Sensitivity statistics for regression coefficients — sensitivity

Convenience function that computes the robustness_value, partial_r2 and partial_f2 of the coefficient of interest.

sensitivity_stats(model, ...)

# S3 method for lm
sensitivity_stats(model, treatment, q = 1, alpha = 0.05, reduce = TRUE, ...)

# S3 method for fixest
sensitivity_stats(
  model,
  treatment,
  q = 1,
  alpha = 0.05,
  reduce = TRUE,
  message = T,
  ...
)

# S3 method for numeric
sensitivity_stats(
  estimate,
  se,
  dof,
  treatment = "treatment",
  q = 1,
  alpha = 0.05,
  reduce = TRUE,
  ...
)

Arguments

model: An fixest object with the outcome regression.
...: Arguments passed to other methods. First argument should either be an lm model with the outcome regression or a numeric vector with the coefficient estimate.
treatment: A character vector with the name of the treatment variable of the model.
q: percent change of the effect estimate that would be deemed problematic. Default is 1, which means a reduction of 100% of the current effect estimate (bring estimate to zero). It has to be greater than zero.
alpha: significance level.
reduce: Should the bias adjustment reduce or increase the absolute value of the estimated coefficient? Default is TRUE.
message: should messages be printed? Default = TRUE.
estimate: Coefficient estimate.
se: Standard error of the coefficient estimate.
dof: Residual degrees of freedom of the regression.

Value

A data.frame containing the following quantities:

treatment: a character with the name of the treatment variable
estimate: a numeric vector with the estimated effect of the treatment
se: a numeric vector with the estimated standard error of the treatment effect
t_statistics: a numeric vector with the t-value of the treatment
r2yd.x: a numeric vector with the partial R2 of the treatment and the outcome, see details in partial_r2
rv_q: a numeric vector with the robustness value of the treatment, see details in robustness_value
rv_qa: a numeric vector with the robustness value of the treatment considering statistical significance, see details in robustness_value
f2yd.x: a numeric vector with the partial (Cohen's) f2 of the treatment with the outcome, see details in partial_f2
dof: a numeric vector with the degrees of freedom of the model

References

Cinelli, C. and Hazlett, C. (2020), "Making Sense of Sensitivity: Extending Omitted Variable Bias." Journal of the Royal Statistical Society, Series B (Statistical Methodology).

Examples


## loads data
data("darfur")

## fits model
model <- lm(peacefactor ~ directlyharmed + age + farmer_dar + herder_dar +
             pastvoted + hhsize_darfur + female + village, data = darfur)

## sensitivity stats for directly harmed
sensitivity_stats(model, treatment = "directlyharmed")
#>        treatment   estimate         se t_statistic     r2yd.x      rv_q
#> 1 directlyharmed 0.09731582 0.02325654     4.18445 0.02187309 0.1387764
#>        rv_qa     f2yd.x dof
#> 1 0.07625797 0.02236222 783

## you can  also pass the numeric values directly
sensitivity_stats(estimate = 0.09731582, se = 0.02325654, dof = 783)
#>   treatment   estimate         se t_statistic     r2yd.x      rv_q      rv_qa
#> 1 treatment 0.09731582 0.02325654     4.18445 0.02187309 0.1387763 0.07625795
#>       f2yd.x dof
#> 1 0.02236222 783