Hello Kurniawan Yulianto. It looks like you have a variable with 2 categories. Please tell us the observed frequencies and the expected proportions. Thanks.

Yes, my variable (effectiveness of a medicine) have 2 categories, "Yes" and "No".

• There are 70 observations in "Yes" category and the expected proportions is 0,6.

• There are 30 observations in "No" category and the expected proportions is 0,4.

I have tried that case using binomial test in one-sample nonparametric tests menu and the output have the same problem as above.

Hi Kurniawan Yulianto. I how have v28 of SPSS, and I do not see p = 41,000 as shown in your uploaded image. But I do remember that in previous versions of SPSS, I used to click on Edit > Options, then go to the Output tab and deselect the Model Viewer output. I did that, because I did not find the Model Viewer output particularly helpful! You could try that!

Cheers,

Bruce

By the way, here is some output from Stata using the -bitesti- command to estimate the (exact) binomial test for your data and -prtesti- to estimate an approximate z-test. The latter is equivalent to your Chi-square goodness of fit test (z2 = your Chi-square statistic). (Copy and paste somewhere that you can use a fixed font to make things line up properly.) HTH.

. bitesti 100 70 0.6

N Observed k Expected k Assumed p Observed p

------------------------------------------------------------

100 70 60 0.60000 0.70000

Pr(k >= 70) = 0.024783 (one-sided test)

Pr(k 0.6

Pr(Z < z) = 0.9794 Pr(|Z| > |z|) = 0.0412 Pr(Z > z) = 0.0206

. display "Chisq = z^2 = " r(z)^2

Chisq = z^2 = 4.1666667

EDIT- Adding a text file attachment with the Stata output, because copying and pasting the output above does not work as I hoped it would.

as my experience, what is your problem the program told you to reject the null hypothesis, so it means significant; and when you copy it to word, correct the number

Thank you Bruce Weaver for the solution. I've try that and it's works.

Yes Sahar A Esa Al Shatari, actually the test's result is reject null hypothesis, but I just want to demonstrate it in class. So if the output give unreasonable p-value, it can make my audience confuse. But thank you for your suggestion.