Re: significance test for non-normal data

Good afternoon,

Another possibility is to subject your data to a power transformation,
such as ln(x) or x^lambda. This will have the effect of making the
distribution of values more symmetric, and the resulting symmetry may
allow you to use the ttest. This will only work if your data a unimodal
(i.e., a single humped distribution). If taking the natural log does not
produce the desired level of symmetry, try raising all values (x) in
your distribution by some power (lambda; either negative or positive
lambda). Then re-plot the data using a histogram of similar graphical
technique to examine the resulting level of symmetry.

Otherwise, the R Statistical Language has a large range of formal
statistical tests that are non-parametric, including the
Kolmogorov-Smirnov two sample test available in a future release of NCL.
Another very good non-parametric test is the Wilcoxon-Mann-Whitney test,
which is implemented in R as wilcox.test function.

http://en.wikipedia.org/wiki/Mann-Whitney_U
http://stat.ethz.ch/R-manual/R-patched/library/stats/html/wilcox.test.ht
ml

Hope this helps,

Daran

> Message: 14
> Date: Tue, 17 Sep 2013 08:40:30 -0600
> From: Dennis Shea <shea@ucar.edu>
> Subject: Re: significance test for non-normal data
> To: Melissa Anne Pfeffer <m.a.pfeffer@geo.uio.no>
>
> The current release of NCL (6.1.2) does not have a function.
>
> In the next release of NCL (6.2.0), the following function will be
> available:
>
> http://www.ncl.ucar.edu/Document/Functions/Built-in/kolsm2_n.shtml
>
> Would this satisfy u needs?
>
>
>
> On 9/17/13 6:29 AM, Melissa Anne Pfeffer wrote:
> Hello,
>
> Does anybody have any ncl code to test for significance for data that
is
> not normally distributed? I can't use ttest because of the
non-normality of
> my data. I want to show if results from sensitivity simulations are
> significantly different from results from the reference simulation. I
think
> what I need is a permutation test, and I appreciate any example code
people
> may have.
>
> Thanks,
> Melissa

Daran Rife
Global Head of Mesoscale Modeling
DNV GL - Energy

9665 Chesapeake Drive, Suite 435, San Diego, CA 92123, USA
Tel: +1 303 271 1450 / Fax: +1 303 271 1450 / Mobile: +1 303 968 6821
www.gl-garradhassan.com

This email is issued on behalf of Garrad Hassan America, Inc. (trading
as GL Garrad Hassan) and is subject to the GL Garrad Hassan email
disclaimer (http://www.gl-garradhassan.com/emaildisclaimer). Garrad
Hassan America, Inc. is registered in America No. 94-3402236 Registered
Office: 45 Main Street, Suite 302, Peterborough, NH 03458, USA

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Sep 17 13:30:48 2013