Trim observations with extreme propensity scores by replacing them with NA,
effectively removing those units from downstream analyses. The returned object
has the same length (or dimensions) as the input, with trimmed entries set to
NA. After trimming, refit the propensity score model on the retained
observations with ps_refit().
Usage
ps_trim(
.propensity,
method = c("ps", "adaptive", "pctl", "pref", "cr", "optimal", "density", "resid"),
lower = NULL,
upper = NULL,
.exposure = NULL,
.focal_level = NULL,
.reference_level = NULL,
...,
.sigma = NULL,
.density = "normal",
.treated = NULL,
.untreated = NULL,
ps = lifecycle::deprecated()
)Arguments
- .propensity
A numeric vector of propensity scores in (0, 1) for binary exposures, a matrix / data frame where each column gives the propensity score for one level of a categorical exposure, or a fitted model (see Fitted models and Trimming a dose model in Details). A data frame trimmed for a binary exposure is reduced to a single column: the second column of a two column data frame, which is the probability of the second level in the layout model predictions come in, and the first column otherwise. The column taken is announced;
options(propensity.quiet = TRUE)silences the announcement. A matrix is held to the same open interval as a vector, so a score of exactly 0 or 1 in any cell is refused and a separated multinomial fit cannot be repaired by trimming it: setting an extreme score to missing gains nothing from a score already at an endpoint.ps_trunc()reads the closed interval for a categorical matrix and is the repair; see Propensity scores at 0 and 1 inwt_ate(). A data frame holding a.pred_classcolumn, which a fitted tidymodels classification model returns when no prediction type is named, carries predicted levels rather than probabilities and is refused with an error of classpropensity_df_class_column_error.- method
Trimming method. One of:
"ps"(default): Fixed threshold. Observations with propensity scores outside[lower, upper]are trimmed. For categorical exposures, observations where any column falls belowlower(the symmetric threshold delta) are trimmed."adaptive": Data-driven threshold that minimizes the asymptotic variance of the IPW estimator (Crump et al., 2009). Thelowerandupperarguments are ignored. The threshold adapts to the estimated scores, tightening under poor overlap, and the estimand becomes the population it keeps.ps_trunc(method = "adaptive")instead adapts to the sample size alone and keeps every unit."pctl": Quantile-based. Observations outside the[lower, upper]quantiles of the propensity score distribution are trimmed. Defaults:lower = 0.05,upper = 0.95."pref": Preference score trimming. Transforms propensity scores to the preference scale (Walker et al., 2013) and trims outside[lower, upper]. Requires.exposure. Binary exposures only. Defaults:lower = 0.3,upper = 0.7."cr": Common range (clinical equipoise). Trims to the overlap region of the propensity score distributions across exposure groups. Requires.exposure. Binary exposures only. Thelowerandupperarguments are ignored. When the two distributions do not overlap at all, so that the lowest score among the focal units sits above the highest among the reference units, the overlap region is empty and every observed unit is trimmed. That is a truthful record of an empty region rather than an error, and it differs deliberately fromps_trunc(), which refuses the same data with an error of classpropensity_no_overlap_errorbecause there is no range left to bound the scores to."optimal": Multi-category optimal trimming (Yang et al., 2016). Categorical exposures only. Requires.exposure."density": Quantile floor on the conditional density of a continuous exposure. Units whose conditional density at their observed dose falls below itslowerquantile are trimmed. Default:lower = 0.01."resid": Bound on the absolute standardized residual of a continuous exposure. Units more thanupperspreads from their predicted dose are trimmed.upperhas no default.
"density"and"resid"need the residuals and the family of the model that fit the exposure's conditional mean, so they accept only a dose model (see Trimming a dose model in Details). They refuse a vector or matrix of values, a binomial or quasibinomialglm, and amultinomwith an error of classpropensity_method_error. A dose model accepts only these two methods, and refuses every other one, including the default"ps"whenmethodis not supplied, with the same class.For categorical exposures, only
"ps"and"optimal"are supported.- lower, upper
Numeric thresholds whose interpretation depends on
method:"ps": absolute propensity score bounds (defaults: 0.1, 0.9). For categorical exposures, onlyloweris used, as the symmetric threshold delta, and it defaults to 0.1, the defaultps_trunc()uses as well. Withkexposure levels, a threshold of1/kor larger cannot be met by every column of a row that sums to one, and is an error."pctl": quantile probabilities (defaults: 0.05, 0.95)."pref": preference score bounds (defaults: 0.3, 0.7)."adaptive","cr","optimal": ignored (thresholds are data-driven)."density":loweris the quantile probability of the floor, in (0, 0.5) (default 0.01).upperis refused with an error of classpropensity_unsupported_arg_error."resid":upperis the bound on the absolute standardized residual, a single positive finite number, and is required (an error of classpropensity_missing_arg_errorwithout it).loweris refused with an error of classpropensity_unsupported_arg_error.
