Displaying 1 result from an estimated 1 matches for "fixedhet".
2010 Mar 22
0
using lmer weights argument to represent heteroskedasticity
...ghts estimated
res[["lme.esthet"]] <- sumLME( lme(fixed = y ~ x, data = d, random =
~1 | groupid, weights = varIdent(form = ~1|verrorf)) )
## lme, heteroskedastic model with true fixed weights
v <- c(v0.5 = 0.5, v1 = 1, v2 = 2, v4 = 4)
sdrats <- sqrt(v / v[1])[-1]
res[["lme.fixedhet"]] <- sumLME( lme(fixed = y ~ x, data = d, random =
~1 | groupid, weights = varIdent(form = ~1|verrorf, fixed = sdrats)) )
detach("package:nlme")
library(lme4)
## lmer, ignoring heteroskedasticity
res[["lmer.nohet"]] <- sumLMER( lmer(y ~ x + (1|groupid), data = d) )...