Skip to contents

Wraps a long data frame – one row per unit and period – with the names of its unit, time and outcome columns, so that specifications can refer to columns by name. Any other columns (further outcomes, covariates, a display label) stay available: a specification's data$outcome and covariates$vars name them.

Usage

sdid_panel(data, unit, time, outcome = NULL, treatment = NULL, label = NULL)

Arguments

data

A data frame in long format.

unit, time

Names of the unit and time columns.

outcome

Name of the default outcome column, used when a specification does not name one. Optional.

treatment

Name of a 0/1 treatment column. Optional. When given, a specification that leaves design$treated_unit or design$treat_time unset takes them from it: the one unit that is ever treated, and its first treated period.

label

Name of a column holding a display name for each unit, used in plots and tables. Optional.

Value

An sdid_panel object: a list with data (the data frame, with the unit column coerced to character) and the column names.

Details

Units are stored as character, so numeric identifiers and names behave the same way everywhere, and times must be numeric (years, or any ordered numeric period index).

Examples

data("california_prop99", package = "synthdid")
panel <- sdid_panel(california_prop99, unit = "State", time = "Year",
                    outcome = "PacksPerCapita", treatment = "treated")
panel
#> 
#> ── sdid_panel 
#> • 39 units (State) x 31 periods (Year, 1970-2000)
#> • outcome PacksPerCapita
#> • treatment treated
#> • columns PacksPerCapita and treated