Prob > |z| vs. Exact Prob (Wilconxon signed-ranked test)

[Loan Tran]

Hello,
I run a Wilcoxon signed-rank test for a sample of 7 observations in STATA 16.1. The results were:
z = -1.976
Prob > |z| = 0.0482
Exact Prob = 0.1250
Which p-value should be used? Why is there a big difference between p-value and exact p-value?
I would be very grateful if someone can share the source for explaining the different p-values in this test. Thanks a lot.

Details of the Wilcoxon signed-rank test:

[Rich Goldstein]

the following is a quote from the manual entry for this test: "The approximate p-value is based on a normal approximation to the randomization distribution. By default, the exact p-value is computed for sample sizes n <=  200 because the normal approximation may not be precise in small samples"

you are clearly in a small sample situation

[Loan Tran]

So, I should use the exact p-value to intepret the result, right? Btw, could you recommend me papers or books about this matter? Thank you very much, Rich for the reply

[Rich Goldstein]

there are citations in the manual entry and clearly you should use the exact p-value (based on what it says in the manual)