sim_ame()
is a wrapper for sim_apply()
that computes average marginal effects, the average effect of changing a single variable from one value to another (i.e., from one category to another for categorical variables or a tiny change for continuous variables).
Usage
sim_ame(
sim,
var,
subset = NULL,
contrast = NULL,
outcome = NULL,
type = NULL,
eps = 1e-05,
verbose = TRUE,
cl = NULL
)
# S3 method for clarify_ame
print(x, digits = NULL, max.ests = 6, ...)
Arguments
- sim
a
clarify_sim
object; the output of a call tosim()
ormisim()
.- var
either the name of a variable for which marginal effects are to be computed or a named list of length one containing the values the variable should take. If a list is supplied or the named variables is categorical (factor, character, or having two values), categorical calculations will be triggered. Otherwise, continuous calculations will be triggered. See Details.
- subset
optional; a vector used to subset the data used to compute the marginal effects. This will be evaluated within the original dataset used to fit the model using
subset()
, so nonstandard evaluation is allowed.- contrast
a string containing the name of a contrast between the average marginal means when the variable named in
var
is categorical and takes on two values. Allowed options include"diff"
for the difference in means (also"rd"
),"rr"
for the risk ratio (also"irr"
),"log(rr):
for the log risk ratio (also"log(irr)"
),"or"
for the odds ratio,"log(or)"
for the log odds ratio, and"nnt"
for the number needed to treat. These options are not case sensitive, but the parentheses must be included if present.- outcome
a string containing the name of the outcome or outcome level for multivariate (multiple outcomes) or multi-category outcomes. Ignored for univariate (single outcome) and binary outcomes.
- type
a string containing the type of predicted values (e.g., the link or the response). Passed to
marginaleffects::get_predict()
and eventually topredict()
in most cases. The default and allowable option depend on the type of model supplied, but almost always corresponds to the response scale (e.g., predicted probabilities for binomial models).- eps
when the variable named in
var
is continuous, the value by which to change the variable values to approximate the derivative. See Details.- verbose
logical
; whether to display a text progress bar indicating progress and estimated time remaining for the procedure. Default isTRUE
.- cl
a cluster object created by
parallel::makeCluster()
, or an integer to indicate the number of child-processes (integer values are ignored on Windows) for parallel evaluations. Seepbapply::pblapply()
for details. IfNULL
, no parallelization will take place.- x
a
clarify_ame
object.- digits
the minimum number of significant digits to be used; passed to
print.data.frame()
.- max.ests
the maximum number of estimates to display.
- ...
optional arguments passed to
FUN
.
Value
A clarify_ame
object, which inherits from clarify_est
and is similar to
the output of sim_apply()
, with the additional attribute "var"
containing
the variable named in var
. The average marginal means will be named
E[Y({v})]
, where {v}
is replaced with the values the focal variable
(var
) takes on. The average marginal effect for a continuous var
will be
named dY/d({x})
where {x}
is replaced with var
.
Details
sim_ame()
operates differently depending on whether continuous or categorical calculations are triggered. To trigger categorical calculations, var
should be a string naming a factor, character, or binary variable or a named list with specific values given (e.g., var = list(x1 = c(1, 2 ,3))
). Otherwise, continuous calculations are triggered.
Categorical calculations involve computing average marginal means at each level of var
. The average marginal mean is the average predicted outcome value after setting all units' value of var
to one level. (This quantity has several names, including the average potential outcome, average adjusted prediction, and standardized mean). When var
only takes on two levels (or it is supplied as a list and only two values are specified), a contrast between the average marginal means can be computed by supplying an argument to contrast
. Contrasts can be manually computed using transform()
afterward as well.
Continuous calculations involve computing the average of marginal effects of var
across units. A marginal effect is the instantaneous rate of change corresponding to changing a unit's observed value of var
by a tiny amount and considering to what degree the predicted outcome changes. The ratio of the change in the predicted outcome to the change in the value of var
is the marginal effect; these are averaged across the sample to arrive at an average marginal effect. The "tiny amount" used is eps
times the standard deviation of the focal variable.
If unit-level weights are included in the model fit (and discoverable using insight::get_weights()
), all means will be computed as weighted means.
Effect measures
The effect measures specified in contrast
are defined below. Typically only "diff"
is appropriate for continuous outcomes and "diff"
or "irr"
are appropriate for count outcomes; the rest are appropriate for binary outcomes. For a focal variable with two levels, 0
and 1
, and an outcome Y
, the average marginal means will be denoted in the below formulas as E[Y(0)]
and E[Y(1)]
, respectively.
contrast | Formula |
"diff" | E[Y(1)] - E[Y(0)] |
"rr" | E[Y(1)] / E[Y(0)] |
"or" | O[Y(1)] / O[Y(0)] , where O[Y(.)] = E[Y(.)] / (1 - E[Y(.)]) |
"nnt" | 1 / (E[Y(1)] - E[Y(0)]) |
The log(.)
versions are defined by taking the log()
(natural log) of the corresponding effect measure.
See also
sim_apply()
, which provides a general interface to computing any
quantities for simulation-based inference; plot.clarify_est()
for plotting the
output of a call to sim_ame()
; summary.clarify_est()
for computing
p-values and confidence intervals for the estimated quantities.
marginaleffects::marginaleffects()
, marginaleffects::comparisons()
, and margins::margins()
for delta method-based implementations of computing average marginal effects.
Examples
data("lalonde", package = "MatchIt")
# Fit the model
fit <- glm(I(re78 > 0) ~ treat + age + race + re74,
data = lalonde, family = binomial)
# Simulate coefficients
set.seed(123)
s <- sim(fit, n = 100)
# Average marginal effect of `age`
est <- sim_ame(s, var = "age", verbose = FALSE)
summary(est)
#> Estimate 2.5 % 97.5 %
#> dY/d(age) -0.00678 -0.00981 -0.00403
# Contrast between average marginal means for `treat`
est <- sim_ame(s, var = "treat", contrast = "rr",
verbose = FALSE)
summary(est)
#> Estimate 2.5 % 97.5 %
#> E[Y(0)] 0.744 0.698 0.784
#> E[Y(1)] 0.809 0.758 0.851
#> RR 1.088 0.987 1.198
# Average marginal means for `race`; need to follow up
# with contrasts for specific levels
est <- sim_ame(s, var = "race", verbose = FALSE)
est <- transform(est,
`RR(h,b)` = `E[Y(hispan)]` / `E[Y(black)]`)
summary(est)
#> Estimate 2.5 % 97.5 %
#> E[Y(black)] 0.704 0.637 0.764
#> E[Y(hispan)] 0.832 0.734 0.899
#> E[Y(white)] 0.800 0.753 0.843
#> RR(h,b) 1.181 1.026 1.360