calls <- readxl::read_excel("state-data.xlsx", sheet = "calls")
attach(calls)

## graphical summary
boxplot(duration)

## checking for normality
shapiro.test(duration)

## left-tailed and two-tailed sign test - by hand
calls.stat <- sum(duration > 5)
calls.n <- length(duration)
# left-tailed p-value
1 - pbinom(calls.n - calls.stat - 1, size = calls.n, prob = 1/2)
# two-tailed p-value
2*(1 - pbinom(max(calls.stat, calls.n - calls.stat) - 1, size = calls.n, prob = 1/2))

## direct solution
binom.test(calls.stat, n = calls.n, p = 1/2, alternative = "less")
binom.test(calls.stat, n = calls.n, p = 1/2)