HonestDiD (version 0.2.6)

createSensitivityPlot: Constructs sensitivity plot for \(\Delta = \Delta^{SD}(M)\), \(\Delta^{SDB}(M)\) and \(\Delta^{SDM}(M)\)

Description

This function constructs sensitivity plots that examine how the robust confidence sets change as the parameter M varies for \(\Delta = \Delta^{SD}(M)\), \(\Delta^{SDB}(M)\) and \(\Delta^{SDM}(M)\). Similar plots are constructed in Section 6 of Rambachan & Roth (2021).

Usage

createSensitivityPlot(robustResults, originalResults,
                      rescaleFactor = 1, maxM = Inf, add_xAxis = TRUE)

Value

Returns ggplot object of the sensitivity plot.

Arguments

robustResults

Dataframe that contains the upper/lower bounds of robust confidence sets for each choice of M. Contains columns: method -- Method of constructing robust confidence set (e.g., "FLCI"), lb -- Lower bound of robust confidence set, ub -- Upper bound of robust confidence set, M -- M values associated with each robust confidence set.

originalResults

Dataframe that contains the original confidence set for the parameter of interest. Contains columns: method -- Method of constructing confidence set (e.g., "Original"), lb -- Lower bound of confidence set, ub -- Upper bound of confidence set.

rescaleFactor

Scalar that is used to rescale the user specified choices of M and the upper/lower bounds of the confidence sets. Default equals one.

maxM

Scalar that specifies the maximum M value to plot in the sensitivity plot. Default equals infinity (no truncation).

add_xAxis

Logical specifying whether to plot the x-axis in the sensitivity plot. Default equals TRUE.

Author

Ashesh Rambachan

References

Rambachan, Ashesh and Jonathan Roth. "An Honest Approach to Parallel Trends." 2021.

Examples

# \donttest{
  # Simple use case; for more detailed examples,
  # see 
  robustResults <-
    createSensitivityResults(betahat        = BCdata_EventStudy$betahat,
                             sigma          = BCdata_EventStudy$sigma,
                             numPrePeriods  = length(BCdata_EventStudy$prePeriodIndices),
                             numPostPeriods = length(BCdata_EventStudy$postPeriodIndices),
                             alpha          = 0.05)
  originalResults <-
    constructOriginalCS(betahat        = BCdata_EventStudy$betahat,
                        sigma          = BCdata_EventStudy$sigma,
                        numPrePeriods  = length(BCdata_EventStudy$prePeriodIndices),
                        numPostPeriods = length(BCdata_EventStudy$postPeriodIndices),
                        alpha          = 0.05)
  createSensitivityPlot(robustResults, originalResults)
# }