format_ci: Format Confidence Intervals

In svenhalvorson/SvenR: Tools for exploration

Format Confidence Intervals

Description

Round and format confidence intervals for presentation.

Usage

format_ci(
  point,
  lower,
  upper,
  null_value = NA_real_,
  digits = NA_real_,
  max_its = 4,
  unacceptables = NA_real_
)

Arguments

point, lower, upper	Point estimates, lower bounds, and upper bounds.
null_value	A vector of null values to be compared to.
digits	A vector of digits to begin rounding at.
max_its	The maximum number of additiona digits to give

Details

The goal of this function is to round confidence intervals for presentation. You can supply digits to round to or let the function guess. It will try to improve the CI until it hits the max_its. The function attempts to have the point estimate and each end of the CI as different numbers. If you supply a null_value it will try not to use that as either endpoint and keep the null value on the same side of the CI as the unformatted value is. If unacceptables is supplied, these values will not be allowed as endpoints for the CI.

Author(s)

Sven Halvorson

Examples

# Set up a dataset for a logistic regression model:
df = nycflights13::flights %>%
  mutate(on_time = dep_delay <= 0)

analysis_table = glm(
  formula = on_time ~ month + sched_dep_time + origin + air_time,
  data = df,
  family = binomial
) %>%
  broom::tidy(
    conf.int = TRUE,
    exponentiate = TRUE
  )

cis = format_ci(
  point = analysis_table['estimate'],
  lower = analysis_table['conf.low'],
  upper = analysis_table['conf.high'],
  null_value = 1,
  unacceptables = 0
)

analysis_table %>%
  select(
    term,
    estimate,
    conf.low,
    conf.high
  ) %>%
  bind_cols(cis)

svenhalvorson/SvenR documentation built on Aug. 25, 2023, 1:31 p.m.