Various situations call for testing whether an empirical sample can be
presumed to have been drawn from a normally (Gaussian) distributed
population, especially because many downstream significance tests depend upon
the assumption of normality. Statistics::Normality implements some of the
more well-known normality tests from the mathematical statistics literature,
though there are also others that are not included. The tests here are all
so-called omnibus tests that find departures from normality on the basis of
skewness and/or kurtosis.
.
Note that, although the Kolmogorov-Smirnov test can also be used in this
capacity, it is a distance test and therefore not advisable. This, and other
distance tests (e.g. Chi-square) are not implemented here.
Installed Size: 41.0 kB
Architectures: all