0
$\begingroup$

I collected data for my study to compare with the published norms of two other studies. All three groups are from different populations (based on occupation). I will be using a two-sample independent t-test to compare population A with B, then another t-test to compare population A with C.

Unfortunately, I do not have access to each individual participants' scores for populations B and C, and it does not seem possible to obtain the data (authors have not been responding to email requests).

Is it still acceptable for journal publications to conduct a two-sample independent t-test without doing a Levene's test for equality of variance, simply using the mean, standard deviation, and sample size?

$\endgroup$
  • 1
    $\begingroup$ 1. "acceptable" to whom? 2. to do the t-test you need the sample variances; while I wouldn't do a formal test of different variances if you're concerned that reviewers might complain you can at least discuss their relative size $\endgroup$ – Glen_b Aug 11 '17 at 2:14
0
$\begingroup$

If the variances of the two population groups are different, and the sample sizes for the two groups are different (especially when the sample sizes are not large enough), computing the standard error simply using the square root of the summation of estimated variances for the two groups may cause problems, since the t-statistic constructed with such standard error will not follow a t-distribution under the null hypothesis.

| 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.