Package {nonabsdid}

Type:	Package
Title:	Visualize Heterogeneity-Robust Event Studies for Non-Absorbing Treatments
Version:	0.4.1
Date:	2026-06-23
Description:	Runs several heterogeneity-robust difference-in-differences (DID) event-study estimators for non-absorbing (i.e., treatment can switch on and off over time, allowing treatment reversal) binary treatments through their respective packages, harmonizes their output onto a common time axis and tidy data structure, and overlays them in a single 'ggplot2' panel for visual comparison. Supported estimators include those provided by 'DIDmultiplegtDYN', 'PanelMatch', and 'fect', with an optional naive two-way fixed-effects reference series via 'fixest'. The underlying methods are respectively described in Clement de Chaisemartin and Xavier D'Haultfoeuille. "Difference-in-Differences Estimators of Intertemporal Treatment Effects." The Review of Economics and Statistics (2026) <doi:10.1162/rest_a_01414>, Kosuke Imai, In Song Kim, and Erik H. Wang. "Matching methods for causal inference with time‐series cross‐sectional data." American Journal of Political Science 67.3 (2023) <doi:10.1111/ajps.12685>, Licheng Liu, Ye Wang, and Yiqing Xu. "A practical guide to counterfactual estimators for causal inference with time‐series cross‐sectional data." American Journal of Political Science 68.1 (2024) <doi:10.1111/ajps.12723>, and Laurent R. Bergé, Kyle Butts, and Grant McDermott. "Fast and user-friendly econometrics estimations: The R package 'fixest'." arXiv preprint (2026) <doi:10.48550/arXiv.2601.21749>. A single nabs_event_study() wrapper runs any supported estimator with a common interface; nabs_event_study_simple() provides a one-line front door for quick exploratory runs; the S3 generic as_nabs_event_study() coerces estimator output into a tidy stable schema; and nabs_event_plot() overlays multiple methods on a single 'ggplot2' panel, with optional naive two-way fixed effects drawn in a neutral color as a reference.
License:	MIT + file LICENSE
Encoding:	UTF-8
Depends:	R (≥ 4.1.0)
Imports:	cli, dplyr, ggplot2, rlang, stats, tibble
Suggests:	DIDmultiplegtDYN, polars, PanelMatch, fect, fixest, haven, knitr, rmarkdown, testthat (≥ 3.0.0), vdiffr, withr
Config/testthat/edition:	3
URL:	https://github.com/takuma1102/nonabsdid, https://takuma1102.github.io/nonabsdid/
BugReports:	https://github.com/takuma1102/nonabsdid/issues
VignetteBuilder:	knitr
Additional_repositories:	https://rpolars.r-universe.dev
Config/roxygen2/version:	8.0.0
NeedsCompilation:	no
Packaged:	2026-06-23 07:21:40 UTC; 81809
Author:	Takuma Iwasaki [aut, cre]
Maintainer:	Takuma Iwasaki <iwasakit@stanford.edu>
Repository:	CRAN
Date/Publication:	2026-06-23 08:50:08 UTC

nonabsdid: Side-by-Side Event-Study Comparison for Heterogeneous DiD

Description

The 'nonabsdid' package provides a single, consistent interface for running, tidying, and plotting event-study estimates from several heterogeneity-robust difference-in-differences estimators that support non-absorbing (switching on/off) treatments:

Details

• **DCDH** — de Chaisemartin & D'Haultfoeuille, via 'DIDmultiplegtDYN::did_multiplegt_dyn()'.

• **PanelMatch** — Imai, Kim & Wang, via 'PanelMatch::PanelMatch()' / 'PanelMatch::PanelEstimate()' with pre-treatment results from 'PanelMatch::placebo_test()'.

• **IFE / Imputation** — Liu, Wang & Xu, via 'fect::fect()'.

The user-facing API has three pieces:

• [nabs_event_study()] runs one of the estimators with a unified argument set and returns its native object plus a tidy tibble.

• [as_nabs_event_study()] is an S3 generic that coerces native estimator objects into a stable tidy tibble (the *nabs_event_study_tbl* schema).

