0
$\begingroup$

Looking for some help. Here's my qq plot. I already deleted the most extreme outliers, but some remain. Also, what about the distributional assumption?

enter image description here

$\endgroup$
2
  • $\begingroup$ Seems like you are dealing with a fat tailed distribution. Note that for such distribution the statistical properties are much worse. $\endgroup$ – Dole May 29 '17 at 18:08
  • 1
    $\begingroup$ Without some hint of what's on what axis this plot isn't much use $\endgroup$ – Glen_b May 30 '17 at 6:55
1
$\begingroup$

I would say it fails to meet normality assumptions because of its divergence at the tails. However, you could use something like an Anderson-Darling test to test for normality more precisely. You can not particularly assume outliers from the qqplot -- but just that some points fray from a normal distribution.

As @Dole noted, you have fat tails and your data could be a t-distribution.

I would also check out this wonderful answer for further qq plot questions: How to interpret a QQ plot

$\endgroup$
1
  • 1
    $\begingroup$ There just as easily could be short tails: it all depends on what the axes mean. I believe we can figure out which axis is which, but that merits some explanation. $\endgroup$ – whuber May 29 '17 at 18:14