2
$\begingroup$

I have a data to compare two group let say Group A and Group B. The sample size for group A is 20 and Group B is 57. The outcomes variables are normally distributed as according to Shapiro-Wilk test in SPSS. The variance between these two groups are equal (which is showed by leven's equality test). it seemed that all the assumptions of independent T-test are fulfilled.

THE t-test results showed that there is significance difference between the mean of two groups.

I am wondering whether results like this is valid? if small sample size can cause loss of power (type II ERROR high- the probablity of false negative result- probability not reject instead the hypothesis is not true), how if the sample size small and significance still presence?

Thanks.

Improve this question
$\endgroup$
2
  • 2
    $\begingroup$ "The outcomes variables are normally distributed as according to Shapiro-Wilk test in SPSS." -- no. Failure to reject the null doesn't imply the null is true. $\endgroup$
    – Glen_b
    Commented Oct 11, 2015 at 9:02
  • $\begingroup$ One of the main points of significance and p-values is that they have the same meaning regardless of sample size (even though, as you correctly state, small samples tend to have little power). $\endgroup$
    – whuber ♦
    Commented Oct 26, 2015 at 22:32

1 Answer 1

Reset to default
-4
$\begingroup$

What is the p-value you got? You must check the p-value which is the best approach to reject or accept a hypothesis in hypothesis testing. The smaller the p-value the more significant is the result.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ -1 does not adress the concerns and presents a very naive approach to hypothesis testing. $\endgroup$
    – Erik
    Commented Oct 11, 2015 at 13:54

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.