A
"density"loweror"resid"upperout of range is refused with an error of classpropensity_range_error.- .exposure
An exposure variable. Required for
"pref","cr"(binary vector), and"optimal"(factor or character). For"density"and"resid", the dose is read from the model's response unless supplied here. Not required for other methods.- .focal_level
The value of
.exposurerepresenting the focal (treated) group, used by"pref"and"cr". Every binary coding honors it: 0/1 numeric, logical, two-level factor, and two-level character exposures are all coded with the named level as focal, and a level the exposure never takes is an error. With no level named, a binary exposure defaults to its higher level, which is1for a 0/1 exposure,TRUEfor a logical one, and the second of the two levels a factor or character exposure takes. Levels a factor declares but never takes are not candidates. Naming any other level reverses the coding, so.propensitymust then hold the probability of the named level.- .reference_level
The value of
.exposurerepresenting the reference (control) group. Naming it makes the exposure's other level focal, with the same consequence for.propensity, and a level the exposure never takes is an error. Automatically detected if not supplied.- ...
Additional arguments passed to methods.
- .sigma
For
"density"and"resid", a single residual spread to read the conditional density at, recorded as"supplied". With none supplied, the spread is the one the density family estimates from the residuals, as inwt_ate(): the root mean square, unless the family estimates a scale of its own. A spread for each unit is refused with an error of classpropensity_sigma_error, and a spread supplied with a family that estimates its own scale is refused with an error of classpropensity_density_error. The other methods refuse any.sigmawith an error of classpropensity_sigma_error.- .density
For
"density"and"resid", the family of the conditional density, in any formwt_ate()accepts:"normal"(the default),"laplace","kernel", a specification such asdens_t(), or a function of the standardized residual."resid"accepts onlydens_normal(),dens_t(), anddens_laplace(), and refuses a kernel or user-written density with an error of classpropensity_density_error. The other methods refuse any family but the normal with the same class.- .treated
![[Deprecated]](figures/lifecycle-deprecated.svg)
Use .focal_levelinstead.- .untreated
- Use
.reference_levelinstead. - ps
- Use
.propensityinstead. A call that namespsmust name the arguments after it as well, since a positional argument binds to.propensity.
Value
A ps_trim object (a numeric vector with class "ps_trim", or a
matrix with class "ps_trim_matrix"). Trimmed observations are NA.
Metadata is stored in the "ps_trim_meta" attribute and can be accessed
with ps_trim_meta(). Key fields include:
method: the trimming method usedkeep_idx: integer indices of retained observationstrimmed_idx: integer indices of trimmed (NA) observationsn_obs: the number of observations those indices describeMethod-specific fields such as
cutoff(adaptive),q_lower/q_upper(pctl),cr_lower/cr_upper(cr),delta(categorical ps), orlambda(optimal)focal_inverted(vector scores only):TRUEwhen the scores are one minus the probability a fitted model reports, because the model was trimmed with its first level named as focal, andFALSEotherwise, including for every vector of scores supplied directly
A trim of a dose model records method, lower ("density", else
NULL), upper ("resid", else NULL), threshold (the realized floor
on the conditional density), sigma (the spread it was read at),
sigma_kind ("pooled", "mle", or "supplied"), density (the
density specification), keep_idx, trimmed_idx, and n_obs.
Details
How trimming works
Trimming identifies observations with extreme (near 0 or 1) propensity
scores and sets them to NA. These observations are excluded from
subsequent weight calculations and effect estimation. The goal is to
remove units that lack sufficient overlap between exposure groups, which
would otherwise receive extreme weights and destabilize estimates.
