confidence intervals lab updates

[rpruim]

mosaic::sample() handles a multi-column data frame just fine. No need to select just one variable.

[rpruim]

Do we want to show cnorm()?

cnorm(.95)
##          lower    upper
## [1,] -1.959964 1.959964

[rpruim]

If we use cnorm(), then creating an interval looks like this:

zstar <- cnorm(.95)
se <- sd(~ area, data = sample60) / sqrt(60)
sample_mean + zstar * se
##         lower    upper
## [1,] 1249.123 1482.044

[rpruim]

Not sure I'm a fan of this style:

with(samp_ci, plot_ci(lower, upper, m = mean(~ area, data = ames)))

Possible solutions:

• leave as is
• write a different wrapper that avoids using with and starts from the raw data
• use native ggformula. Except for calculating coverage, it is dead simple. But not sure I like the part about computing which intervals cover.
• use CIsim()
• other ideas?

# This can also be done in native ggformula -- just need to do some work to 
# to calculate which ones cover.
pop.mean <- mean(~ area, data = ames)
samp_ci <- samp_ci %>%
  mutate(covers = pop.mean >= lower & pop.mean <= upper)
gf_pointrange(mean + lower + upper ~ .index, data = samp_ci, color = ~covers)