Displaying 1 result from an estimated 1 matches for "rohwer1".
Did you mean:
rohwer
2010 Sep 14
0
influence measures for multivariate linear models
...for multivariate response
models fit with lm().
By bugs, I mean that results returned are silently wrong, or at best,
misleading.
My test case: a subset of the Rohwer data anayzed by Barrett & Ling (&
others)
> library(heplots)
> data(Rohwer, package="heplots")
> Rohwer1 <- Rohwer[Rohwer$SES=='Hi',]
> rohwer.mod <- lm(cbind(SAT, PPVT, Raven) ~ n + s + ns + na + ss,
data=Rohwer1)
> (B <- coef(rohwer.mod))
SAT PPVT Raven
(Intercept) -28.467471 39.6970896 13.2438356
n 3.257132 0.0672825 0.0593474
s...