• [nabs_event_plot()] takes one or more *nabs_event_study_tbl* objects and overlays them on a single ggplot2 panel, optionally with a naive two-way fixed effects (TWFE) reference series in a neutral color.

Tidy schema

All tidiers return a tibble with class 'c("nabs_event_study_tbl", "tbl_df", ...)' and the following columns:

'time'

Integer relative period (0 = treatment onset).

'estimate'

Point estimate.

'std.error'

Standard error (may be 'NA' when the estimator only reports CI bounds, e.g. some 'fect' configurations).

'conf.low', 'conf.high'

Lower / upper bound of the 'conf.level' confidence interval.

'window'

'"pre"' if 'time < 0', otherwise '"post"'.

'method'

Method label, e.g. '"DCDH"', '"PanelMatch"', '"IFE"', or '"TWFE"'.

'outcome'

Outcome variable name (when known), else 'NA'.

Author(s)

Maintainer: Takuma Iwasaki iwasakit@stanford.edu (ORCID)

Authors:

• Takuma Iwasaki iwasakit@stanford.edu (ORCID)

See Also

Useful links:

• https://github.com/takuma1102/nonabsdid

• https://takuma1102.github.io/nonabsdid/

• Report bugs at https://github.com/takuma1102/nonabsdid/issues

Collapse effect cells back onto an event-study path

Description

Aggregates a 'nabs_effect_cell_tbl' over cohorts to recover a one-dimensional path, returning a 'nabs_event_study_tbl' that plugs straight into [nabs_event_plot()]. This makes explicit that the event study is the cohort-collapsed view of the same cells.

Usage

aggregate_effects(cells, by = c("event_time", "calendar_time"))

Arguments

Details

Point estimates are averaged across cohorts (weighted by 'n' when present). Re-aggregated standard errors are **not** computed here – collapsing SEs correctly needs the estimator's replicate draws – so 'std.error' and the CI columns are returned as 'NA'. Use this for a quick overlay, not for inference.

Value

A 'nabs_event_study_tbl' (with 'NA' standard errors).

Examples

raw <- expand.grid(cohort = 3:6, event_time = -2:4)
raw$estimate <- with(raw, ifelse(event_time < 0, 0, 0.2 * event_time))
cells <- as_nabs_effect_cells(raw, method = "FE")
aggregate_effects(cells)

Coerce an estimator result to a tidy cohort-by-time effect-cell tibble

Description

'as_nabs_effect_cells()' is an S3 generic that converts the native output of a supported estimator into a *cohort x time* effect-cell schema – the input for [plot_effect_matrix()] heatmaps. It is the two-dimensional companion to [as_nabs_event_study()]: where the event-study schema collapses everything onto a single relative-time axis, this schema keeps the cohort (treatment onset period) as a second dimension so heterogeneity *across* cohorts stays visible.

Usage

as_nabs_effect_cells(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'data.frame'
as_nabs_effect_cells(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'did_multiplegt_dyn'
as_nabs_effect_cells(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'fect'
as_nabs_effect_cells(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  axis = c("event", "calendar"),
  weighted = TRUE,
  ...
)

Arguments

Details

## DCDH method

Expects a 'did_multiplegt_dyn' object **run with the 'by' option**, where the ‘by' variable is a unit-level onset cohort (e.g. each unit’s first treated period). When 'by' is set, the object is reshaped into one sublist per 'by' level, each carrying its own event-study 'plot$data' ('Time', 'Estimate', 'LB.CI', 'UB.CI', and sometimes 'SE'). This method walks those sublists and stacks them into the cohort-by-time schema, shifting the axis so onset sits at 'event_time = 0' (the same '-1' shift the event-study tidier applies).

Building the cohort 'by' variable and running DCDH for you is exactly what [nabs_effect_cells()] with 'method = "DCDH"' does; call the generic directly only when you already have a 'by'-run object in hand.

