inst/tinytest/test-pkg-estimatr.R

In marginaleffects: Predictions, Comparisons, Slopes, Marginal Means, and Hypothesis Tests

source("helpers.R")
using("marginaleffects")
requiet("estimatr")
requiet("emmeans")
requiet("margins")
requiet("broom")

Km <<- read.csv("https://vincentarelbundock.github.io/Rdatasets/csv/sem/Kmenta.csv")
dat <- mtcars
dat$cyl <- factor(dat$cyl)
dat <<- dat

# lm_lin: no validity
mod <- lm_lin(mpg ~ am, ~ hp + cyl, data = dat)
expect_slopes(mod)
expect_slopes(mod, n_unique = 9)

# iv_robust vs. stata
stata <- readRDS(testing_path("stata/stata.rds"))$estimatr_iv_robust
model <- iv_robust(
    Q ~ P + D | D + F + A,
    se_type = "stata",
    data = Km)
mfx <- slopes(model)
tid <- tidy(mfx)
expect_slopes(model)
mfx <- merge(tid, stata)
expect_equivalent(mfx$estimate, mfx$dydxstata)
expect_equivalent(mfx$std.error, mfx$std.errorstata, tolerance = .1)



# lm_robust vs. stata vs. emtrends
model <- lm_robust(carb ~ wt + factor(cyl),
    se_type = "HC2",
    data = dat)
stata <- readRDS(testing_path("stata/stata.rds"))$estimatr_lm_robust
mfx <- tidy(slopes(model))
mfx$term <- ifelse(mfx$contrast == "6 - 4", "6.cyl", mfx$term)
mfx$term <- ifelse(mfx$contrast == "8 - 4", "8.cyl", mfx$term)
mfx <- merge(mfx, stata)
expect_equivalent(mfx$estimate, mfx$dydxstata)
expect_equivalent(mfx$std.error, mfx$std.errorstata, tolerance = .1)
# emtrends
mfx <- slopes(model, newdata = datagrid(cyl = 4, wt = 2, newdata = dat), variables = "wt")
em <- emtrends(model, ~wt, "wt", at = list(cyl = 4, wt = 2))
em <- tidy(em)
expect_equivalent(mfx$estimate, em$wt.trend, tolerance = .001)
expect_equivalent(mfx$std.error, em$std.error, tolerance = .001)
# margins does not support standard errors
tmp <- mtcars
tmp$cyl <- factor(tmp$cyl)
mod <- lm_robust(carb ~ wt + cyl, data = tmp, se_type = "stata")
mar <- margins(mod, data = head(tmp))
mfx <- slopes(mod, newdata = head(tmp))
expect_true(expect_margins(mfx, mar, se = FALSE))


# iv_robust: predictions: no validity
# skip_if_not_installed("insight", minimum_version = "0.17.1")
model <- iv_robust(Q ~ P + D | D + F + A,
    se_type = "stata",
    data = Km)
expect_predictions(predictions(model), n_row = nrow(Km))
expect_predictions(predictions(model, newdata = head(Km)), n_row = 6)


# lm_robust: marginalmeans predictions: no validity
# skip_if_not_installed("insight", minimum_version = "0.17.1")
tmp <- mtcars
tmp$cyl <- as.factor(tmp$cyl)
tmp$am <- as.logical(tmp$am)
model <- lm_robust(carb ~ wt + am + cyl,
    se_type = "stata",
    data = tmp)
expect_predictions(predictions(model), n_row = nrow(tmp))
expect_predictions(predictions(model, newdata = head(tmp)), n_row = 6)
expect_marginal_means(marginal_means(model))




source("helpers.R")
rm(list = ls())