### Forums

### Most Recent

- Scotch Gambit
- 10/21/2017 - V Korolkov, Revista De Sah, 1957
- Comments welcome for this game
- Fairly Even Game (Where did I go wrong?)
- Morphy had some stones
- Mean people when playing daily
- goi.bin opening book
- Opening Theory Is Pointless For Most People That Will Ever Play. Why Bother?
- Time for a new defense repertoire
- AlphaGoZero masters Go entirely on its own?

### Forum Legend

- Following
- New Comments
- Locked Topic
- Pinned Topic

Last week I went to a reunion chess match of people I was at secondary school with. We had a great time, playing chess, chatting and watching a tournament (see http://blog.chess.com/ArnesonStidgeley/bournemouth-school-old-boys---2012 if you're having a slow afternoon).

There were ten of us - all born in the 1960s. What struck me was that only three of us had married and only two of those unions had produced children (and no children outside those).

What does that show - if anything?

A sample of 10 people from the same school?

It shows you need more data

it doesnt show anything :)

statistics doesnt apply in every corner of life. or you will imply that if a butterfly move its wings a tornado will appear somewhere else :)

A sample of 10 people from the same school?

It shows you need more data

Hello, Waffle and ionutg

Here's where we need the stats nerds: in any given sample of ten men born in the 60s, what is the number of child-bearing unions (or marriages) that will result? I would have thought that the mean would be around five, with 95% confidence limits of one to nine. Two - as in this sample - seems low.

Our school was a 'state grammar' school - ie, tax-payer funded and entrance based on results of an entrance exam taken at the age of 11. Students were therefore academically able.

Well, yeah that's reasonable I guess, but there's sure to be many groups of 10 who have 0 marriages. Sure if it's something like 6 standard deviations out there, it's unlikely, but how many groups of 10 are there from 1960's schools? Millions I'd guess. So 2 marriages doesn't seem extraordinary... maybe if you'd all been married 4 times each heh.

No matter the skill of the statistician, your calculations can't be better than your data, and I think we'd all agree a sample of 10 is pretty small :)

The minimum sample size in statistics is 30.

It all depends on what the general probability of marriage in general is. "The internet" may claim 70% or 90% on average. The table below shows the distribution (what percentage of the time will a random group of 10 people have so many marriages) based on simple combinatorics for various underlying probabilities (50% change of getting married means look at the "0.5" column. In any case, your outcome appears fairly unlikely. For the nerds, this is just a tabulation of p^k*(1-p)^(10-k)*(10 choose k).