library(tidyverse)
library(broom)
library(rlang)

#' Compare coefficients values across two different models
#'
#' @param model1 A model object (from lm for example) with a tidy method.
#' @param model2 A model object (from lm for example) with a tidy method.
#' @param … Bare (unquoted) coefficient names present in both models.

compare_coefs <- function(model1, model2, ...) {
  compare_single_coef <- function(coef) {
    coef_name <- quo_name(coef)
    t1 <- dplyr::filter(tidy(model1), term == !!coef_name)
    t2 <- dplyr::filter(tidy(model2), term == !!coef_name)

    if (nrow(t1) == 0 || nrow(t2) == 0) {
      stop(paste0("Coefficient `", quo_name(coef),
                  "` not found in one or both models."), call. = FALSE)
    }

    diff <- t1$estimate - t2$estimate
    se <- sqrt(t1$std.error^2 + t2$std.error^2)
    statistic <- diff / se

    tibble(term = coef_name,
           difference = diff,
           conf.low = diff - 1.96 * se,
           conf.high = diff + 1.96 * se,
           statistic = statistic,
           p.value = 2 * pnorm(abs(statistic), lower.tail = FALSE),
           estimate1 = t1$estimate,
           estimate2 = t2$estimate,
           method = "Z-test")
  }
  coefs <- quos(...)
  map_dfr(coefs, compare_single_coef)
}

model <- lm(mpg ~ hp + drat, mtcars)
other <- lm(mpg ~ hp * disp + drat, mtcars)

# check if there is a significant difference in `hp` and `drat` coefs
# between the models
compare_coefs(model, other, hp, drat)

#> # A tibble: 2 x 9
#>   term  difference conf.low conf.high statistic p.value estimate1
#>   <chr>      <dbl>    <dbl>     <dbl>     <dbl>   <dbl>     <dbl>
#> 1 hp        0.0486  -0.0115     0.109      1.58  0.113    -0.0518
#> 2 drat      5.00     0.796      9.21       2.33  0.0197    4.70
#> # ... with 2 more variables: estimate2 <dbl>, method <chr>