/sci/ - Science & Math

>>16054293
>Childhood_Friends.jpg

>>15037179
incoming autism warning....
>I don't think that's what's been agreed to,
If we provide birth order info it goes to 1/2 and others have agreed. I was merely putting that in context to an entire population.
>what's been agreed to is that if we know a girl was born first, or separately we know a girl was born second, those two different populations of families would be 50% likely to have a boy as the other child
You can call them different populations but that doesn't have any significance since they're all part of an original population. If we have 1 million 2 child families with random distribution of male/female we'd expect roughly 125k of each of below combinations. I'll rearrange what I wrote earlier for clarity (prime denotes first born thats whats rearranged):

known girl is 1st
A' girl - B boy
A boy - B' girl

known girl is 2nd
A girl - B' boy
A' boy - B girl

Family's w/ g-g & known girl can be 1st/2nd
A' girl - B girl
A girl - B' girl

excluded:
A' boy - B boy
A boy - B' boy

in our discussion we exclude b-b of course so we're left with 750k families who all have AT LEAST 1 girl. It was agreed if we reveal order an individual family has 1/2 chance of g g. With 750k families we'd expect 375k g-g s (an incorrect amount)

If we aren't given birth order we'd have:
A girl - B girl
A girl - B boy
A boy - B girl

And still have 750k families with at least 1 girl, each combination should roughly be 250k. It was agreed above each of individual family has 1/3 chance of girl girl if no order is given. With 750k families we'd expect 250 g-g s (the correct amount)

The number of families did not change and these are the same families out of that 1 million population, but the odds for all 750k individual families did change merely by gaining information. It doesn't make sense. Thus I don't see how pointing out there are different sub-populations in the former has significance.