CCG_TRM2x - Forms the trimmed mean of the given set of ordered data, returning flags showing which values are used

Description:

The routine forms the trimmed mean of the given dataset. The alpha (as a fraction) upper and lower ordered values are removed from consideration. Then the remaining values are added and averaged. The SVAR value is a (given) best estimate of the original un-ordered population variance. The variance of the output value is formed assuming that the original dataset was normal in distribution, and is now fairly represented by the ordered statistics variances-covariances supplied in packed form in COVAR. The elements of the input array which actually contribute to the final value are flagged in the used array.

Invocation:

CALL CCG_TRM2x( ALPHA, EL, ORDDAT, ORDVAR, USED, COVAR, TMEAN, TVAR, STATUS )

Arguments:

ALPHA = REAL (Given)

The fraction of data to trim from upper and lower orders. (MUST BE GREATER THAN 0.0 AND LESS 0.5)

EL = INTEGER (Given)

The number of entries in ORDDAT.

ORDDAT( EL ) = ? (Given)

The set of ordered data for which a trimmed mean is required.

SVAR = DOUBLE PRECISION (Given)

The variance of the unordered sample now ordered in ORDDAT.

USED( EL ) = LOGICAL (Returned)

If the corresponding value in ORDDAT contributes to the final value then this is set true.

COVAR( $*$ ) = DOUBLE PRECISION (Given)

The packed variance-covariance matrix of the order statistics from a normal distribution of size EL.

TMEAN = DOUBLE PRECISION (Returned)

The trimmed mean.

TVAR = DOUBLE PRECISION (Returned)

A variance estimate of the trimmed mean.

STATUS = INTEGER (Given and Returned)

The global status.

Prior Requirements

• The variance-covariance array must have been generated in a fashion similar to that of ORDVAR.