Displaying 1 result from an estimated 1 matches for "dfres".
Did you mean:
dfree
2006 Jul 21
2
glm cannot find valid starting values
...s an integer. The model will have the lowest
deviance
#when x = 0.
x <- 0
Mdif <- N.est - M + x
qpglm2 <- glm(S ~ -1 + Mdif, family = quasipoisson(link = identity), sdat)
summary(qpglm2)
#Use analysis of deviance to estimate confidence limits on N.est:
disp <- summary(qpglm)$dispersion
dfres <- df.residual(qpglm)
dev.res <- deviance(qpglm)
#From MASS4, p. 210, assume that changes in deviance scaled by
#dispersion as |x| increases have an F distribution
dev.crit <- dev.res+qf(0.95,1,dfres)*disp
dev.crit
#values of x for which the deviance = dev.crit give approximate 95%
confid...