oldschoolcarto
01-17-2011, 04:03 PM
Hello all:
I just received a reference text in quality control – Statistical Quality Control by Montgomery 6ed. In describing the (2 sample unequal variance) Welch t-test, the author notes the formula for the Satterthwaite degrees of freedom estimator as (for lack of a better way): ((var(x1)/n1 + var(x2)/n2)^2 / ((var(x1)/n1)^2/n1-1) + ((var(x2)/n2)^2/n2-1))) which is the standard method I’ve always used. However, the author subtracts another 2 degrees of freedom. Why take away an extra 2df, isn’t that taken care of the denominator? Has this formula changed since the Dark Ages when I took my first stat class?
Thanks in advance.
I just received a reference text in quality control – Statistical Quality Control by Montgomery 6ed. In describing the (2 sample unequal variance) Welch t-test, the author notes the formula for the Satterthwaite degrees of freedom estimator as (for lack of a better way): ((var(x1)/n1 + var(x2)/n2)^2 / ((var(x1)/n1)^2/n1-1) + ((var(x2)/n2)^2/n2-1))) which is the standard method I’ve always used. However, the author subtracts another 2 degrees of freedom. Why take away an extra 2df, isn’t that taken care of the denominator? Has this formula changed since the Dark Ages when I took my first stat class?
Thanks in advance.