1
$\begingroup$

Paired sample t-test and Repeated Measures ANOVA with a dichotomous predictor are two tests that answer the same question, and output the same p-value as far as I can tell.

This question has been asked before, but it was focused on the assumption of both tests. What strikes me more here is that the test statistics are different: one is based on the $t$ distribution while the other on the $F$ distribution. Is there any theoretical way to explain why these two distributions should be close to each other in the case of a dichotomous predictor?

$\endgroup$
1
$\begingroup$

If $X\sim t(\nu)$, i.e. $X$ has a $t$-distribution with $\nu$ degrees of freedom, then $X^{2}\sim F(1, \nu_{2} = \nu)$. It follows that in the case of two repeated measurements, the squared $t$ value of the paired $t$-test is equal to the $F$ value of the repeated measures ANOVA.

Here is an example using SPSS. There are two timepoints with repeated measurements. First the paired $t$-test:

SPSS_pairedttest

The $t$ value is $-0.151$ with $173$ degrees of freedom. Now the repeated measures ANOVA:

SPSS_rmanova

The $F$ value is $0.023$ which is equal to $(-0.151)^{2}$, as expected. The degrees of freedom are $1$ and $173$. The $p$-values are identical.

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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