Tidy an eventStudy object

Returns a broom-style tidy data frame for the output of eventStudy(). Renames existing columns to broom conventions (se \(\to\) std.error, p_value \(\to\) p.value) and adds a term column ("e<event_time>") plus a statistic column (estimate / std.error) so the schema matches tidy.<estimator>() for downstream bind_rows() consumers.

Usage

# S3 method for class 'eventStudy'
tidy(x, conf.int = TRUE, conf.level = 0.95, ...)

Arguments

x

An object of class "eventStudy" returned by eventStudy().

conf.int

Logical; include conf.low / conf.high columns.

conf.level

Numeric in (0, 1). Retained for broom-convention parity (default 0.95) but no longer recomputes the event-study CIs: as of version 1.16.0 (#197) the conf.low / conf.high columns pass through the eventStudy object's stored ci_low / ci_high (reflecting the fit's ci_type). To change the confidence level, recompute eventStudy(fit, alpha = ...) at the desired alpha first.

...

Unused.

Value

A data frame with one row per event-time and columns term, event_time, n_cohorts, estimate, std.error, statistic, p.value, and (when conf.int = TRUE) conf.low / conf.high.

Details

The eventStudy() output stores its confidence-interval bounds (ci_low / ci_high), which reflect the fit's ci_type (#197): simultaneous (family-wise, uniform) by default, or pointwise when the fit used ci_type = "pointwise". When conf.int = TRUE (the default), conf.low / conf.high PASS THROUGH those stored bounds rather than recomputing from estimate +/- z * se — so the tidied event-study CIs agree with print / summary / plot and with simultaneousCIs() under the default. When conf.int = FALSE, the CI columns are omitted. (Degenerate event times carry NA bounds under both ci_type settings.)

Examples

if (FALSE) { # \dontrun{
  res <- fetwfeWithSimulatedData(
    simulateData(genCoefs(G = 3, T = 6, d = 2, density = 0.5, eff_size = 2),
                 N = 120, sig_eps_sq = 1, sig_eps_c_sq = 0.5, seed = 123)
  )
  broom::tidy(eventStudy(res))
} # }