Per-cohort average treatment effects

Extracts per-cohort ATT estimates from a fitted FETWFE / ETWFE / BETWFE / twfeCovs object as a tidy data frame. Parallel to eventStudy() for event-time aggregation: the per-cohort information is already available via result$catt_df, and cohortStudy() surfaces it through a discoverable function with its own help page (so users can reach it via ?cohortStudy without having to know the slot name).

The function is a pass-through on result$catt_df modulo class: the columns, their values, and their order are unchanged. The returned object carries class c("cohortStudy", "catt_df", "data.frame"). The cohortStudy class dispatches the broom tidier tidy.cohortStudy(); the catt_df class preserves the helpful-error layer (introduced in fetwfe 1.11.0) that intercepts pre-1.11.0 Title-Case column names (Cohort, Estimated TE, etc.) with a migration message; the data.frame base preserves the standard data-frame methods (print, head, nrow, dplyr::filter, etc.).

Usage

cohortStudy(result)

Arguments

result

A fitted object from fetwfe(), etwfe(), betwfe(), or twfeCovs() (or their *WithSimulatedData() wrapper analogs, which return the same classes).

Value

A data frame with class c("cohortStudy", "catt_df", "data.frame") containing one row per treated cohort and columns:

cohort

Character; the cohort label (the calendar time at which the cohort first received treatment).

estimate

Numeric; the per-cohort ATT estimate.

se

Numeric; standard error for the per-cohort ATT (NA when the Gram matrix is singular or, for fetwfe() / betwfe(), the bridge penalty zeroed out the cohort).

ci_low, ci_high

Numeric; the stored lower and upper confidence-interval bounds, reflecting the fit's ci_type – simultaneous (family-wise) by default, or pointwise 1 - alpha Wald bounds when the fit used ci_type = "pointwise" (alpha is the value passed at fit time).

p_value

Numeric; follows the fit's ci_type. Under "pointwise", the two-sided Wald p-value (2 * pnorm(-|estimate / se|)); under "simultaneous" (the default), the single-step max-T multiplicity-adjusted (family-wise) p-value matching the simultaneous band (#200). NA when se is 0 or NA.

selected

(fetwfe() / betwfe() only.) Logical; TRUE when the bridge penalty left the cohort's ATT nonzero. Absent for etwfe() and twfeCovs(), which do not perform selection.

Use tidy(cohortStudy(result)) (with the broom package loaded) to reshape to broom convention (term, estimate, std.error, statistic, p.value, conf.low, conf.high, optionally selected); see tidy.cohortStudy().

See also

eventStudy() for the parallel event-time accessor; cohortTimeATTs() for the fully disaggregated per-(cohort, time) accessor; tidy.cohortStudy() for broom-shape translation.

Examples

if (FALSE) { # \dontrun{
  coefs <- genCoefs(G = 3, T = 6, d = 2, density = 0.5, eff_size = 2)
  dat <- simulateData(coefs, N = 120, sig_eps_sq = 1, sig_eps_c_sq = 0.5, seed = 123)
  res <- fetwfeWithSimulatedData(dat)
  cs <- cohortStudy(res)
  cs
  # Broom-shape translation:
  if (requireNamespace("broom", quietly = TRUE)) {
    broom::tidy(cs)
  }
} # }