lm.beta: Add Standardized Regression Coefficients to...

View source: R/lm.beta.R

lm.betaR Documentation

Add Standardized Regression Coefficients to Linear-Model-Objects

Description

Adds standardized regression coefficients to objects created by lm.

Usage

lm.beta(object, complete.standardization = FALSE)

Arguments

object

An R object of type lm

complete.standardization

Logical. (See Details.)

Details

Calculates the standardized regression coefficients by common method used for example in SPSS. For translating the formula, functions model.matrix (for the right-hand side) and model.frame (for the left-hand side) are used. Additionally the case weights are regarded. So all options saved in the lm-object are supported.

In the case of models with intercept, the standardization results in the same estimates as lm(..., data = scale(data)).

In the case of models without intercept, there are two different types of standardization available. (1) Complete standardization (complete.standardization = TRUE) results in the same estimates as lm(..., data = scale(data)) and therefore results in the same estimates as the same model with intercept. (2) Incomplete standardization (complete.standardization = FALSE, the standard value) results in the same estimates as lm(..., data = scale(data, center = FALSE)). This estimation is implemented in IBM SPSS Statistics. For a theoretical justification see Eisenhauer 2003.

Please regard:

Package lm.beta standardizes the coefficients after estimating them using the standard deviations or similar measures of the used variables. So there are unstandardized and standardized coefficients available simultaneously.

Standardizing before estimating is not (yet) available in this package, but by using the function scale you can do this by using basic commands. Hereby please regard that the option center influences the way of interpretation of the intercept.

Package lm.beta standardizes all coefficients disregarding the use in interpretation. In this version, all types of scales of the variables (metrical, categorical, ...), all types of contrasts, interaction effects and additional terms on both sides of the formula can be handled if lm can handle them. The sensitive use in interpretation has to be regarded by the user.

Value

A list of class lm.beta like a lm-object extended by

  • standardized.coefficients named vector of the standardized coefficients.

Note

Some S3 methods, where standardized coefficients mind, are extended, the others work unchanged.

Author(s)

Stefan Behrendt, r@behrendt-stefan.de

References

Eisenhauer, J.G. (2003). Regression through the Origin. In Teching Statistics, 25(3).

Urban, D., Mayerl, J., Sackmann, R. (Hrsg.) Regressionsanalyse : Theorie, Technik und Anwendung. VS-Verlag, 4th ed.

Vittinghoff, E. et al (2005) Regression methods in biostatistics: Linear, logistic, survival, and repeated measures models, Springer, p 75

See Also

lm for creating the demanded object and print.lm.beta, summary.lm.beta and coef.lm.beta as well as xtable.lm.beta for extended S3-methods.

Examples

## Taken from lm help
##
## Annette Dobson (1990) "An Introduction to Generalized Linear Models".
## Page 9: Plant Weight Data.
ctl <- c(4.17,5.58,5.18,6.11,4.50,4.61,5.17,4.53,5.33,5.14)
trt <- c(4.81,4.17,4.41,3.59,5.87,3.83,6.03,4.89,4.32,4.69)
group <- gl(2, 10, 20, labels = c("Ctl","Trt"))
weight <- c(ctl, trt)
lm.D9 <- lm(weight ~ group)

# standardize
lm.D9.beta <- lm.beta(lm.D9)
print(lm.D9.beta)
summary(lm.D9.beta)
coef(lm.D9.beta)
xtable::xtable(lm.D9.beta)

lm.beta documentation built on March 31, 2023, 8:23 p.m.