graph a normal distribution with skew and kurtosis

MarkS

unread,

Dec 17, 2006, 11:54:01 PM12/17/06

to

Hi,
I know the mean, max, min, standard deviation, skew and kurtosis.

I need to plot a graph with this data and I have got as far as using
NORMDIST to create the normal distribution curve. How do I add in the skew
and kurtosis

Thanks

David Biddulph

unread,

Dec 18, 2006, 12:56:47 AM12/18/06

to

Your subject line talks of a normal distribution, in which case presumably
your skew is zero and kurtosis is 3? If not, you presumably don't have a
normal distribution? Similarly if you have a normal distribution your max
and min are at plus and minus infinity?
--
David Biddulph

"MarkS" <Ma...@discussions.microsoft.com> wrote in message
news:95423B3D-DA55-41FE...@microsoft.com...

Mike Middleton

unread,

Dec 18, 2006, 1:12:15 AM12/18/06

to

MarkS -

The normal curve (bell-shaped) is completely described by only two
parameters: the mean and standard deviaton.

For charting instructions, see

http://www.tushar-mehta.com/excel/charts/normal_distribution/index.htm

- Mike
http://www.mikemiddleton.com

"MarkS" <Ma...@discussions.microsoft.com> wrote in message
news:95423B3D-DA55-41FE...@microsoft.com...

MarkS

unread,

Dec 18, 2006, 1:54:04 AM12/18/06

to

This is what I know
Min -5.83% -5.83% -3.88%
Avg 0.10% -0.41% -0.17%
Max 5.84% 2.17% 1.92%
Std Dev 1.86% 1.45% 1.43%
Skew 0.2621 -2.4322 -0.8981
Kurt 3.0580 9.6449 0.8925

MarkS

unread,

Dec 18, 2006, 1:55:01 AM12/18/06

to

this is what I know
Min -5.83% -5.83% -3.88%
Avg 0.10% -0.41% -0.17%
Max 5.84% 2.17% 1.92%
Std Dev 1.86% 1.45% 1.43%
Skew 0.2621 -2.4322 -0.8981
Kurt 3.0580 9.6449 0.8925

Jerry W. Lewis

unread,

Dec 18, 2006, 7:47:01 AM12/18/06

to

"Kurtosis" may mean "excess kurtosis"
http://mathworld.wolfram.com/Kurtosis.html
which is 0 for the normal distribution and is estimated by the Excel KURT()
function.

Jerry

Jerry W. Lewis

unread,

Dec 18, 2006, 7:56:01 AM12/18/06

to

Is this one data set with 18 observations (in which case Min, Max, etc are
not labeling estimates from the distribution), or are these summary
statistics for 3 data sets (what is the sample size?)

IF this is a complete enumeration of one data set, please clarify the
relationship between values that are and are not expressed as %.

Jerry

James Silverton

unread,

Dec 18, 2006, 9:44:07 AM12/18/06

to

Hello, Mike!
You wrote on Sun, 17 Dec 2006 22:12:15 -0800:

MM> The normal curve (bell-shaped) is completely described by
MM> only two parameters: the mean and standard deviaton.

MM> For charting instructions, see

MM>
http://www.tushar-mehta.com/excel/charts/normal_distribution/index.htm

MM> - Mike
MM> http://www.mikemiddleton.com

This is doubtless correct but I suppose you could modify the
values given by a normal curve to show kurtosis and skewness by
using a suitable non-linear scale. Still, I'm not sure how I
would achieve a given value of either.

MM> "MarkS" <Ma...@discussions.microsoft.com> wrote in message
MM> news:95423B3D-DA55-41FE...@microsoft.com...
??>> Hi,
??>> I know the mean, max, min, standard deviation, skew and
??>> kurtosis.
??>>
??>> I need to plot a graph with this data and I have got as
??>> far as using NORMDIST to create the normal distribution
??>> curve. How do I add in the skew and kurtosis
??>>
??>> Thanks

James Silverton
Potomac, Maryland

E-mail, with obvious alterations:
not.jim.silverton.at.comcast.not

Jerry W. Lewis

unread,

Dec 18, 2006, 12:25:01 PM12/18/06

to

A couple of approaches would be an Edgeworth expansion
http://www.math.sfu.ca/~cbm/aands/page_935.htm
or the so called "Skew Normal" distribution
http://tango.stat.unipd.it/SN/

Jerry