Suppose I run a t-test and I get the following result:

enter image description here

The p-value is very small, so I reject the null hypothesis. However, I was wondering if I could say something on the sign of two mean and their relative dimension from this data. Also, can I say something with a significant level of 5% or 0.0001%?

  1. You can't decide on the significance level of a test after calculating a p-value. If the prespecified alpha level was 0.05, you can say the test is significant at that level. (You don't reject the null hypothesis just because p-value is small, suppose your alpha level was even lower.

  2. You can't say anything about the sign of the two samples that you put in this test. The 95% confidence interval just tells you the mean of X is lower than Y. (the average of the CI will give you the actual mean difference). But X mean could be 100 and Y mean could be 102 or X mean could be -100 and Y mean could be -98 and the results are the same.

  • $\begingroup$ I see. What about the relative dimension of the two mean samples? $\endgroup$ – qcc101 May 15 '19 at 5:35
  • $\begingroup$ @qcc101 what is "relative dimension"? Do you mean relative direction? $\endgroup$ – AdamO May 15 '19 at 14:24
  • $\begingroup$ Nope, I mean how small or big one is with respect to the other. $\endgroup$ – qcc101 May 15 '19 at 18:31

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.