Calculate the beta coefficients for linear regression

lm_beta_coefficients.R

#generate random numbers
set.seed(1234)
x1 <- rnorm(n = 1000, mean = 3, sd = 3)
x2 <- rnorm(n = 1000, mean = 0, sd = 1)
noise <- runif(n = 1000, min = 0, max = 1)

# Set formula
y <- 5 * x1 - 3 * x2 + noise

# Run regression
md <- lm(y ~ x1 + x2)

summary(md)

# Calculate the beta coefficients 
# Extract the Xs coefficients
b <- summary(md)$coef[-1, 1]
# Calculate the SD for the Xs and Y
sx <- sapply(md$model[-1], sd)
sy <- sd(md$model[1][,1])

# Calculate the beta coefficients
beta <- b * sx/sy
beta

# Alternatively use the lm.beta function from the QuantPsyc package
library(QuantPsyc)
lm.beta(md)