and I don't have the t-critical values anywhere in my house... can you calculate the Confidence interval of 90% and 99% with an N of 2 and xbar equal to four hundred to see if it captures 415(xbar 2(aka HKS gain). I imagine with a small N the standard of error will make this fairly hard, with a large gap between the lower and upper limits of the confidence interval, but I'd still like to see the results. I'd do it, but I don't have a tcritical chart 15/Sx CI of 99%=400+-tcrit(Sxbar) Sxbar=S^/sqrtof 2 S^=sqrt((E(summation)x^2 - ((Ex)^2)/2))/1)) Oops, can't calculateSxbar with an N of 2, you get 0 which obviously screws up a confidence interval. **If N was 3(since two can't be used-obviously there are not really three dynos- so this is just for arguments sake) S^=346 So CI%+-(tcritical value of 10% or 1% with N of 3) * 346 I have a feeloing these numbers of variation are going to be huge and not very useful.
 |