Computes standardized ("beta") regression coefficients for a fitted lm
or mlm object, i.e. the coefficients that would result from fitting the
same model with all numeric variables rescaled to mean 0, SD 1.
For an "mlm" object, standardization follows stdmodel()'s convention:
the response(s) and numeric predictors are standardized; factor
predictors are left raw.
Usage
stdcoef(object, ...)
# S3 method for class 'lm'
stdcoef(object, ...)
# S3 method for class 'mlm'
stdcoef(object, ...)Value
For an "lm" object, a named numeric vector of standardized
coefficients (intercept excluded). For an "mlm" object, a numeric
matrix of standardized coefficients (parameters x responses,
intercept row excluded).
See also
stdmodel(), coefplot.mlm(). For standard errors, test statistics
and p-values on the standardized scale, run lmtest::coeftest() (and, for
a tidy data frame, broom::tidy.coeftest()) on stdmodel()'s result.
Other multivariate linear models:
coefplot(),
glance.mlm(),
stdmodel()
Examples
data(Prestige, package = "carData")
prestige.mod <- lm(prestige ~ income + education, data = Prestige)
stdcoef(prestige.mod)
#> income education
#> 0.3359243 0.6561537
rohwer.mod <- lm(cbind(SAT, PPVT, Raven) ~ SES + n + s + ns + na + ss, data = Rohwer)
stdcoef(rohwer.mod)
#> SAT PPVT Raven
#> SES1 0.147027414 0.5067869165 0.255713484
#> n 0.202149151 0.0005263032 0.018048991
#> s 0.003769722 -0.0924254802 0.256124149
#> ns -0.481088194 -0.0983270477 0.197100216
#> na 0.438041696 0.4835442475 -0.019473052
#> ss 0.187478718 0.1734997047 -0.008684568