Choosing a method
Use
"ps"when you have a specific threshold in mind or want a simple default.Use
"adaptive"for a principled, data-driven cutoff that targets variance reduction.Use
"pctl"to trim a fixed percentage of extreme values from each tail.Use
"pref"when you want to restrict to the region of clinical equipoise based on the preference score.Use
"cr"to restrict to the common support region where both exposure groups have observed propensity scores.Use
"optimal"for multi-category (3+) exposures; this is the only data-driven method available for categorical treatments.Use
"density"or"resid"for a continuous exposure, on its dose model.
For a binary or categorical exposure, trimming and refitting is the first
recourse when the analysis can accept a change of estimand. When every unit
must be kept, bound the scores with ps_trunc() instead, for a binary
exposure with method = "adaptive", or bound the weights with
wt_trunc().
Typical workflow
Fit a propensity score model
Apply
ps_trim()to flag extreme valuesCall
ps_refit()to re-estimate propensity scores on the retained sampleCompute weights with
wt_ate()or another weight function
Fitted models
.propensity can be the fitted propensity score model instead of the scores
it reports. A binomial stats::glm() and a two-level nnet::multinom() are
read as one score per unit; a nnet::multinom() of three or more levels is
read as one column per level. Those are the shapes predict(fit, type = "response") and fitted(fit) give, and trimming a fit trims exactly what
trimming those values would.
The methods that read an exposure ("pref", "cr", and every method on the
categorical route) take it from the model when .exposure is not supplied,
announcing the variable they read; options(propensity.quiet = TRUE)
silences the announcement. An .exposure you supply is used instead, and a
categorical model's columns are matched to its levels by name, so an exposure
whose levels are ordered differently is still trimmed against the right
column.
Trimming a dose model
For a continuous exposure, .propensity can be the model of the exposure's
conditional mean: a stats::lm(), a glm of the gaussian() family (or
another family whose variance is constant), a MASS::rlm(), or an
mgcv::gam(), the models wt_ate() reads a dose from. A glm whose spread
changes with its mean, such as poisson(), is refused with an error of
class propensity_model_family_error. Only "density" and "resid" apply.
For a dose, the generalized propensity score is the conditional density
\(f(a \mid x)\) (Hirano and Imbens, 2004), and a unit whose observed dose
has a small conditional density gets a large weight, as a unit with a
propensity score near 0 does for a binary exposure. Both methods read the
fitted conditional mean mu, the spread sigma of the residuals under the
family in .density, the standardized residual z = (a - mu) / sigma, and
the conditional density f = g(z) / sigma, where g is the family's
standardized density. "density" keeps the units whose f is at least its
lower quantile. "resid" keeps those whose |z| is at most upper; for
the symmetric unimodal families it accepts, that is the floor
g(upper) / sigma on f, which is the threshold it records. The trim is
one-sided, at the low end of the density, since only a small density makes
a large weight (Branson et al., 2024).
The retained values are the conditional means, and the trimmed ones are
NA. A unit with a missing mean or dose takes no part in the trim.
.focal_level, .reference_level, and their deprecated forms are refused
with an error of class propensity_focal_level_error, since a dose has no
levels.
A trim changes the estimand. The analysis describes the units whose observed dose was plausible under the model, not the full sample, and which units those are depends on the threshold. Branson et al. (2024) define the trimmed dose response at each dose among the units whose conditional density at that dose exceeds the threshold, a population that changes with the dose; this trim is its sample analogue at each unit's own observed dose rather than the same population. A dose-response curve fit to the trimmed sample is a curve for the units it keeps.
After trimming, ps_refit() refits the conditional mean on the retained
rows and re-estimates the spread there under the recorded family, and
wt_ate() builds the weights from the refit under the family and spread
the record holds. ipw() refuses those weights, as it refuses every
weight built from a trimmed score.
The other way to hold down extreme weights for a dose is wt_trunc(),
which bounds the weights and keeps every unit; see Truncating weights or
trimming the density there for the simulation results behind this
summary. With a density family that fits the residuals, a density trim at
lower = 0.01 did not help: at an effective sample size of 64% of the
sample, its interval covered 0.843, against 0.933 for the untrimmed
weights under the same fixed-weight interval. A loose bound on the weights
left that coverage where it was and lowered the root mean squared error.
With heavy-tailed residuals read through a normal density, or a residual
spread the model left out, the trim covered 0.979 to 0.989, against 0.618
to 0.881 for the untrimmed weights and 0.644 to 0.906 for the bound,
because it removed the units the misspecified density fit worst. Two
cautions go with those numbers. The trim's intervals were computed on the
trimmed sample, held the weights fixed, and were conservative, and its bias
was measured against the effect in the full population, so part of that
bias is the change of estimand. "resid" was not simulated; the results
carry over to it only because a bound on the standardized residual is a
floor on the conditional density.
