1
$\begingroup$

I am running a two-sample t-test with unequal variances. However, my standard deviations are so small (as in .03 in some cases) that I cannot get a number greater than 0 for the degrees of freedom (despite sample sizes in the hundreds and sometimes thousands), which makes it impossible to compute the p value.

What do I do here? Is there a different way to compute the degrees of freedom, or is there a way to manipulate the data so that the degrees of freedom formula actually gives me an appropriate number?

Degrees of Freedom is calculated as (s1^2/n1 + s2^2/n2)^2/((1/(n1-1))*s1^2/n1 + (1/(n2-1))*s2^2/n2)

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ Can you give a reference for your formula? $\endgroup$
    – Glen_b
    Commented Sep 16, 2015 at 12:53

1 Answer 1

Reset to default
1
$\begingroup$

The formula for degrees of freedom for Welch's t-test given by Wikipedia is

$$\nu \quad \approx \quad {{\left( \; {s_1^2 \over N_1} \; + \; {s_2^2 \over N_2} \; \right)^2 } \over { \quad {s_1^4 \over N_1^2 (N_1-1)} \; + \; {s_2^4 \over N_2^2 (N_2-1) } \quad }}$$

The formula you give has the same numerator but quite a different denominator; and as a result it isn't unitless (which it should definitely be!). This makes the formula you used get small when you have small variances, which is the cause of your problem.

Use the correct formula and your problem should disappear.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.