Talk:Welch–Satterthwaite equation

Underlying normal?
Some other statements of the Welch–Satterthwaite equation state that the underlying distribution must be normal. I can also see mention of normality in the abstract of the referenced paper by Welch. However, the article doesn't mention any constraints on the underlying distribution. Alex.sayers (talk) 09:26, 27 February 2018 (UTC)

Article is not very exact
The following holds:

$$ \nu_{\chi'}\frac{ \chi'}{E[\chi']} $$

follows approximately a chi-square distribution with $$ \nu_{\chi'}$$ degrees of freedom.

k_i
The article talks about $$k_i$$ which is nowhere defined. Can I take them arbitrary numbers? Frankmeulenaar (talk) 11:07, 19 April 2012 (UTC)
 * Yes. JMiall  ₰  12:22, 19 April 2012 (UTC)

nu
The expression given has nu, which is nowhere referenced in the rest of the article. —Preceding unsigned comment added by 98.118.142.50 (talk) 20:25, 22 March 2011 (UTC)
 * "For $n$ sample variances $s_{i}^{2} (i = 1, ..., n)$, each respectively having $ν_{i}$ degrees of freedom..." $ν_{i}$ is the degrees of freedom for sample variance $i$. I've changed the font as the nu may have looked more like a v, depending on your browser font settings. --Qwfp (talk) 22:25, 23 March 2011 (UTC)

signs of a?
Hello,

Does this mean that if I compute a difference

$$ f(x_1, x_2) = x_1 - x_2 $$

the effective degrees of freedom are zero? — Preceding unsigned comment added by 75.150.66.10 (talk) 19:23, 10 January 2012 (UTC)