tidy.mfx: Tidy a(n) mfx object

Description

Tidy summarizes information about the components of a model. A model component might be a single term in a regression, a single hypothesis, a cluster, or a class. Exactly what tidy considers to be a model component varies across models but is usually self-evident. If a model has several distinct types of components, you will need to specify which components to return.

The particular functions below provide generic tidy methods for objects returned by the mfx package, preserving the calculated marginal effects instead of the naive model coefficients. The returned tidy tibble will also include an additional "atmean" column indicating how the marginal effects were originally calculated (see Details below).

Usage

# S3 method for mfx
tidy(x, conf.int = FALSE, conf.level = 0.95, ...)
# S3 method for logitmfx
tidy(x, conf.int = FALSE, conf.level = 0.95, ...)
# S3 method for negbinmfx
tidy(x, conf.int = FALSE, conf.level = 0.95, ...)
# S3 method for poissonmfx
tidy(x, conf.int = FALSE, conf.level = 0.95, ...)
# S3 method for probitmfx
tidy(x, conf.int = FALSE, conf.level = 0.95, ...)

Value

A tibble::tibble() with columns:

conf.high: Upper bound on the confidence interval for the estimate.
conf.low: Lower bound on the confidence interval for the estimate.
estimate: The estimated value of the regression term.
p.value: The two-sided p-value associated with the observed statistic.
statistic: The value of a T-statistic to use in a hypothesis that the regression term is non-zero.
std.error: The standard error of the regression term.
term: The name of the regression term.
atmean: TRUE if the marginal effects were originally calculated as the partial effects for the average observation. If FALSE, then these were instead calculated as average partial effects.

Arguments

x

A logitmfx, negbinmfx, poissonmfx, or probitmfx object. (Note that betamfx objects receive their own set of tidiers.)

conf.int

Logical indicating whether or not to include a confidence interval in the tidied output. Defaults to FALSE.

conf.level

The confidence level to use for the confidence interval if conf.int = TRUE. Must be strictly greater than 0 and less than 1. Defaults to 0.95, which corresponds to a 95 percent confidence interval.

...

Additional arguments. Not used. Needed to match generic signature only. Cautionary note: Misspelled arguments will be absorbed in ..., where they will be ignored. If the misspelled argument has a default value, the default value will be used. For example, if you pass conf.lvel = 0.9, all computation will proceed using conf.level = 0.95. Two exceptions here are:

tidy() methods will warn when supplied an exponentiate argument if it will be ignored.
augment() methods will warn when supplied a newdata argument if it will be ignored.

Details

The mfx package provides methods for calculating marginal effects for various generalized linear models (GLMs). Unlike standard linear models, estimated model coefficients in a GLM cannot be directly interpreted as marginal effects (i.e., the change in the response variable predicted after a one unit change in one of the regressors). This is because the estimated coefficients are multiplicative, dependent on both the link function that was used for the estimation and any other variables that were included in the model. When calculating marginal effects, users must typically choose whether they want to use i) the average observation in the data, or ii) the average of the sample marginal effects. See vignette("mfxarticle") from the mfx package for more details.

Examples

Run this code

if (FALSE) { # rlang::is_installed("mfx")

# load libraries for models and data
library(mfx)

# get the marginal effects from a logit regression
mod_logmfx <- logitmfx(am ~ cyl + hp + wt, atmean = TRUE, data = mtcars)

tidy(mod_logmfx, conf.int = TRUE)

# compare with the naive model coefficients of the same logit call
tidy(
  glm(am ~ cyl + hp + wt, family = binomial, data = mtcars), conf.int = TRUE
)

augment(mod_logmfx)
glance(mod_logmfx)

# another example, this time using probit regression
mod_probmfx <- probitmfx(am ~ cyl + hp + wt, atmean = TRUE, data = mtcars)

tidy(mod_probmfx, conf.int = TRUE)
augment(mod_probmfx)
glance(mod_probmfx)
}