[R] Comparing two regression slopes

Martin Biuw emb7 at st-andrews.ac.uk
Wed Jul 30 16:23:46 CEST 2003

I've written a simple (although probably overly roundabout) function to 
test whether two regression slope coefficients from two linear models on 
independent data sets are significantly different. I'm a bit concerned, 
because when I test it on simulated data with different sample sizes and 
variances, the function seems to be extremely sensitive both of these. I am 
wondering if I've missed something in my function? I'd be very grateful for 
any tips.



TwoSlope <-function(lm1, lm2) {

## lm1 and lm2 are two linear models on independent data sets

coef1 <-summary(lm1)$coef
coef2 <-summary(lm2)$coef

sigma <-(sum(lm1$residuals^2)+sum(lm2$residuals^2))/(lm1$df.residual + 
SSall <-sum(lm1$model[,2]^2) + sum(lm2$model[,2]^2)
SSprod <-sum(lm1$model[,2]^2) * sum(lm2$model[,2]^2)

F.val <-(as.numeric(coefficients(lm1)[2]) - as.numeric(coefficients(lm2) 

p.val <-1-pf(F.val, 1, (lm1$df.residual + lm2$df.residual-4))

cat("\n\nTest for equality between two regression slopes\n\n")
cat("\nCoefficients model 1:\n\n")

cat("\nCoefficients model 2:\n\n")

cat("\nF-value on 1 and", lm1$df.residual + lm2$df.residual-4, "degrees of 
freedom:" ,F.val, "\n")
cat("p =", ifelse(p.val>=0.0001, p.val, "< 0.0001"), "\n")


More information about the R-help mailing list