SEs are the estimator's own ('se_method = "native"') when 'Time'-level SEs are present in the plot data; otherwise they are recovered from the symmetric ‘LB.CI'/'UB.CI' bounds ('se_method = "ci"'), which is exact for DCDH’s normal CIs, so 'show_se' works either way.

## fect method

Uses 'fect::imputed_outcomes()' (fect >= 2.4.0), the documented long-form accessor that returns one row per treated cell with columns 'id', 'time', 'event.time', 'cohort', 'eff', and 'W.agg'. The cell estimate for each '(cohort, event_time)' group is the 'W.agg'-weighted mean of the cell-level effects 'eff' (set 'weighted = FALSE' for an unweighted mean).

Standard errors come from the bootstrap surface: when the fit was produced with 'se = TRUE' and 'keep.sims = TRUE', 'imputed_outcomes(replicates = TRUE)' is re-aggregated within each replicate, and the cell SE is the standard deviation across replicates (with percentile CIs). Without stored sims the SE / CI columns are 'NA' and 'se_method' is '"none"'.

Value

A tibble of class '"nabs_effect_cell_tbl"', one row per '(cohort, event_time)' cell, with columns documented in [new_effect_cell_tbl()].

Status

This is an **experimental** feature line, separate from the stable event-study API. Only the 'fect' family ('IFE' / 'FE' / 'MC') and 'DCDH' ('DIDmultiplegtDYN') are supported; 'PanelMatch' is deliberately omitted for now because a faithful cohort breakdown there needs the matched-set bootstrap to be re-aggregated by cohort, which is out of scope for this pass.

Cohort and event-time conventions

* 'cohort' is the treatment **onset calendar period** (the first period a unit is treated). For repeated on/off treatment this is the *first* onset, so interpret later periods through the estimator's own carryover handling. * 'event_time' is the relative period with '0' at onset, matching the 'nabs_event_study_tbl' convention. For 'fect' this is computed directly as 'calendar_time - cohort'; for 'DCDH' it is the native event-study axis shifted so onset sits at '0'. * The 'fect' surface only covers **treated** cells, so its matrix spans 'event_time >= 0'. 'DCDH' run with placebos additionally yields the pre-period ('event_time < 0') cells.

See Also

[plot_effect_matrix()] to draw the heatmap, [nabs_effect_cells()] to fit and tidy in one step, [aggregate_effects()] to collapse cells back onto an event-study path.

Examples

# The data.frame escape hatch needs no estimator packages.
raw <- expand.grid(cohort = 3:5, event_time = 0:3)
raw$estimate  <- with(raw, 0.1 * event_time + 0.05 * (cohort - 4))
raw$std.error <- 0.08
cells <- as_nabs_effect_cells(raw, method = "DCDH", outcome = "y")
cells

Coerce an estimator result to a tidy event-study tibble

Description

'as_nabs_event_study()' is an S3 generic that converts the native output object of a supported estimator into the unified *nabs_event_study_tbl* schema used by [nabs_event_plot()]. Methods exist for objects of class '"did_multiplegt_dyn"' (from 'DIDmultiplegtDYN'), '"PanelEstimate"' (from 'PanelMatch'), '"fect"' (from 'fect'), and '"fixest"' (from 'fixest', used for the naive TWFE reference series).

Usage

as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'fixest'
as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'did_multiplegt_dyn'
as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'fect'
as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'list'
as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'nabs_event_study_result'
as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'nabs_event_study_simple'
as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  ...
)

## S3 method for class 'PanelEstimate'
as_nabs_event_study(
  x,
  method = NULL,
  outcome = NA_character_,
  conf.level = 0.95,
  pre_obj = NULL,
  add_reference = TRUE,
  ...
)

Arguments

Details

A 'data.frame' method is also provided as an escape hatch: it accepts any frame that already contains 'time' and 'estimate' columns and fills in the rest of the schema if missing.

## fixest method

Extracts coefficients on 'time_to_event' interactions of the form 'time_to_event::<k>' or 'time_to_event::<k>:<interaction>', the coefficient names produced by 'fixest::i()'. These are treated as event-study *levels* (the classic absorbing-treatment parametrisation). Standard errors come from the model's clustered VCOV; confidence intervals use the normal approximation and 'conf.level'.

