lol rlumley wtf

I thought I explained myself pretty well...

i can look up some calculation if you give me an hour or so. ill submit a formula to you and youd simply have to find the variables to plug in.

I read something about Bayesian Analysis, but I don't see how it applies to my current quandary...

i've no idea what a T test is, but my father is a statitics lecturer, i can ask him.

Subtract one set of scores from the other, assume normal distribution? Null hypothesis is X-Y=0

Will this work??

Do you mean my brother and I's highest score - my parent's highest score, etc.? I tried that, and it was obviously wrong.

Would you really get the "probability" that you and your brother are better than your parents, or would you actually get the statistical degree to which you and your brother are better than your parents? I mean, are you looking for the numbers a bookie would use to calculate the odds of you and your brother winning against your parents if an outsider were to bet at even odds (without bookie commission)?

I'm assuming you have a lot of scores logged.

We have about 50 scores logged.

I'm looking for this statement: There is an X percent chance of obtaining results this extreme if the variation in the results were due to chance alone.

"I thought I explained myself pretty well..." - Rlumley, you're a fag.

Look at simple statistical analysis, or go hog wild and adjust the ghost maker ranking system to use with the 4 of you...

Simple win/loss will give you a simple statistic, or you can start delving deeper... when you win, how many 'points' (sorry, I don't know bridge) do you win over your parents? Is it complete domination? This could be added to the analysis...

Honestly, it shouldn't be terribly tough to do this analysis...

You could use a chi-squared test for statistical significance, putting in the results of who won.

If I understood Bridge's scoring better I would be able to be much more helpful.

Also, you could model it as a binomial distribution (Each hand is an event, 0.5 chance of winning, hands' results are independent) and find the odds that you win the number of hands you did or more.

Again, understanding Bridge would help.

I think a binomial distribution is a good idea as well.
If you say there are 2 possibilities: 1. you have a better hand. 2. they have a better hand
Then you use binomial distribution to work out the chances of you getting a better hand x out of 50 times. Tell me how many you've won and i can work it out for you.

The problem is I want to account for points. Doing a simple win/loss would work, but some rubbers you win by 2000 , some you win by 20 . These two are obviously different, and thus a binomial distribution wouldn't truly represent it.

The problem if you count points is that the distribution of My brother and I (Bridge is a team game, so only two teams) - My parents and I is hugely binomial, clustered around +1000 and -1000, since you get about a 700 point bonus for winning the rubber. (But it's possible to "win" the rubber and get these points, and still have fewer points than the other team.)

The scoring is incredibly complex. If you want to understand it, google it. It can explain a lot better than I can.