Hacker News Reader

Comment by hackathonguy | Hacker News Reader

The article doesn't indicate how the villages were sampled.

Suppose a village has a birth rate of 8 babies a year (2 babies every 3 months). There's (roughly) a 50% chance the babies will be male and female, 25% chance that they're both female and 25% chance that they're both male.

That means that if you're sampling from a district that has 1,000 villages (with the average birth rate of 2 babies / 3 months, roughly the birth rate indicated by the article), one quarter of the villages will have 2 male babies - that's 250 villages with 0 female babies born in the last 3 months.

Not making any claims regarding foeticide in India - just saying that the way they chose the villages matters a lot.

View on HN

https://timesofindia.indiatimes.com/city/dehradun/in-3-month... has a bit more info.

You can find the census for Uttarkashi district at https://cdn.s3waas.gov.in/s3ef575e8837d065a1683c022d2077d342...

Page 12 says there are 694 villages and 330,086 people in the district.

Ah! That sort of proves the point, doesn't it? 25% of 694 is 173, which is even higher than 132 villages where there were no female babies. That means there are actually more female babies than you'd predict in advance. Or is my math wrong somehow?

The phrase "in area of India covering 132 villages" seems to imply that it's a compact region, and if that's the case the suspicion of cherry picking would lose foundation.

It would still be possible to find a compact subregion of a large region where no girls were born. However, the fact that it was 132 villages makes this seem unlikely, unless the large region contained many thousands of villages.

132 villages giving birth to only boys for 3 months straight is probably a 0.001% chance of happening naturally

1,000 births (a random guess) with 50% chance of girl, never happening once?

yikes

I know that was probably a random percentage, but for a different perspective a 0.001% chance of an airline crash would still result in an accident each day just because of the sheer amount of air traffic, and that's just for one of the world's more "exotic" modes of transport. In a world of 7 billion people things that should happen extremely rarely still happen every day/month/year only because the dice get rolled so often

1/2^216 is 9e-66, so it's closer to 0.000000000000000000000000000000000000000000000000000000000000009%

I think that 9e-66 is rather irrelevant.

If I flipped 216 coins, the exact sequence that I get back also has a probability of 9e-66.

If I flipped 532 coins, the probability that I get any specific subset that share the same side is the sum of all probability that gives that result.

What we know is that 132 out 694 villages, with the average number of born children for 3 months being 1.6, had zero girls. The other 562 villages with unknown average of children had girls. As such we can conclude that the probability must be the sum of all sequences of births that results in 132 out of 694 villages having exactly no girls born under the period of 3 months.

The probability of the 132 villages is thus undefined as we simply do not have enough data on the spectrum of sequences. Based on the global census of 943 females per 1,000 males we can suspect a bias and create a ball park guess, but there is simply not enough to say an exact number like 9e-66.

> 1,000 births (a random guess)

You don't have to guess, it's right there in the first sentence of the article. There have been 216 births, so about 1.6 per village, which makes having zero female birth statistically normal in each village individually, but possibly not over the whole area if these villages are all in the same area.

no girl in 216 birth is a probability of 1/(2^216) : it's either cherry picking or criminal. The article is ambiguous on the way the villages were identified ("a red zone" ???).

Yeah this seems really important.

Reply on news.ycombinator.com