class Statsample::Test::BartlettSphericity

Bartlett's test of Sphericity.

Test the hyphotesis that the sample correlation matrix comes from a multivariate normal population where variables are independent. In other words, the population correlation matrix is the identity matrix.

Reference

Bartlett's test of Sphericity.

Test the hyphotesis that the sample correlation matrix comes from a multivariate normal population where variables are independent. In other words, the population correlation matrix is the identity matrix.

Reference

Attributes

df[R]
name[RW]
ncases[R]
nvars[R]
value[R]

Public Class Methods

new(matrix,ncases) click to toggle source

Args

  • matrix: correlation matrix

  • ncases: number of cases

# File lib/statsample/test/bartlettsphericity.rb, line 21
def initialize(matrix,ncases)
  @matrix=matrix
  @ncases=ncases
  @nvars=@matrix.row_size
  @name=_("Bartlett's test of sphericity")
  compute
end

Public Instance Methods

compute() click to toggle source

Uses SPSS formula. On Dziuban & Shirkey, the minus between the first and second statement is a *!!!

# File lib/statsample/test/bartlettsphericity.rb, line 32
def compute
  @value=-((@ncases-1)-(2*@nvars+5).quo(6))*Math::log(@matrix.determinant)
  @df=(@nvars*(@nvars-1)).quo(2)
end
probability() click to toggle source
# File lib/statsample/test/bartlettsphericity.rb, line 36
def probability
  1-Distribution::ChiSquare.cdf(@value,@df)
end