The goal of konfound is to carry out sensitivity analysis as described in Frank, Maroulis, Duong, and Kelcey (2013) based on Rubin's (1974) causal model.

Installation

# install.packages("devtools")
devtools::install_github("jrosen48/rsensitivity")

Use of konfound

pkonfound() for published studies

pkonfound(), for published studies, calculates (1) how much bias there must be in an estimate to invalidate/sustain an inference; (2) the impact of an omitted variable necessary to invalidate/sustain an inference for a regression coefficient:

library(konfound)
pkonfound(2, .4, 100, 3, to_return = "df")
#> # A tibble: 1 x 3
#>       inference percent observations
#>           <chr>   <dbl>        <dbl>
#> 1 to_invalidate   60.31           60
pkonfound(2, .4, 100, 3, to_return = "print")
#> To invalidate the inference, 60.31 % of the estimate would have to be due to bias.
#> To invalidate the inference, 60 observations would have to be replaced with cases for which there is no effect.
pkonfound(2, .4, 100, 3, to_return = "plot")

pkonfound(.4, 2, 100, 3, to_return = "print")
#> To sustain the inference,  89.92 % of the estimate would have to be due to bias.
#> To sustain the inference,  90  of the cases with 0 effect would have to be replaced with cases at the threshold of inference.

konfound() for models fit in R

konfound() calculates the same for models fit in R. For example, here is the output from a linear model fit with lm() using the built-in dataset mtcars:

m1 <- lm(mpg ~ wt + hp, data = mtcars)
arm::display(m1)
#> lm(formula = mpg ~ wt + hp, data = mtcars)
#>             coef.est coef.se
#> (Intercept) 37.23     1.60  
#> wt          -3.88     0.63  
#> hp          -0.03     0.01  
#> ---
#> n = 32, k = 3
#> residual sd = 2.59, R-Squared = 0.83

Sensitivity analysis for the effect for wt on mpg can be carried out as follows, specifying the fitted model object:

konfound(m1, wt)
#> To invalidate the inference, 66.72 % of the estimate would have to be due to bias.
#> To invalidate the inference, 21 observations would have to be replaced with cases for which there is no effect.

mkonfound for meta-analyses including sensitivity analysis

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
df <- tribble(
  ~unstd_beta, ~standard_error, ~n_obs, ~n_covariates,
  2,           .3,              70,      3,
  10,          2.9,             405,     4,
  1.7,         1.5,             200,     1,
  -3,          1.3,             125,     2
)

mkonfound(df)
#> # A tibble: 4 x 7
#>   unstd_beta standard_error n_obs n_covariates     inference percent
#>        <dbl>          <dbl> <dbl>        <dbl>         <chr>   <dbl>
#> 1        2.0            0.3    70            3 to_invalidate   70.06
#> 2       10.0            2.9   405            4 to_invalidate   42.99
#> 3        1.7            1.5   200            1    to_sustain   42.53
#> 4       -3.0            1.3   125            2 to_invalidate   14.22
#> # ... with 1 more variables: observations <dbl>

Shiny Version for published studies

A shiny version for sensitivity analysis for published studies is available here