Programming the statistical procedures from SAS

How to write the syntax to perform Lund and prescott tests for outliers in Bioequivalence?

Reply
Occasional Contributor
Posts: 16

How to write the syntax to perform Lund and prescott tests for outliers in Bioequivalence?

Hi All, 

I want to know how to write the syntax to get the lund and prescott test to identify outliers in Bioequivalence.

 

Thanks.

Super User
Posts: 11,343

Re: How to write the syntax to perform Lund and prescott tests for outliers in Bioequivalence?

Posted in reply to MohamedYehia

You might post a reference to what this test is. Sometimes specific fields will use a common statistical test and attach a name to the first in a field that used if for some specific purpose or interpretation rules.

Contributor
Posts: 22

Re: How to write the syntax to perform Lund and prescott tests for outliers in Bioequivalence?

Posted in reply to MohamedYehia

Here is a link to a community thread on someone asking a similar question a few years ago.

Perhaps some of the code / websites linked in the thread will be helpful for you.

 

https://communities.sas.com/t5/SAS-Procedures/Lund-s-Test-for-Outlier-Analysis/td-p/151289

SAS Super FREQ
Posts: 3,752

Re: How to write the syntax to perform Lund and prescott tests for outliers in Bioequivalence?

Posted in reply to MohamedYehia

According to this article on the Lund-Prescott test, the test requires computing the critical value of an F distribution and them flagging standardized residuals whose lengths that exceed that value. You could certainly implement that in SAS by using the OUTPUT statement in a regression procedure to generate the standardized residuals and then writing a DATA step to flag the Std Resid that exceed the critical value.

 

However, I suggest you look at the ROBUSTREG procedure, which can automatically flag outliers and high-leverage points based on robust linear models. See the Getting Started example in the ROBUSTREG documentation, which shows example data, syntax, and interpretation of the results.

 

Ask a Question
Discussion stats
  • 3 replies
  • 197 views
  • 0 likes
  • 4 in conversation