Note that [naive_twfe()] does not fit this absorbing parametrisation itself – it uses a distributed-lag design in treatment levels – but this method is retained so that models you fit yourself with 'fixest::i()' can still be tidied.

## fect method

'fect::fect()' returns event-study coordinates in '$time' and '$att', with confidence-interval bounds in the two-column matrix '$att.bound'. Standard errors are pulled from '$est.att[, "S.E."]' when available; if the object was fit without 'se = TRUE', only the point estimates are returned and SE / CI columns are filled with 'NA'.

The 'method' label is auto-detected from 'x$method', the option that was passed to 'fect::fect()':

• '"fe"' -> '"FE"' (two-way fixed-effects imputation; Borusyak-style)

• '"ife"' -> '"IFE"' (interactive fixed effects; Bai 2009)

• '"mc"' -> '"MC"' (matrix completion; Athey et al. 2021)

Pass an explicit 'method' argument to override this auto-detected label.

## PanelMatch method

For 'PanelMatch::PanelEstimate()' the post-treatment leads are stored as '$estimate' / '$standard.error' (singular). The pre-treatment placebo results from 'PanelMatch::placebo_test()' use '$estimates' / '$standard.errors' (plural). To produce a single event-study path, pass the placebo object via 'pre_obj':

  pm <- PanelMatch::PanelMatch(...)
  pe <- PanelMatch::PanelEstimate(pm, panel.data = pd)
  pl <- PanelMatch::placebo_test(pm, panel.data = pd, plot = FALSE)
  tidy <- as_nabs_event_study(pe, pre_obj = pl)

A 'time = -1' reference point with 'estimate = 0' is inserted so that the event-study path is anchored at t = -1, matching common practice and the 'did' / 'fixest::iplot' convention. Disable with 'add_reference = FALSE'.

Value

A tibble of class '"nabs_event_study_tbl"' with one row per relative period and the columns documented in the package overview.

Examples

# The data.frame escape hatch needs no estimator packages: pass a frame
# that already has `time` and `estimate`; the remaining schema columns
# (including CIs derived from `std.error`) are filled in automatically.
raw <- data.frame(
  time      = -3:4,
  estimate  = c(-0.05, 0.01, 0.00, 0.02, 0.30, 0.42, 0.38, 0.50),
  std.error = 0.12
)
tidy_fit <- as_nabs_event_study(raw, method = "DCDH", outcome = "y")

# With the DCDH estimator installed, coerce its native object directly.
if (requireNamespace("DIDmultiplegtDYN", quietly = TRUE) &&
    requireNamespace("polars", quietly = TRUE)) {
  set.seed(1)
  library(polars)
  panel <- expand.grid(id = 1:60, t = 1:10)
  panel$d <- with(panel, as.integer(
    (id %% 4 == 1 & t %in% 4:7) |
    (id %% 4 == 2 & t %in% 5:8) |
    (id %% 4 == 3 & t %in% 6:9)
  ))
  panel$y <- 0.2 * panel$t + 0.5 * panel$d + rnorm(nrow(panel))

  fit <- DIDmultiplegtDYN::did_multiplegt_dyn(
    df = panel,
    outcome = "y",
    group = "id",
    time = "t",
    treatment = "d",
    effects = 3,
    placebo = 2
  )
  as_nabs_event_study(fit, outcome = "y")
}

Fit an estimator and return cohort-by-time effect cells

Description

'nabs_effect_cells()' is the cohort-matrix counterpart to [nabs_event_study()]: it fits one supported estimator and returns the result already tidied into the 'nabs_effect_cell_tbl' schema, ready for [plot_effect_matrix()]. It wires up the per-estimator machinery that a cohort breakdown needs – a unit-level onset cohort for 'DCDH', and 'keep.sims = TRUE' for 'fect' bootstrap cell SEs – so you do not have to.

