 1984. On the contrary, even during the Gujarat pogrom of 2002, the police in the Bhuj area were able to ensure that communal violence was significantly contained.

Does this mean that Muslim representation in the police is an irrelevant issue? Far from it. Fuller representation for Muslims in State police forces is one among several instruments needed to build public confidence in the institution and foster dialogue and cooperation between communities and the State's coercive apparatus.

It is, however, important to be aware of the limitations of the politics of representation. Self-conscious efforts to recruit more members of minority ethnic groups or races into police forces in the United Kingdom and the United States have had mixed results; notably, the eradication of institutional racism has not been an inevitable outcome.

Representative policing is a seductive slogan, offering a one-pill solution to an infinitely complex and apparently incurable malaise. At best, however, it is a placebo - not a prescription for building professional police forces.