Lying with statistics – # 1

Here is a recent article  about the  purported rise in murders in Bihar (which of course  was promptly and endlessly  recirculated on Social Media).

The report does not mention when the data is from but dog-whistles that new government had some thing to do with it. Thats kinda funny  because the new govt. was only sworn in the mid of November. So its definitely not 2 months yet.

So, scouring for the data  I found this from Bihar Police website. I assume this is what they were looking at because the last two months for which the data is available (Sept and Oct of 2015) do add up to 578 – the number of murders they mention in the article.

Any way, when stated without context the number of murders do look significant and the article does make it seem as if there has been a significant rise in murders (or “Jungle Raaj” as the article puts it) after the new/old government took over (even though that is clearly not true).

So, I thought I’d put it in context. (Data from here )


See the big jump in murders in September and October …neither do I.

But Bihar is a violent state, every one knows that. These numbers are probably higher than those for other states(an exercise for another time) and hence sound scary. But I don’t see how these are any thing different from what they were in previous months.

Just for comparison,  look at the what they were in 2014.


The numbers look much better in 2015  than they were in 2014. But that doesn’t make a great headline does it ?

Digging in further, I found an interesting pattern in the number of murders (gruesome as it sounds) , and  the specific the time of the year when they seem to rise. More about it in the next post.

Data Scientist by training with diversions into economics, politics,technology and trivia in no certain order.