weightSign: Sign ambiguity corrections

In mronkko/matrixpls: Matrix-Based Partial Least Squares Estimation

Sign ambiguity corrections

Description

Sign ambiguity corrections adjust the signs of the weights to satisfy a criterion.

Usage

weightSign.Wold1985(W, S)

weightSign.dominantIndicator(W, S)

Arguments

W	Weight matrix, where the indicators are on colums and composites are on the rows.
S	Covariance matrix of the data.

Details

Instead of fixing a weight to a particular value, composite variables are typically provided a scale by standardization. This leads to sign indeterminacy because standardized weights W and -W both satisfy the scaling constraint. The sing ambiguity corrections add additional constraints that make

The sign indeterminacy corrections should not be confused with sign chance corrections applied to boostrap samples (See signChange).

Value

W after sign correction.

Functions

• weightSign.Wold1985: Adjust the signs of W so that the majority of the indicators are positively correlated with the composite as proposed by Wold (1985).

• weightSign.dominantIndicator: Adjust the signs of W so that the first indicator of each composite has positive weight.

References

Wold, H. (1985). Partial Least Squares. In S. Kotz & N. L. Johnson (Eds.), Encyclopedia of statistical sciences (Vol. 6, pp. 581–591). New York: Wiley.

See Also

matrixpls;

mronkko/matrixpls documentation built on April 19, 2024, 4:23 p.m.