Usage

nabs_effect_cells(
  data,
  outcome,
  treatment,
  unit,
  time,
  method = c("DCDH", "IFE", "FE", "MC"),
  lags = 6L,
  leads = 8L,
  controls = NULL,
  cluster = unit,
  conf.level = 0.95,
  axis = c("event", "calendar"),
  dcdh_strategy = c("loop", "by"),
  nboots = 200L,
  max_cohorts = 30L,
  ...
)

Arguments

Value

A list of class '"nabs_effect_cells_result"' with elements 'cells' (an 'nabs_effect_cell_tbl'), 'fit' (native object, or a list of them for the DCDH loop), and 'call'.

Status

Experimental, and intentionally limited to 'DCDH' and the 'fect' family ('IFE' / 'FE' / 'MC'). 'PanelMatch' is not supported here.

See Also

[plot_effect_matrix()], [as_nabs_effect_cells()].

Examples

if (requireNamespace("fect", quietly = TRUE)) {
  set.seed(1)
  panel <- expand.grid(id = 1:80, t = 1:12)
  onset <- c(`1` = 4, `2` = 6, `3` = 8)[as.character(panel$id %% 4)]
  panel$d <- as.integer(!is.na(onset) & panel$t >= onset)
  panel$y <- 0.2 * panel$t + 0.4 * panel$d + rnorm(nrow(panel))
  res <- nabs_effect_cells(panel, outcome = "y", treatment = "d",
                           unit = "id", time = "t", method = "FE",
                           nboots = 50)
  res$cells
}

Plot one or more event-study tibbles on a single panel

Description

Overlays event-study estimates from any combination of supported estimators on a single ggplot2 panel. Two visual encodings are available via 'style':

Usage

nabs_event_plot(
  ...,
  style = c("prepost_color", "method_shape"),
  connect = FALSE,
  connect_linewidth = 0.4,
  reference = NULL,
  reference_color = "grey20",
  palette = "default",
  shapes = NULL,
  xlim = NULL,
  ylim = NULL,
  dodge = 0.5,
  point_size = 2.5,
  errorbar_width = 0.1,
  x_break_by = 2,
  show_pre_post_legend = TRUE,
  xlab = "Relative time to treatment change",
  ylab = "Estimated effect",
  base_size = 11
)

Arguments

Details

* '"prepost_color"' (default) – each method gets its own color, with separate shades for pre- and post-treatment periods, mirroring common conventions in DCDH-style plots. Points are drawn as circles throughout. * '"method_shape"' – each method gets a single color *and* a single marker shape. Pre and post periods share both the color and the shape; they are told apart only by their position relative to time 0. Because method is double-encoded (color + shape), this style stays legible in grayscale.

An optional 'reference' series – typically a naive TWFE fit from [naive_twfe()] – is drawn in a neutral color (default black) so the reader can see what the heterogeneity-robust estimators are correcting against.

Set ‘connect = TRUE' to join each series’ point estimates with a thin line, in addition to the points and error bars.

Value

A 'ggplot' object.

Examples

dcdh_tidy <- as_nabs_event_study(
  data.frame(
    time = -2:3,
    estimate = c(-0.06, -0.02, 0.10, 0.22, 0.28, 0.31),
    std.error = 0.08
  ),
  method = "DCDH",
  outcome = "y"
)

ife_tidy <- as_nabs_event_study(
  data.frame(
    time = -2:3,
    estimate = c(-0.04, 0.00, 0.08, 0.18, 0.25, 0.27),
    std.error = 0.10
  ),
  method = "IFE",
  outcome = "y"
)

nabs_event_plot(dcdh_tidy, ife_tidy, xlim = c(-2, 3))
nabs_event_plot(dcdh_tidy, ife_tidy, style = "method_shape", connect = TRUE)

Run an event-study estimator with a unified interface

Description

'nabs_event_study()' is a thin wrapper around the three supported estimators (DCDH, PanelMatch, IFE/fect) that takes a single, common argument set and dispatches to the correct underlying package. It is **not** intended to expose every option of every estimator; for that, call the underlying packages directly and tidy their output with [as_nabs_event_study()].

