Help System (web edition)

Some short examples of usage follow. All matrices presented here are produced
by Survo operations like CORR, FACTA, ROTATE, /FCOEFF, LINREG etc., but
the weights may also be chosen freely, e.g. based on the theory of the
application, earlier experience etc.
In these examples the factors are assumed uncorrelated (orthogonal),
but they could be oblique as well.

Using the sucro /RELIAB:

The reliabilities of the factor images, the factor scores and the
unweighted sum, including Cronbach's alpha for the sum:
/RELIAB
The above command corresponds to the following with the default matrices:
/RELIAB CORR.M,AFACT.M,MSN.M


Using the RELIAB module:

The reliabilities of the factor images and the unweighted sum:
RELIAB CORR.M,AFACT.M,END+2

The reliabilities of the factor scores:
RELIAB CORR.M,AFACT.M,END+2 / WEIGHT=FCOEFF.M

The reliability of a weighted sum of the observed variables:
RELIAB CORR.M,AFACT.M,END+2 / WEIGHT=REG.M

The reliability of a second order scale:
RELIAB CORR.M,AFACT.M,END+2 / WEIGHT=FCOEFF.M WEIGHT2=REG.M
(Here, the factor scores are used as regressors in regression analysis,
and the weights are the regression coefficients.)

The reliabilities of second order scales:
RELIAB CORR.M,AFACT.M,END+2 / WEIGHT=FCOEFF.M WEIGHT2=DISCRL.M
(Here, the factor scores are used in discriminant analysis, and the weights
are the discriminant coefficients.)

In the last two examples, the resulting second order scale coefficients
are saved in the matrix file WEIGHT2.M .

  M = More information on reliability of measurement scales