Multiple placebocheks unified in just one list or coefplot

Unifies shift_border, cutoff2polygon, assign_treated in one function to carry out a myriad of placebo checks at once. The output is either a data.frame (with or without geometry of the respective placeboline) or a coefplot. Requires operations data.frame that contains all desired operations (columns shift.x, shift.y, scale, angle, orientation.1, orientation.2, endpoint.1, endpoint.2), if you don't need a certain operation just use default values (e.g. 0 for angle and 1 for scale), but the column has to be there.

create_placebos(
  data,
  cutoff,
  formula,
  operations,
  bw_dist,
  coefplot = FALSE,
  geometry = FALSE
)

Arguments

data: sf data.frame that contains all units of observation
cutoff: initial RD cutoff as an sj object
formula: provide the formula you want to use for OLS, omit the treatetment dummy (if you want a univariate regression just on "treated", then provide y ~ 1 as formula)
operations: container that has all the information in it on how to change the border for each placeboregression
bw_dist: what is the distance for the bandwith (in CRS units, thus ideally metres)
coefplot: provide coefplot instead of a data.frame
geometry: set to TRUE if you want to plot all the lines of the used placebo borders

Value

either a coefplot or data.frame containing results of placebo regressions

Examples

points_samp.sf <- sf::st_sample(polygon_full, 100) # create points
# make it an sf object bc st_sample only created the geometry list-column (sfc):
points_samp.sf <- sf::st_sf(points_samp.sf)
# add a unique ID to each observation:
points_samp.sf$id <- 1:nrow(points_samp.sf)
points_samp.sf$treated <- assign_treated(points_samp.sf, polygon_treated, id = "id")
#> Warning: attribute variables are assumed to be spatially constant throughout all geometries
operations.df <- data.frame(operation = c("shift"),
                            shift.x = c(0),
                            shift.y = c(0),
                            scale = 1,
                            angle = 0,
                            orientation.1 = c("west"),
                            orientation.2 = c("west"),
                            endpoint.1 = c(.8),
                            endpoint.2 = c(.2))
create_placebos(data = points_samp.sf, cutoff = cut_off,
formula = id ~ 1, operations = operations.df, bw_dist = 3000)
#> Simple feature collection with 1 feature and 0 fields
#> Geometry type: MULTILINESTRING
#> Dimension:     XY
#> Bounding box:  xmin: 366034.5 ymin: 1673763 xmax: 400561.6 ymax: 1734272
#> Projected CRS: WGS 84 / UTM zone 43N
#>                       border_sfc
#> 1 MULTILINESTRING ((366034.5 ...
#> Warning: attribute variables are assumed to be spatially constant throughout all geometries
#> [1] "HERE"
#>   id estimate std.error statistic   p.value rd.estimate     const       HC0
#> 1  1 8.966667  14.53169  0.617042 0.5471117    34.49697 0.6695952 0.6263489
#>         HC1       HC2       HC3 HC4 operation shift.x shift.y scale angle
#> 1 0.6205987 0.5747843 0.6113464  NA     shift       0       0     1     0
#>   orientation.1 orientation.2 endpoint.1 endpoint.2
#> 1          west          west        0.8        0.2