A family that fits the residuals is the better repair where it applies.
Under heavy tails, dens_t() with 4 degrees of freedom at its default
scale had about a third of the root mean squared error of the bounded
normal weights and an M-estimation interval that covered 0.926 to 0.939
with standard errors close to right, and it kept every unit. The trim's
coverage was higher, but its interval was conservative. Check the family
first; trim when the units at the low-density end are ones the analysis
should not describe, and say so when reporting the estimand.
Object behavior
Arithmetic operations on ps_trim objects return plain numeric vectors,
since transformed propensity scores (e.g., 1/ps) are no longer propensity
scores. Trimmed values propagate as NA in calculations; use na.rm = TRUE
where appropriate.
When combining ps_trim objects with c(), trimming parameters must match.
Mismatched parameters trigger a warning and return a numeric vector.
Use ps_trim_meta() to inspect the trimming metadata, including the method,
cutoffs, and which observations were retained or trimmed.
Missing values
A propensity score that arrives missing is not one this function removed, so
it takes no part in the trimming record: it joins neither the retained nor
the trimmed positions, and is_unit_trimmed() reports FALSE for it. The
value propagates as NA into the result. For a matrix of categorical
propensity scores, a row with a missing cell comes back exactly as it
arrived, since there is no complete probability vector to place against a
threshold.
The methods that read a cutoff off the scores read it off the scores they
have. "adaptive", "pctl", "cr", and "optimal" all work their cutoffs
out from the complete scores or rows, so the cutoff is the one the same call
would produce with the missing observations dropped.
"pref" centers its preference scores on the proportion exposed across the
whole sample, which is a fact about the exposure rather than about the
scores, so a unit whose propensity score is missing still counts toward it
and the cutoff is not the one the shorter call would produce. That unit's own
preference score is missing, which leaves it outside both the retained and
the trimmed positions like any other missing score.
A missing exposure is a different matter, and "pref" and "cr" refuse one
with an error of class propensity_missing_value_error. Their cutoffs come
from the exposure groups, and a unit that belongs to neither leaves them
undefined. Remove or impute the missing exposure values first.
The trimming record
A ps_trim records which units were trimmed as positions among the
observations it was written for, along with how many observations that was.
Operations that hand this package the subscript re-index those positions onto
the result: subsetting with [, sort(), rep(), and na.omit() all
return a record written for what they return, and a subscript naming a
position more than once reports that unit at every place it now holds.
unique() does the same when it can, as described below.
Operations that change how many observations there are without supplying a
subscript cannot re-index the record, and it is dropped rather than worked
out from the values, since reading membership back from the NA pattern
would report a propensity score that arrived missing as one this package
removed. vctrs::vec_slice(), which is how filtering, joining, and grouped
verbs in dplyr reach a column, is the usual route, and combining two or more
vectors with c() is another, because concatenation appends one set of
observations to another. The record is dropped without comment on every
route: most of these length changes build vectors the caller never holds,
such as the pieces a grouped verb slices a column into or the rows a tibble
slices off to print, so a warning would mostly describe something that is not
the result. The values, the class, and the method and its cutoffs are
untouched.
A combine drops the positions even when it is handed a single vector. c()
of one ps_trim returns it unchanged, record included, but
vctrs::vec_c(x), dplyr::bind_rows(df), vctrs::vec_rbind(df), and an
ungrouped dplyr::reframe() rebuild the column, so a later ps_refit() or
is_unit_trimmed() on the result refuses it.
unique() keeps one element for each distinct value, or one row for each
distinct row of a matrix of scores, and that element or row stands for every
unit holding the same scores. A matrix comes back as a matrix of the same
class with its column names. The record is re-indexed onto the result when
all of the merged units share one status, and dropped otherwise: a trimmed
score and one that arrived missing are both NA, and a trimmed row is NA
throughout, so a vector or matrix holding both returns a single NA element
or row that neither status describes.
as.data.frame() and tibble::as_tibble() turn a matrix of scores into a
data frame whose columns are ps_trim vectors carrying the matrix's record, one
row per unit, so the weight functions read the data frame as they read the
matrix and a subset of its rows re-indexes the record. A data frame whose
columns are not all trimmed with the same record is refused by the weight
functions with an error of class propensity_matrix_type_error. A matrix
without column names gives columns named V1, V2, and so on, which the
weight functions then refuse because they name no exposure level, so name
the columns after the exposure levels before converting.
