[hrothgar]

AL78, on 2020-August-02, 07:02, said:

I've use KS to test for likely deviation of a distribution from normality, I've never heard of the G test until now.

Two sample KS can check if two distributions differ from one another

[hrothgar]

pilowsky, on 2020-August-02, 06:55, said:

The correct test is the G-test for independence, not the KS.
But you have to know about statistics to know that.

Just about every reference I've seen for a G test explicitly warns NOT TO USE THIS FOR SMALL DATA SETS

[hrothgar]

Pilowsky

I don't know what the ***** your problem is, but over the course of the last month or so you've gone out of your way to pick fights with half a dozen people on the forums.

In all seriousness, go get some help, because this is just sad.

Acting out on internet forums isn't going to get you what you actually need.

[pilowsky]

You are misunderstanding the difference between sets of continuous (KS) and discontinuous (G-test) groups.
The G-test has been around for more than 40 years.
It replaces Chi-square for larger groups.

[PeterAlan]

On a seating detail, what I have observed in virtual club pairs games is that the partner who Invites in the Registration process gets to sit S / W, and the invited partner N / E: I've yet to notice an exception to this.

[PeterAlan]

AL78, on 2020-August-02, 04:48, said:

The bridge hand HCP distribution is unusual compared to real life distributions in that we know exactly what it is, because there are only a finite number of bridge deals. The distribution is very close to normal, not actually normal because it is bounded by zero and 37. I would think it is close enough to normal that statistical methods could be used that rely on normality as an assumption.

The HCP distribution of pairs of hands is normal, for all 'normal' purposes. It is, of course, symmetrical about its mean of 20: for every pair of hands with X points, the other pair of hands has (40-X) points.

[pilowsky]

One interesting peculiarity that I have recently noticed in the prime area is that pretty much everyone who starts a table starts with themselves in the South seat.
This has the effect of humans always playing three robots. Unless they bring someone with them.
When I foolishly sat myself East a few times my scores deteriorated badly - of course, this could be a complete coincidence, but I'm sticking with South. At least in the prime area.

[barmar]

PeterAlan, on 2020-September-30, 18:51, said:

On a seating detail, what I have observed in virtual club pairs games is that the partner who Invites in the Registration process gets to sit S / W, and the invited partner N / E: I've yet to notice an exception to this.

Yes, that's how it always works. I'm not sure what the reason is, but it's intentional in the code. Not just virtual clubs, all pair tourneys work this way.