Can we get a definitive verdict in what concerns the Jarque-Bera Test?
The main question: Can we assert that a sample could be normal simply because the test-value falls in the acceptance region? I do not agree . . . I would say instead: What we can state is that the sum of the Skewness Coefficient and that of Excess Kurtosis each one reduced (by its sample standard deviations) are so that are in conformity with that the normal samples have. NOTE this statement is about sums, never about each quantity *PER SE*, one of each (not both, of course) could be insupportably high (one tail test) by the tested hypotheses of normality.