Usage

nabs_event_study(
  data,
  outcome,
  treatment,
  unit,
  time,
  method = c("DCDH", "PanelMatch", "IFE", "FE", "MC"),
  lags = 6L,
  leads = 8L,
  controls = NULL,
  cluster = unit,
  conf.level = 0.95,
  cv = NULL,
  nboots = NULL,
  r = NULL,
  k = NULL,
  nlambda = NULL,
  vartype = NULL,
  se = NULL,
  parallel = FALSE,
  cores = NULL,
  number.iterations = NULL,
  se.method = NULL,
  run_placebo = NULL,
  num.cores = NULL,
  ...
)

Arguments

Details

What it does cover:

• Variable names (outcome, treatment, unit, time),

• Pre/post window length ('lags', 'leads'),

• Optional covariates and clustering,

• Reasonable defaults that match the three packages' typical use.

Value

A list of class '"nabs_event_study_result"' with elements:

'tidy'

An 'nabs_event_study_tbl'.

'fit'

The native estimator object (for diagnostics).

'call'

The call that produced it.

Examples

 if (requireNamespace("DIDmultiplegtDYN", quietly = TRUE) &&
     requireNamespace("polars", quietly = TRUE)) {
  set.seed(1)
  library(polars)
  panel <- expand.grid(id = 1:60, t = 1:10)
  panel$d <- with(panel, as.integer(
    (id %% 4 == 1 & t %in% 4:7) |
    (id %% 4 == 2 & t %in% 5:8) |
    (id %% 4 == 3 & t %in% 6:9)
  ))
  panel$y <- 0.2 * panel$t + 0.5 * panel$d + rnorm(nrow(panel))

  res_dcdh <- nabs_event_study(
    panel,
    outcome = "y",
    treatment = "d",
    unit = "id",
    time = "t",
    method = "DCDH",
    lags = 2,
    leads = 2
  )
  res_dcdh$tidy
}

One-line exploratory front door for non-absorbing event studies

Description

'nabs_event_study_simple()' is a deliberately opinionated convenience wrapper for the *first 30 seconds* of an analysis. You give it your data and the four column names that identify outcome / treatment / unit / time, and it tries to give you a sensible event-study figure with as little typing as possible.

Usage

nabs_event_study_simple(
  data,
  outcome,
  treatment,
  unit,
  time,
  methods = c("DCDH", "FE"),
  include_twfe = TRUE,
  lags = NULL,
  leads = NULL,
  controls = NULL,
  verbose = TRUE,
  full = FALSE,
  max_units = 5000L,
  sample_seed = 1L,
  keep_fits = FALSE,
  ...
)

Arguments

Details

By default it runs **all three** heterogeneity-robust estimators (DCDH, PanelMatch, IFE) plus a naive TWFE reference, and returns a single overlay plot along with the tidy tibbles and raw fits. Use it to *see the picture quickly*; for a careful, publication-ready result, switch to [nabs_event_study()] and tune options per estimator.

If a particular estimator's package is not installed, that estimator is silently skipped with a message and the rest are still attempted. This is intentional: the goal of '_simple()' is to give you *something* to look at even if your environment isn't fully provisioned.

Errors from a single estimator (for instance, PanelMatch failing because there are too few clean controls in the lag window) are caught, reported as a warning, and the remaining estimators continue.

Value

A list of class '"nabs_event_study_simple"' with elements:

'plot'

A 'ggplot' object; the overlay figure.

'tidy'

A single combined 'nabs_event_study_tbl' with all methods.

'per_method'

Named list of per-method tidy tibbles.

'fits'

Named list of native estimator objects.

'twfe'

The TWFE reference (or 'NULL').

'call'

The matched call.

