Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.


Virgil
Posts:
8,833
Registered:
1/6/11


Re: Squares intead Module in standard deviation
Posted:
Oct 12, 2013 3:15 AM


In article <X9dnYiwhco8c8XPnZ2dnUVZ_oWdnZ2d@giganews.com>, fom <fomJUNK@nyms.net> wrote:
> On 10/11/2013 11:56 PM, Christiano wrote: > > Why in the formula of standard deviation uses squares instead of simply > > putting differences within module? > > > > Example: > > standard deviation² = Sum[(yym)²]/(n1) > > > > Is not it better this? > > standard deviation = Sum[yym]/(n1)
This one, or rather Sum[yym]/n, is more commonly called the mean absolute deviation rather than standard deviation.
http://pic.dhe.ibm.com/infocenter/spssas/v1r0m0/index.jsp?topic=%2Fcom.ib m.spss.analyticcatalyst.help%2Fanalytic_catalyst%2Fmeanmedianabsolutedevi ation.html > > > > Your answer is probably here: > > http://en.wikipedia.org/wiki/Standard_deviation#Relationship_between_standard_ > deviation_and_mean > > The standard deviation satisfies a minimization problem with > respect to the mean. > > As a general criterion for dispersions to points of central > tendency, there is a small table here: > > http://en.wikipedia.org/wiki/Central_tendency#Solutions_to_variational_problem > s > > Click on the other dispersions to find if any of the others > relates to your "better" choice. 



