confidence limits for method of moments estimators of variance components
function for getting confidence intervals on variance components estimated by the method of moments
vci(confl,c1,ms1,nu1,c2,ms2,nu2)
confl |
input- confidence level |
c1 |
input - linear combination coefficient of ms1 in the estimated variance component |
ms1 |
input - Anova mean square 1 |
nu1 |
input - Anova degrees of freedom for mean square 1 |
c2 |
input - linear combination coefficient of ms2 in the estimated variance component |
ms2 |
input - Anova mean square 2 |
nu2 |
input - Anova degrees of freedom for mean square 2 |
returned delta, Lower and Upper limits
John Lawson
vci(.90,.05,.014852,2,.05,.026885,18)
## The function is currently defined as
vci<-function(confl,c1,ms1,nu1,c2,ms2,nu2){
delta<-c1*ms1-c2*ms2
alpha<-1-confl
Falpha1<-qf(confl,nu1,10000000)
Falpha12<-qf(confl,nu1,nu2)
Fconf2<-qf(alpha,nu2,10000000)
Fconf12<-qf(alpha,nu1,nu2)
Falpha2<-qf(confl,nu2,10000000)
Fconf1<-qf(alpha,nu1,10000000)
Fconf12<-qf(alpha,nu1,nu2)
G1<-1-(1/Falpha1)
H2<-(1/Fconf2)-1
G12<-((Falpha12-1)**2-G1**2*Falpha12**2-H2**2)/Falpha12
VL<-G1**2*c1**2*ms1**2+H2**2*c2**2*ms2**2+G12*c1*c2*ms1*ms2
H1<-(1/Fconf1)-1
G2<-1-(1/Falpha2)
H12<-((1-Fconf12)**2-H1**2*Fconf12**2-G2**2)/Fconf12
VU<-H1**2*c1**2*ms1**2+G2**2*c2**2*ms2**2
L<-delta-sqrt(VL)
U<-delta+sqrt(VU)
cat("delta=",delta," Lower Limit=",L," Upper Limit=",U,"\n")
}Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.