A record can also outlive the observations it describes, because it travels
by routes vctrs does not see: growing a ps_trim by subassignment carries it
across the length change. is_unit_trimmed() and ps_refit() therefore
check that the record covers the object they are given and raise an error of
class propensity_missing_meta_error when it does not, rather than name
trimmed units at stale positions.
That check compares how many observations the record was written for against
how many the object holds, which a reordering does not change, so a route
that could reorder the observations without saying how drops the positions
instead, at any length: vctrs::vec_slice(), dplyr::arrange(),
dplyr::filter(), and vctrs::vec_assign() and the helpers built on it
return a ps_trim whose record keeps its method and cutoffs and names no
units, and is_unit_trimmed() and ps_refit() refuse it. Subsetting with
[, sort(), unique(), and rep() know where the units went and
re-index the record, and [<- and is.na<- move no unit and keep it, so
reorder with [, or put the propensity scores in the order you want before
trimming them.
Casting a numeric vector into a ps_trim with vctrs::vec_cast() is a type
operation and not a trimming. The result is described by the method and
cutoffs of the target and records that none of the arriving values was
trimmed, so it can hold scores outside those cutoffs, including 0 and 1.
Call ps_trim() on the scores to trim them.
References
Branson, Z., Kennedy, E. H., Balakrishnan, S., & Wasserman, L. (2024). Causal effect estimation after propensity score trimming with continuous treatments. arXiv preprint arXiv:2309.00706.
Crump, R. K., Hotz, V. J., Imbens, G. W., & Mitnik, O. A. (2009). Dealing with limited overlap in estimation of average treatment effects. Biometrika, 96(1), 187–199.
Hirano, K., & Imbens, G. W. (2004). The propensity score with continuous treatments. In Applied Bayesian Modeling and Causal Inference from Incomplete-Data Perspectives (pp. 73–84).
Walker, A. M., Patrick, A. R., Lauer, M. S., et al. (2013). A tool for assessing the feasibility of comparative effectiveness research. Comparative Effectiveness Research, 3, 11–20.
Yang, S., Imbens, G. W., Cui, Z., Faries, D. E., & Kadziola, Z. (2016). Propensity score matching and subclassification in observational studies with multi-level treatments. Biometrics, 72(4), 1055–1065.
See also
ps_trunc() for bounding (winsorizing) instead of discarding,
ps_refit() to re-estimate propensity scores after trimming,
ps_calibrate() for calibration-based adjustment,
ps_trim_meta() to inspect trimming metadata,
is_ps_trimmed() and is_unit_trimmed() for logical queries.
Examples
set.seed(2)
n <- 300
x <- rnorm(n)
z <- rbinom(n, 1, plogis(1.3 * x))
fit <- glm(z ~ x, family = binomial)
ps <- predict(fit, type = "response")
# Fixed threshold trimming (default)
trimmed <- ps_trim(ps, method = "ps", lower = 0.1, upper = 0.9)
trimmed
#> <ps_trim; trimmed 44 of 300[300]>
#> 1 2 3 4 5 6 7 8
#> 0.1780112 0.4934483 0.8725819 0.1353577 0.4025855 0.4752489 0.6683913 0.3506150
#> 9 10 11 12 13 14 15 16
#> NA 0.3831809 0.5738044 0.7467782 0.3038553 0.1508037 0.8997256 NA
#> 17 18 19 20 21 22 23 24
#> 0.7187207 0.4419184 0.7548592 0.5787647 NA 0.1244366 0.8728587 NA
#> 25 26 27 28 29 30 31 32
#> 0.4313639 NA 0.5939254 0.2474276 0.6938170 0.5298266 0.6778670 0.5399668
#> 33 34 35 36 37 38 39 40
#> 0.7707828 0.3366773 0.2037945 0.2476600 NA 0.1768609 0.2572600 0.3484614
