pkonfound: Perform sensitivity analysis for published studies
In konfound: Quantify the Robustness of Causal Inferences

pkonfound

R Documentation

Perform sensitivity analysis for published studies

Description

For published studies, this command 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.

Usage

pkonfound(
  est_eff,
  std_err,
  n_obs,
  n_covariates = 1,
  alpha = 0.05,
  tails = 2,
  index = "RIR",
  nu = 0,
  n_treat = NULL,
  switch_trm = TRUE,
  model_type = "ols",
  a = NULL,
  b = NULL,
  c = NULL,
  d = NULL,
  two_by_two_table = NULL,
  test = "fisher",
  replace = "control",
  sdx = NA,
  sdy = NA,
  R2 = NA,
  far_bound = 0,
  eff_thr = NA,
  FR2max = 0,
  FR2max_multiplier = 1.3,
  to_return = "print"
)

Arguments

`est_eff`	the estimated effect (such as an unstandardized beta coefficient or a group mean difference)
`std_err`	the standard error of the estimate of the unstandardized regression coefficient
`n_obs`	the number of observations in the sample
`n_covariates`	the number of covariates in the regression model
`alpha`	probability of rejecting the null hypothesis (defaults to 0.05)
`tails`	integer whether hypothesis testing is one-tailed (1) or two-tailed (2; defaults to 2)
`index`	whether output is RIR or IT (impact threshold); defaults to "RIR"
`nu`	what hypothesis to be tested; defaults to testing whether est_eff is significantly different from 0
`n_treat`	the number of cases associated with the treatment condition; applicable only when model_type = "logistic"
`switch_trm`	whether to switch the treatment and control cases; defaults to FALSE; applicable only when model_type = "logistic"
`model_type`	the type of model being estimated; defaults to "ols" for a linear regression model; the other option is "logistic"
`a`	cell is the number of cases in the control group showing unsuccessful results
`b`	cell is the number of cases in the control group showing successful results
`c`	cell is the number of cases in the treatment group showing unsuccessful results
`d`	cell is the number of cases in the treatment group showing successful results
`two_by_two_table`	table that is a matrix or can be coerced to one (data.frame, tibble, tribble) from which the a, b, c, and d arguments can be extracted
`test`	whether using Fisher's Exact Test or A chi-square test; defaults to Fisher's Exact Test
`replace`	whether using entire sample or the control group to calculate the base rate; default is control
`sdx`	the standard deviation of X
`sdy`	the standard deviation of Y
`R2`	the unadjusted, original R2 in the observed function
`far_bound`	whether the estimated effect is moved to the boundary closer (default 0) or further away (1);
`eff_thr`	for RIR: unstandardized coefficient threshold to change an inference; for IT: correlation defining the threshold for inference
`FR2max`	the largest R2, or R2max, in the final model with unobserved confounder
`FR2max_multiplier`	the multiplier of R2 to get R2max, default is set to 1.3
`to_return`	whether to return a data.frame (by specifying this argument to equal "raw_output" for use in other analyses) or a plot ("plot"); default is to print ("print") the output to the console; can specify a vector of output to return

Value

pkonfound prints the bias and the number of cases that would have to be replaced with cases for which there is no effect to nullify the inference. If to_return = "raw_output," a list will be given with the following components:

obs_r: correlation between predictor of interest (X) and outcome (Y) in the sample data.
act_r: correlation between predictor of interest (X) and outcome (Y) from the sample regression based on the t-ratio accounting for non-zero null hypothesis.
critical_r: critical correlation value at which the inference would be nullified (e.g., associated with p=.05).
r_final: final correlation value given CV. Should be equal to critical_r.
rxcv: correlation between predictor of interest (X) and CV necessary to nullify the inference for smallest impact.
rycv: correlation between outcome (Y) and CV necessary to nullify the inference for smallest impact.
rxcvGz: correlation between predictor of interest and CV necessary to nullify the inference for smallest impact conditioning on all observed covariates (given z).
rycvGz: correlation between outcome and CV necessary to nullify the inference for smallest impact conditioning on all observed covariates (given z).
itcvGz: ITCV conditioning on the observed covariates.
itcv: Unconditional ITCV.
r2xz: R2 using all observed covariates to explain the predictor of interest (X).
r2yz: R2 using all observed covariates to explain the outcome (Y).
delta_star: delta calculated using Oster's unrestricted estimator.
delta_star_restricted: delta calculated using Oster's restricted estimator.
delta_exact: correlation-based delta.
delta_pctbias: percent of bias when comparing delta_star with delta_exact.
cor_oster: correlation matrix implied by delta_star.
cor_exact: correlation matrix implied by delta_exact.
beta_threshold: threshold value for estimated effect.
beta_threshold_verify: estimated effect given RIR. Should be equal to beta_threshold.
perc_bias_to_change: percent bias to change the inference.
RIR_primary: Robustness of Inference to Replacement (RIR).
RIR_supplemental: RIR for an extra row or column that is needed to nullify the inference.
RIR_perc: RIR as % of total sample (for linear regression) or as % of data points in the cell where replacement takes place (for logistic and 2 by 2 table).
fragility_primary: Fragility. the number of switches (e.g., treatment success to treatment failure) to nullify the inference.
fragility_supplemental: Fragility for an extra row or column that is needed to nullify the inference.
starting_table: Observed 2 by 2 table before replacement and switching. Implied table for logistic regression.
final_table: The 2 by 2 table after replacement and switching.
user_SE: user entered standard error. Only applicable for logistic regression.
needtworows: whether double row switches are needed.
analysis_SE: the standard error used to generate a plausible 2 by 2 table. Only applicable for logistic regression.
Fig_ITCV: figure for ITCV.
Fig_RIR: figure for RIR.

Examples

# using pkonfound for linear models
pkonfound(2, .4, 100, 3)
pkonfound(-2.2, .65, 200, 3)
pkonfound(.5, 3, 200, 3)
pkonfound(-0.2, 0.103, 20888, 3, n_treat = 17888, model_type = "logistic")

pkonfound(2, .4, 100, 3, to_return = "thresh_plot")
pkonfound(2, .4, 100, 3, to_return = "corr_plot")

# using pkonfound for a 2x2 table
pkonfound(a = 35, b = 17, c = 17, d = 38)
pkonfound(a = 35, b = 17, c = 17, d = 38, alpha = 0.01)
pkonfound(a = 35, b = 17, c = 17, d = 38, alpha = 0.01, switch_trm = FALSE)
pkonfound(a = 35, b = 17, c = 17, d = 38, test = "chisq")

# use pkonfound to calculate delta* and delta_exact 
pkonfound(est_eff = .4, std_err = .1, n_obs = 290, sdx = 2, sdy = 6, R2 = .7,
 eff_thr = 0, FR2max = .8, index = "COP", to_return = "raw_output")
# use pkonfound to calculate rxcv and rycv when preserving standard error
pkonfound(est_eff = .5, std_err = .056, n_obs = 6174, eff_thr = .1,
sdx = 0.22, sdy = 1, R2 = .3, index = "PSE", to_return = "raw_output")

konfound documentation built on Oct. 17, 2024, 5:08 p.m.