Significant Outliers

JJS

New Member
#1
I have measured 4 characteristics for ~600 individuals (2 biological replicates per individuals). Some of the characteristics are not normally distributed (Anderson-Darling). I have access to minitab or microsoft excel (2010).

Does anyone have suggestions as to what tests I could use to find significant outliers for each measured parameter?

Thank you in advance.

-JJS
 

noetsi

Fortran must die
#2
What method are you running? For regression, and I assume ANOVA, the easiest way to test for outliers is to look at the standardized or studentized residuals. A value more than 3(some say 2) is usually taken as a sign of a meaningful outlier.

I am fairly certain excel will not do this. I don't know minitab well enough to know if it will or not. I don't think, even with the analytics add on, that excel has any way to do any outlier anlysis.