t****g 发帖数: 715 | 1 Hi Guys,
There exist so many formal tests for normality (see below). Which test can
be considered "optimal"?
I am an outsider of the statistics literature, but have a simple application
in hand. With n i.i.d data, I want to test for normality. A quick look
at these tests confuses me: there are too many of them.
Which is the best:
D'Agostino's K-squared test, the Jarque–Bera test, the Anderson–Darling
test, the Cramer–von-Mises criterion, the Lilliefors test for normality (
itself an adaptation o | s*r 发帖数: 2757 | 2 the best test is the informal test by plotting and eye
all formal tests have problems | D******n 发帖数: 2836 | 3 agree
【在 s*r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : the best test is the informal test by plotting and eye : all formal tests have problems
| t****g 发帖数: 715 | 4 Is there any paper comparing the performances of them? I do have to pick one
"formal" test, and I think I need to explain why I pick it to the audience.
.
Eyeballing is fine, but it is hard to include it in your written work.
【在 s*r 的大作中提到】![](/moin_static193/solenoid/img/up.png) : the best test is the informal test by plotting and eye : all formal tests have problems
|
|