test for normality

[bayenCPG]

I'd like to test a data set for normality before running a bunch of stats on it, but can't find much on this topic. Can anyone suggest a test for normality?

I haven't seen anything, so I'm trying to write my own test VI. My sample size is small (n=10) so the Shapiro-Wilk test seems like the best option. Finding W from the correlation between the samples' values vs order rank is possible, but I don't know how to get a proper P value from W. Maybe a W-P lookup table exists that I can use, but I haven't found it yet.

Any help is greatly appricaited. Thanks!

-Bayen

[chris.b]

There is not an out-of-the-box normality test in LabVIEW... one option would be to compare the histogram of a normal distribution with a histogram of your data.

Histogram VI

 

Message Edited by CXB on 10-29-2008 02:26 PM

[chris.b]

...or the Gaussian Peak Fit VI

[chris.b]

...the Residue and Standard Deviation outputs of the Gaussian Peak Fit might be helpful.  Just some quick thoughts.

[bayenCPG]

Thanks for the suggestions. These are all subjective measures I've played around with, but I'm looking for something that outputs a P value so I can make confidence level decisions.