#> 41 42 43 44 45 46 47 48
#> 0.3065403 NA 0.1895191 NA 0.6415565 NA 0.3300895 0.3990495
#> 49 50 51 52 53 54 55 56
#> 0.3683877 0.1246121 0.1902500 NA 0.2564150 0.8160957 0.1494022 NA
#> 57 58 59 60 61 62 63 64
#> 0.3247367 0.7345271 0.7859021 0.8849770 NA NA 0.2208817 0.4841803
#> 65 66 67 68 69 70 71 72
#> 0.6036086 0.1941979 NA 0.2790100 0.4585602 0.1783019 0.1731103 0.5439312
#> 73 74 75 76 77 78 79 80
#> 0.3822372 0.5796397 0.4114856 0.1759469 0.8218240 0.6877562 0.7649259 NA
#> 81 82 83 84 85 86 87 88
#> 0.7505008 NA 0.2641422 0.1005949 NA NA 0.2330120 0.3365148
#> 89 90 91 92 93 94 95 96
#> 0.3055473 0.5632496 0.8745082 0.8863194 0.1269292 0.1023453 NA 0.1166039
#> 97 98 99 100 101 102 103 104
#> NA 0.4322875 0.1893249 0.2462384 0.7703638 0.5197606 0.3273939 0.2099624
#> 105 106 107 108 109 110 111 112
#> 0.1851823 NA 0.7356255 NA 0.2954896 0.3163021 0.1530042 0.3471981
#> 113 114 115 116 117 118 119 120
#> 0.5921212 0.8328274 0.6227067 0.5867797 0.8065734 0.7877456 0.4663064 0.2023006
#> 121 122 123 124 125 126 127 128
#> 0.8087856 0.4774812 0.8903829 0.2928150 0.1499937 0.6139848 0.2290136 0.6467536
#> 129 130 131 132 133 134 135 136
#> NA NA 0.6627758 0.5441083 0.7165979 NA 0.8025574 0.7998822
#> 137 138 139 140 141 142 143 144
#> 0.7597742 0.6921037 NA NA 0.2509264 0.5711077 0.1970442 0.4590332
#> 145 146 147 148 149 150 151 152
#> 0.6800801 0.2329425 0.6525433 0.6180388 0.1970997 0.1584870 0.7452343 0.3731672
#> 153 154 155 156 157 158 159 160
#> 0.6727629 0.1889404 0.8164247 0.1043218 0.6858279 0.5895482 0.5223260 0.6558189
#> 161 162 163 164 165 166 167 168
#> 0.5672709 0.2368400 0.3418011 0.5540597 0.1083159 0.1806594 NA NA
#> 169 170 171 172 173 174 175 176
#> 0.1187105 0.7490531 0.7738375 0.7077039 0.4491471 0.5416643 0.1764396 0.2333126
#> 177 178 179 180 181 182 183 184
#> 0.3434474 0.1704628 0.7023006 NA 0.1524604 0.1153463 0.5651259 0.1351849
#> 185 186 187 188 189 190 191 192
#> 0.6161453 0.7945146 0.4382952 0.6065642 0.2328341 0.6027459 0.1139088 0.4037497
#> 193 194 195 196 197 198 199 200
#> 0.1040353 0.3422376 0.7735701 0.6661116 0.2893391 0.2011360 0.1863223 0.2107038
#> 201 202 203 204 205 206 207 208
#> 0.5327159 0.1544197 NA 0.5052145 0.1642856 0.5622725 0.3887452 0.3163883
#> 209 210 211 212 213 214 215 216
#> 0.6341619 0.5095796 0.7582940 0.2665601 NA NA 0.4774214 0.5852567
#> 217 218 219 220 221 222 223 224
#> 0.8030404 0.1067663 0.1338595 0.8855459 0.5671613 0.6707177 0.2000938 NA
#> 225 226 227 228 229 230 231 232
#> 0.4538222 0.5912550 0.3993802 0.7230688 0.3094663 NA 0.8702859 NA
#> 233 234 235 236 237 238 239 240
#> 0.2415827 0.4195113 NA 0.1483657 0.2541910 0.5957602 0.4626159 0.2496911
#> 241 242 243 244 245 246 247 248
#> 0.4020747 NA 0.7936546 0.6179515 0.6899355 0.2556513 0.4650462 0.5270157
#> 249 250 251 252 253 254 255 256
#> 0.2803714 0.2850014 0.6020472 0.3002118 0.3705819 0.3257820 0.7087374 0.5960756
#> 257 258 259 260 261 262 263 264
#> 0.3306472 0.3363861 0.7464321 0.3725508 0.8297224 0.3965473 0.3250342 NA
#> 265 266 267 268 269 270 271 272
#> 0.2345937 NA NA 0.5886632 0.6106387 0.4172303 NA 0.4137080
#> 273 274 275 276 277 278 279 280
#> 0.4312953 0.1206438 0.5164038 0.8752400 0.4176263 0.7591012 0.3016993 0.4527210
#> 281 282 283 284 285 286 287 288
#> NA 0.6501483 0.8168531 0.2393292 0.8311549 0.8851143 0.7936328 0.4317872
#> 289 290 291 292 293 294 295 296
#> 0.8222307 0.3983355 0.1356224 0.6302455 0.4602807 0.3527491 0.8562989 0.3153233
