heywood_case.hlp

来自「是一个经济学管理应用软件 很难找的 但是经济学学生又必须用到」· HLP 代码 · 共 38 行

{smcl}
{* 31mar2005}{...}
{* This hlp file is called by clickable output in -factor-}{...}
{title:Heywood case}

{pstd}
Factor analysis output mentioning Heywood cases indicate that a boundary
solution, called a Heywood solution, was produced.  The approximations used in
computing the chi-squared value and degrees of freedom for the
likelihood-ratio test are mathematically justified on the assumption that an
interior solution to the factor maximum likelihood was found.

{pstd}
Heywood solutions often produce uniquenesses of 0, and in that case, at least
at a formal level, the test cannot be justified.  Nevertheless, we believe
that the reported tests are useful, even in such circumstances, provided that
they are interpreted cautiously.  The maximum likelihood method seems to be
particularly prone to producing Heywood solutions.

{pstd}
A message concerning Heywood cases is also printed when, in principle, there
are enough free parameters to completely fit the correlation matrix, another
sort of boundary solution.  We say "in principle" because the correlation
matrix frequently cannot be fitted perfectly, so you will see a positive
chi-squared with zero degrees of freedom.  This warning note is printed
because the geometric assumptions underlying the likelihood-ratio test are
violated.


{title:Also see}

{psee}
Manual:  {bf:[MV] factor}

{psee}
Online:  {helpb factor}
{p_end}