In India’s lower judiciary, the absence of in-group bias
Much of the debate around judicial fairness in Indian courts has focused on Supreme Court or high court judges. The expansive lower judiciary, comprising more than 7,000 district, sessions, and subordinate courts, has largely been missing from this ongoing debate.
To fill this gap, we examined data on over six million court cases filed under India’s criminal codes between 2010–2018 across the country. Women represent 48% of the population, but only 28% of lower court judges. Similarly, India’s 200 million Muslims represent 14% of the population, but only 7% of lower court judges. We wanted to test whether these disparities lead to worse judicial outcomes for women and for Muslims. Measuring bias in criminal proceedings is not straightforward. It is not sufficient to look at conviction rates for different social groups, because other inequalities may result in differential outcomes. For example, it is hardly surprising, and not evidence of bias against men, that only 23% of individuals charged under the penal code are women.
Instead, we applied a statistical test that has uncovered judicial bias in courts in several other countries. We asked whether women or Muslim defendants on average get different decisions on their cases, depending on whether their assigned judge is a man or a non-Muslim. This test is credible, because neither defendants nor prosecutors in the judicial system have much control over the demography of the judge who hears a case. In fact, the legal system explicitly prohibits the practice of shopping for judges, noting that it “must be crushed with a heavy hand”.
This allowed us to compare two defendants who are alike in every way — of the same gender and religion, charged under the same section of the penal code, in the same month, with cases heard in the same district court — whose only difference is that one had her case heard by a female judge and the other by a male judge.
There is evidence from around the world that judges discriminate in favour of litigants who share their identity. For example, in Israeli courts, Arabs whose cases are heard in front of Arab judges get more favourable decisions than Arabs whose cases come before Jewish judges. In the United States, women are more likely to win sexual discrimination lawsuits when their cases are heard by female judges.
We expected to find the same kind of bias in India’s district courts. Our own prior research shows that Muslim men are worse off than Dalits and Adivasis in terms of upward economic mobility. The 2006 Sachar committee reported that Indian Muslims fare worse on every human development indicator, are over-policed, and are often killed in botched police encounters. Given a climate of prejudice, we had reason to suspect bias in judicial decisions.
To our surprise, across six million criminal cases, we found virtually no evidence of in-group bias among judges on either gender or religion. Male defendants did not get better outcomes when they were assigned to male judges, nor did female judges favour women. Equally, the judicial outcomes of Muslims were virtually identical whether their cases were assigned to Muslim or to non-Muslim judges. We found similar results whether we looked at case outcomes (such as acquittal or conviction), or at case processes, like delay.
We did find some evidence that male judges are more lenient overall than their female colleagues. But this is not evidence of bias, because male and female defendants get equally lenient treatment from male judges.
Our study is just one step toward building a body of evidence on the performance of the Indian judicial system. We found an absence of in-group bias in one context, but it does not rule out other forms of bias in the legal system as a whole. We did not have data to study in-group bias along caste or socio-economic lines. We also cannot rule out unequal treatment of Muslims or women in the system as a whole. Our study does not provide any reason to doubt the existence of widespread unfair treatment in other domains.
However, our research is optimistic about lower court judges, at least with regard to this important form of bias. If district and lower court judges display impartiality toward defendant identity in their rulings, they should be celebrated for it, even as other problems are rightly condemned.
Similar studies need to be done on police and prosecutor behaviour, and on outcomes in other courts. Our research was only possible because of the laudable efforts of the courts to make data on cases publicly available. The government should make available detailed microdata on all the other stages of the criminal justice process, such as police stops and arrests. If justice is served fairly, there should be nothing to fear from transparency. In the words of the great American jurist Louis Brandeis, “Sunlight is said to be the best of disinfectants; electric light the most efficient policeman.”
Sam Asher (assistant professor of economics at Johns Hopkins University) and Paul Novosad (associate professor of economics at Dartmouth College) are the founders of Development Data Lab, where Aditi Bhowmick is a research associateThe views expressed are personal