Question: precision of the statistic test of shapiro wilks


I need to work with the shapiro-wilks statistical test for the normaility of the population.

I remark that the precision of this test with Maple is not very good.

I try the same test with STATISTICA, STHDA and  XLSTAT. 

For p-value, I obtained 0.3478, 0.3439 and 0.348 with the 3 others softs and 0.4146 with Maple.

For the W, I obtained 0.986, 0.9825 and 0.983 with the 3 others and 0.9761 with Maple.

My sample has 80 elements, risk 0,05.

Is there anybody to explain this difference between the results and perhaps to confirm if it does exist a best method with Maple.

The test : 

pvalue2 := Statistics:-ShapiroWilkWTest(Mat2, level = 0.05, output = 'pvalue')
statvalue2 := Statistics:-ShapiroWilkWTest(Mat2, level = 0.05, output = 'statistic')

Thank you very much for help.

Please Wait...