Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.


Luis A. Afonso
Posts:
4,520
From:
LIsbon (Portugal)
Registered:
2/16/05


WARNING
Posted:
Jan 9, 2014 6:51 PM


WARNING
It is obvious, but amazingly undetected, that the JarqueBera test DOES NOT check if data is Gaussian (Normal) as generally accepted. It performs a quite different task, namely if the SUM of Skewness and Kurtosis after reduction to a squared standard normal is not incompatible with the Normality Hypotheses. Because we are dealing with a sum we are not allowed to state (if so) the two parameters above are correct towards normality.
Note that statistical tests that are based on the Empirical Distribution, like AndersonDarling or Lilliefors, are free to this kind of criticism, of course.
Luis A. Afonso