Examples

 if (requireNamespace("DIDmultiplegtDYN", quietly = TRUE) &&
     requireNamespace("polars", quietly = TRUE)) {
  set.seed(1)
  library(polars)
  panel <- expand.grid(id = 1:60, t = 1:10)
  panel$d <- with(panel, as.integer(
    (id %% 4 == 1 & t %in% 4:7) |
    (id %% 4 == 2 & t %in% 5:8) |
    (id %% 4 == 3 & t %in% 6:9)
  ))
  panel$y <- 0.2 * panel$t + 0.5 * panel$d + rnorm(nrow(panel))

  res <- nabs_event_study_simple(
    panel,
    outcome = "y",
    treatment = "d",
    unit = "id",
    time = "t",
    methods = "DCDH",
    include_twfe = FALSE,
    lags = 2,
    leads = 2,
    verbose = FALSE
  )
  res$tidy
}

Read a Stata .dta file into an analysis-ready data frame

Description

'nabs_read_dta()' is a thin convenience layer over [haven::read_dta()] that smooths out the two places where freshly imported Stata data tends to trip up R estimation packages:

Usage

nabs_read_dta(
  path,
  labelled = c("factor", "numeric", "keep"),
  missings = c("na", "keep"),
  encoding = NULL,
  verbose = TRUE,
  ...
)

Arguments

Details

• **Labelled columns.** Stata value labels arrive in R as 'haven_labelled' vectors, which many modeling functions (including the estimator packages wrapped by nonabsdid) do not understand. By default these are converted to factors; set 'labelled = "numeric"' to drop the labels and keep the underlying codes instead.

• **Extended missing values.** Stata's '.a'–'.z' arrive as *tagged* 'NA's, which compare and print like ordinary 'NA' but can survive into model matrices in surprising ways. By default all tagged 'NA's are collapsed to regular 'NA'.

Variable labels (Stata's 'label variable') are preserved as '"label"' attributes on each column; they are harmless to the estimators and often useful for plot labels.

You rarely need to call this function yourself: [nabs_event_study()] and [nabs_event_study_simple()] accept a path to a '.dta' file as their 'data' argument and route it through 'nabs_read_dta()' automatically.

Value

A tibble.

See Also

[nabs_write_dta()] for the reverse direction, and the "nonabsdid for Stata users" vignette ('vignette("nonabsdid-for-stata-users")') for a full Stata-to-R walk-through.

Examples

if (requireNamespace("haven", quietly = TRUE)) {
  # Round-trip a small labelled panel through a temporary .dta file.
  tmp <- tempfile(fileext = ".dta")
  panel <- data.frame(id = rep(1:3, each = 2), t = rep(1:2, 3),
                      d = c(0, 1, 0, 0, 1, 1),
                      y = rnorm(6))
  haven::write_dta(panel, tmp)

  mydata <- nabs_read_dta(tmp)
  head(mydata)
}

Write event-study results to a Stata .dta file

Description

'nabs_write_dta()' exports an 'nabs_event_study_tbl' – or anything that [as_nabs_event_study()] can coerce into one, including the result objects returned by [nabs_event_study()] and [nabs_event_study_simple()] – to a Stata '.dta' file via [haven::write_dta()].

Usage

nabs_write_dta(x, path, version = 14, label = NULL, verbose = TRUE)

Arguments

Details

The tidy schema uses dots in some column names ('std.error', 'conf.low', 'conf.high'), which are not valid Stata variable names. These are renamed to underscore versions ('std_error', 'conf_low', 'conf_high') on the way out; any other invalid characters are likewise replaced with '_'.

This makes the "estimate in R, post-process in Stata" workflow a one-liner: a Stata-using coauthor can rebuild the event-study figure with 'twoway rcap'/'scatter', or feed the estimates into their own tables.

Value

The path, invisibly.

See Also

[nabs_read_dta()] for the reverse direction.

Examples

if (requireNamespace("haven", quietly = TRUE)) {
  tidy <- as_nabs_event_study(
    data.frame(time = -2:3,
               estimate = c(0.02, -0.01, 0, 0.4, 0.5, 0.45),
               std.error = 0.1),
    method = "DCDH", outcome = "y"
  )
  tmp <- tempfile(fileext = ".dta")
  nabs_write_dta(tidy, tmp)

  haven::read_dta(tmp)
}

Estimate a naive two-way fixed-effects (TWFE) event study

Description

Runs a basic event-study TWFE regression of 'outcome' on leads and lags of the treatment, with unit and time fixed effects, using 'fixest::feols()'. The result is **deliberately unsophisticated** – the point of 'nonabsdid' is to contrast this naive benchmark against heterogeneity-robust estimators (DCDH, 'fect', PanelMatch).

