konfound: Konfound Analysis for Various Model Types

In jrosen48/konfound: Quantify the Robustness of Causal Inferences

Konfound Analysis for Various Model Types

Description

Performs sensitivity analysis on fitted models including linear models ('lm'), generalized linear models ('glm'), and linear mixed-effects models ('lmerMod'). It calculates the amount of bias required to invalidate or sustain an inference,and the impact of an omitted variable necessary to affect the inference.

Usage

konfound(
  model_object,
  tested_variable,
  alpha = 0.05,
  tails = 2,
  index = "RIR",
  to_return = "print",
  two_by_two = FALSE,
  n_treat = NULL,
  switch_trm = TRUE,
  replace = "control"
)

Arguments

model_object	A model object produced by 'lm', 'glm', or 'lme4::lmer'.
tested_variable	Variable associated with the coefficient to be tested.
alpha	Significance level for hypothesis testing.
tails	Number of tails for the test (1 or 2).
index	Type of sensitivity analysis ('RIR' by default).
to_return	Type of output to return ('print', 'raw_output', 'table').
two_by_two	Boolean; if 'TRUE', uses a 2x2 table approach for 'glm' dichotomous variables.
n_treat	Number of treatment cases (used only if 'two_by_two' is 'TRUE').
switch_trm	Boolean; switch treatment and control in the analysis.
replace	Replacement method for treatment cases ('control' by default).

Value

Depending on 'to_return', prints the result, returns a raw output, or a summary table.

Examples

# using lm() for linear models
m1 <- lm(mpg ~ wt + hp, data = mtcars)
konfound(m1, wt)
konfound(m1, wt, to_return = "table")

# using glm() for non-linear models
if (requireNamespace("forcats")) {
  d <- forcats::gss_cat

  d$married <- ifelse(d$marital == "Married", 1, 0)

  m2 <- glm(married ~ age, data = d, family = binomial(link = "logit"))
  konfound(m2, age)
}

# using lme4 for mixed effects (or multi-level) models
if (requireNamespace("lme4")) {
  library(lme4)
  m3 <- fm1 <- lme4::lmer(Reaction ~ Days + (1 | Subject), sleepstudy)
  konfound(m3, Days)
}

m4 <- glm(outcome ~ condition, data = binary_dummy_data, family = binomial(link = "logit"))
konfound(m4, condition, two_by_two = TRUE, n_treat = 55)

jrosen48/konfound documentation built on Nov. 21, 2024, 4:42 a.m.