okay never mind
great thanks
edits to Kent's dat file
> I used a different formula than Kent for the pooled variance, but in the case of variances drawn from equal population sizes I don't think it makes any difference.
Yes, if the sample sizes for various targets were different, you'd use degrees of freedom to weight the mean variance. That is, if
s(i) = apparent standard error of angles to Target i
and df(i) = degrees of freedom of s(i), i.e. n-1 where n is the number of angles measured to Target i
Then the pooled estimate would be:
SQRT [ ( s(1)^2 x df(1) + s(2)^2 x df(2) + ... + s(i)^2 x df(i) / (df(1) + df(2) + ... + df(i)) ]
But since all of the values s(1), s(2) ... s(9) have the same values of df, they can all be treated as having unit weight when the pooled estimate is formed.
TS-A and TS-B for Conrad
> I ended up with the angle stdev for instrument A at 2.40" and instrument B at 0.46".
Why do I think that TS-A and TS-B have serial numbers whose first two digits and last two digits are identical?
TS-A and TS-B for Conrad
> Why do I think that TS-A and TS-B have serial numbers whose first two digits and last two digits are identical?
Ha ha ha. Why, I'm absolutely quite sure I have no idea!