Usage

naive_twfe(
  data,
  outcome,
  treatment,
  unit,
  time,
  lags = 12L,
  leads = 6L,
  controls = NULL,
  cluster = unit,
  conf.level = 0.95
)

Arguments

Details

Unlike a classic event study, 'naive_twfe()' does **not** assume the treatment is absorbing. It is built for binary treatments that can switch on *and off* over time (e.g. a policy that is repealed, a subsidy that lapses). It fits a distributed-lag TWFE in the treatment *levels*,

y_{it} = \alpha_i + \gamma_t + \sum_{k} \beta_k D_{i,t+k} + \varepsilon_{it},

i.e. the outcome on the leads and lags of the treatment indicator with unit and time fixed effects. The coefficient on lag 'k' is reported at event time '+k' and the coefficient on lead 'k' at event time '-k', so the path is defined relative to a treatment *change* rather than to a single absorbing onset. Event time '-1' is the omitted reference. Each \beta_k is a partial correlation, not a heterogeneity-robust dynamic effect – that is the point of the benchmark.

The naming of 'lags'/'leads' follows the package convention used elsewhere (and in the README): 'lags' counts pre-periods, 'leads' counts post-periods, so 'lags = 6, leads = 8' yields event times on '[-6, 8]'.

Coefficients and standard errors are read directly from the fitted model (clustered as requested); the reference period '-1' is reported as exactly zero.

Missing treatment values are read as untreated ('0') when the leads and lags are constructed. For this naive benchmark that is usually innocuous, but if treatment missingness is itself informative it can bias the reference path; the heterogeneity-robust estimators handle missingness on their own terms.

Value

An 'nabs_event_study_tbl' with 'method = "TWFE"'. The fitted 'fixest' model is attached as the '"fit"' attribute.

Examples

df <- data.frame(
  id = rep(1:4, each = 8),
  yr = rep(1:8, times = 4),
  d  = c(rep(0, 8),
         0, 0, 1, 1, 1, 0, 0, 0,
         0, 0, 0, 1, 1, 1, 1, 0,
         rep(0, 8)),
  y  = rnorm(32)
)
naive_twfe(df, outcome = "y", treatment = "d",
           unit = "id", time = "yr", lags = 2, leads = 3)

Plot a cohort-by-time effect matrix as a heatmap

Description

Draws one or more 'nabs_effect_cell_tbl' objects as cohort (rows) by relative or calendar time (columns) heatmaps, with fill encoding the point estimate on a diverging scale centred at zero.

Usage

plot_effect_matrix(
  ...,
  axis = c("event", "calendar"),
  facet = TRUE,
  show_estimates = FALSE,
  show_se = FALSE,
  digits = 2,
  text_size = 2.6,
  title = NULL,
  caption = NULL,
  low = "#3182BD",
  mid = "#F7F7F7",
  high = "#DE2D26",
  limits = NULL,
  na_color = "grey92",
  xlab = NULL,
  ylab = "Onset cohort",
  legend_title = "Effect",
  base_size = 11
)

Arguments

Details

The intended use is **one method per plot**: a single-method call gets the method as its title automatically. Passing several methods facets them with a shared scale, but that side-by-side view gets crowded quickly, so for careful comparison prefer separate per-method heatmaps.

Value

A 'ggplot' object.

Examples

raw <- expand.grid(cohort = 3:6, event_time = -2:4)
raw$estimate  <- with(raw, ifelse(event_time < 0, 0,
                                  0.15 * event_time + 0.05 * (cohort - 4)))
raw$std.error <- 0.07
cells <- as_nabs_effect_cells(raw, method = "FE", outcome = "y")
plot_effect_matrix(cells)                                   # auto title "FE"
plot_effect_matrix(cells, show_estimates = TRUE, show_se = TRUE)