I was looking through the March and July bar exam results and although there isn't yet too large of a sample set it seems as if definite trends are emerging. Overall, about 70% of those who sit for the exam pass it, although not unexpectedly only about 56% pass it who sit for it a second time. Women seem to do slightly better than men on the exam, but only by 1 or 2%. (for an international comparison, about 65% of those who take the New York bar exam pass it).
The more striking breakdowns are by state and the language in which the exam is taken (the same exam, which is entirely multiple choice, is translated into 9 languages, although I don't think anyone has taken it in Oriya yet). The passage rate for those who take the exam in English, is on average the highest of any language group, at about 80%. English is also by far the preferred language to take the exam in. About a third as many people take the exam in Hindi and the passage rate for them is about 60%. Gujarati is the next most popular language. For July these test takers passage rate was 65%, but in March it had only been 35%. The next largest group that takes it is those who do in Kannada, and they tend to do particularly poorly passing only about 30% of the time. You can look at the links for the results of the other language groups, but for most of the rest not enough people take them in these languages for clear statistical relevance to be established.
The statewise breakdown is also revealing, although not as much because of the passage rate. For example, those who took the exam in Bihar have an extremely high passage rate (98% for March). This might seem odd given the socio-economic indicators in Bihar, no matter what understanding might be out there in the popular imagination concerning their penchant for test-taking. Yet, only 106 people took the exam in the state that month. There are about 100 million people in Bihar. Maharashtra, which has about 110 million people, had almost 4000 people take the test. Considerably smaller Delhi had almost 2700 people take it. Amongst the lowest passage rates from states where lots of people seemed to take the test were interestingly in Andhra Pradesh, Gujarat, and Karnataka. Not naturally places one would assume would do poorly. However, it might just be that they have far more law schools and so far more test takers. Still, it's not obvious why some of the poorer states would have higher passage rates even if they had fewer students taking the test (one could imagine that they would just have fewer and worse students).
It will probably require results from a few more exams before one can come up with any real meaningful comparisons between the states, and have a big enough data set that one can say something concretely about all the language groups. Still, it does seem that if your English is good enough that you feel confident enough to take the exam in English that you are likely to do better than others around the country (this makes intuitive sense since the best law schools in India tend to be English only and most key legal resources, such as Supreme Court judgments, are in English). It also seems clear that different states are producing dramatically different numbers of (qualified) lawyers. Perhaps some of the test takers who took the exam in Maharashtra are from Bihar and will return there. However, it seems like certain states aren't currently producing very many lawyers, or for some reason students graduating in these states don't yet feel pressured enough to take the exam. Certainly, many law graduates don't go on to practice law and so won't take the test, but states like Bihar, Jharkhand, and Orissa currently have a statistically insignificant number of people take the exam in their state. It seems important to get to the bottom of what explains this and whether this should be a point of concern.
(Thanks to the folks at Rainmaker, who conduct the bar exam, for originally pointing me towards these statistics)