Thirty patients were given an anesthetic agent maintained
at a predetermined level (conc) for 15 minutes before making
an incision. It was then noted whether the patient
moved, i.e. jerked or twisted.
This data frame contains the following columns:
a binary numeric vector coded for
patient movement (0 = no movement, 1 = movement)
logarithm of concentration
the complement of move
The interest is in estimating
how the probability of jerking or twisting varies with
increasing concentration of the anesthetic agent.
print("Logistic Regression - Example 8.1.4")z <- table(anesthetic$nomove, anesthetic$conc)
tot <- apply(z, 2, sum) # totals at each concentration
prop <- z[2, ]/(tot) # proportions at each concentration
oprop <- sum(z[2, ])/sum(tot) # expected proportion moving if concentration had no effect
conc <- as.numeric(dimnames(z)[])
plot(conc, prop, xlab = "Concentration", ylab = "Proportion", xlim = c(.5,2.5),
ylim = c(0, 1), pch = 16)
chw <- par()$cxy
text(conc - 0.75 * chw, prop, paste(tot), adj = 1)
abline(h = oprop, lty = 2)pause()anes.logit <- glm(nomove ~ conc, family = binomial(link = logit),
data = anesthetic)
Dataset imported from https://www.r-project.org.