0
$\begingroup$

I have created a random sample of 100 values in SPSS and need to now test for normality with an appropriate goodness of fit test.

I am trying to do a One-Sample t-test via Analyze -> Compare Means -> One Sample T Test and I get a t-statistic of 47.584 and p-value of 0.000. Is this the appropriate test to use? Is there another test that can be done on SPSS that will give better results?

$\endgroup$
5
  • 3
    $\begingroup$ 1. A t-test is not a test of normality. 2. Why are you testing normality? (And why have to? How does the compulsion arise?) $\endgroup$
    – Glen_b
    Apr 20, 2017 at 11:53
  • $\begingroup$ Thank you - this is for a question that I need to answer for a course. I had a feeling what I was doing wasn't right. We had to make a random sample of 100 values and then the question says: test the data for normality using an appropriate goodness of fit test. $\endgroup$
    – glh
    Apr 20, 2017 at 11:57
  • 1
    $\begingroup$ Could possibly the Shapiro-Wilk test be appropriate (doing Analyze -> Descriptive Statistics -> Explore on SPSS)? $\endgroup$
    – glh
    Apr 20, 2017 at 12:04
  • $\begingroup$ That's an appropriate test of normality, yes. Indeed, I'd have suggested that given the context in your prior comment. I guess I'll turn that into an answer. $\endgroup$
    – Glen_b
    Apr 20, 2017 at 13:11
  • $\begingroup$ Thank you so much for your confirmation, I appreciate it $\endgroup$
    – glh
    Apr 20, 2017 at 13:23

1 Answer 1

0
$\begingroup$

A t-test is not a test of normality. (It's used to test equality of two population means or of one population mean with a hypothesized mean.)

You need a goodness of fit test; a widely used test with good power is the Shapiro-Wilk test. It can be found in most stats packages (including SPSS).

$\endgroup$
0

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.