Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

- JMP User Community
- :
- Discussions
- :
- Details about robust standard deviation from "Distribution" module?

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Details about robust standard deviation from "Distribution" module?

Dec 4, 2013 4:07 PM
(2217 views)

The JMP documents guide me to the book by Huber & Ronchetti (2009). But it's still not clear for me how to calculate the robust standard deviation.

Thanks!

3 REPLIES 3

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Details about robust standard deviation from "Distribution" module?

I´m also a bit puzzeld about the robust Standard Deviation in the Distribution Platform.

As far as i´m concerned, the robust standard deviation is someting like IQR/1.35. This is certainly not the case here:

Can someone explain to me, how 2E19 should be robust, if IQR is 8E-12 and median is 2E-10??

Thanks!

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Details about robust standard deviation from "Distribution" module?

JMP employs the same method as RobustReg which is part of SAS. Details can be found in the RobustReg documentation in the section titled "Algorithm".

https://support.sas.com/documentation/onlinedoc/stat/142/rreg.pdf

This is page 8184 in the upper left corner and page 42 in the PDF document.

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Details about robust standard deviation from "Distribution" module?

Thanks for the clarification.

However, I will stick to to the simple IQR/1.35 calculation, since the RobustReg procedure may return misleading results, if there are extreme outliers. That is at least the case for our data sets.

Article Labels

There are no labels assigned to this post.