Jump to content

Talk:Robust measures of scale

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Shouldn't the biweight midvariance be multiplied by the number of samples? (see e.g. http://www.unt.edu/benchmarks/archives/2002/july02/rss.htm , namely the formula at http://www.unt.edu/benchmarks/archives/2002/july02/robust4.jpg ).

Also, in order to get a linear measure of scale (akin to standard deviation, MAD etc.) one should extract the square root from the result.

-- enzomich@gmail.com —Preceding unsigned comment added by 219.78.46.217 (talk) 13:49, 5 June 2010 (UTC)[reply]