MODEL Subcommand (PROXSCAL command)
MODEL
defines
the scaling model for the analysis if more than one source is present. IDENTITY
is the default model. The three
other models are individual differences models.
IDENTITY. Identity model. All sources have the same configuration. This is the default model, and it is not an individual differences model.
WEIGHTED. Weighted Euclidean
model. This model is an individual differences model and
equivalent to the INDSCAL
model
in the ALSCAL procedure. Each source has an individual space, in which
every dimension of the common space is weighted differentially.
GENERALIZED. Generalized
Euclidean model. This model is equivalent to the GEMSCAL
model in the ALSCAL procedure. Each
source has an individual space that is equal to a rotation of the
common space, followed by a differential weighting of the dimensions.
REDUCED. Reduced rank
model. This model is similar to GENERALIZED
, but the rank of the individual space is
equal to n. This number is always smaller than the maximum number
of dimensions and equal to or greater than 1. The default is 2.
- If
IDENTITY
is specified for only one source, this subcommand is silently ignored. - If an individual differences model is specified for
only one source, a warning is issued, and the model is set to
IDENTITY
.
Example
PROXSCAL object01 TO object07
/MODEL=WEIGHTED.
- A weighted Euclidean model is fitted, but only when
the number of cases in the active dataset is a multiple of 7, starting
from 14 (14, 21, 28, and so on). Otherwise, there is only one source,
and the model is set to
IDENTITY
.