#> 297 298 299 300
#> 0.5741435 NA 0.1008201 0.2253829
# How many observations were trimmed?
sum(is_unit_trimmed(trimmed))
#> [1] 44
# Data-driven adaptive trimming
ps_trim(ps, method = "adaptive")
#> <ps_trim; trimmed 44 of 300[300]>
#> 1 2 3 4 5 6 7 8
#> 0.1780112 0.4934483 0.8725819 0.1353577 0.4025855 0.4752489 0.6683913 0.3506150
#> 9 10 11 12 13 14 15 16
#> NA 0.3831809 0.5738044 0.7467782 0.3038553 0.1508037 0.8997256 NA
#> 17 18 19 20 21 22 23 24
#> 0.7187207 0.4419184 0.7548592 0.5787647 NA 0.1244366 0.8728587 NA
#> 25 26 27 28 29 30 31 32
#> 0.4313639 NA 0.5939254 0.2474276 0.6938170 0.5298266 0.6778670 0.5399668
#> 33 34 35 36 37 38 39 40
#> 0.7707828 0.3366773 0.2037945 0.2476600 NA 0.1768609 0.2572600 0.3484614
#> 41 42 43 44 45 46 47 48
#> 0.3065403 NA 0.1895191 NA 0.6415565 NA 0.3300895 0.3990495
#> 49 50 51 52 53 54 55 56
#> 0.3683877 0.1246121 0.1902500 NA 0.2564150 0.8160957 0.1494022 NA
#> 57 58 59 60 61 62 63 64
#> 0.3247367 0.7345271 0.7859021 0.8849770 NA NA 0.2208817 0.4841803
#> 65 66 67 68 69 70 71 72
#> 0.6036086 0.1941979 NA 0.2790100 0.4585602 0.1783019 0.1731103 0.5439312
#> 73 74 75 76 77 78 79 80
#> 0.3822372 0.5796397 0.4114856 0.1759469 0.8218240 0.6877562 0.7649259 NA
#> 81 82 83 84 85 86 87 88
#> 0.7505008 NA 0.2641422 0.1005949 NA NA 0.2330120 0.3365148
#> 89 90 91 92 93 94 95 96
#> 0.3055473 0.5632496 0.8745082 0.8863194 0.1269292 0.1023453 NA 0.1166039
#> 97 98 99 100 101 102 103 104
#> NA 0.4322875 0.1893249 0.2462384 0.7703638 0.5197606 0.3273939 0.2099624
#> 105 106 107 108 109 110 111 112
#> 0.1851823 NA 0.7356255 NA 0.2954896 0.3163021 0.1530042 0.3471981
#> 113 114 115 116 117 118 119 120
#> 0.5921212 0.8328274 0.6227067 0.5867797 0.8065734 0.7877456 0.4663064 0.2023006
#> 121 122 123 124 125 126 127 128
#> 0.8087856 0.4774812 0.8903829 0.2928150 0.1499937 0.6139848 0.2290136 0.6467536
#> 129 130 131 132 133 134 135 136
#> NA NA 0.6627758 0.5441083 0.7165979 NA 0.8025574 0.7998822
#> 137 138 139 140 141 142 143 144
#> 0.7597742 0.6921037 NA NA 0.2509264 0.5711077 0.1970442 0.4590332
#> 145 146 147 148 149 150 151 152
#> 0.6800801 0.2329425 0.6525433 0.6180388 0.1970997 0.1584870 0.7452343 0.3731672
#> 153 154 155 156 157 158 159 160
#> 0.6727629 0.1889404 0.8164247 0.1043218 0.6858279 0.5895482 0.5223260 0.6558189
#> 161 162 163 164 165 166 167 168
#> 0.5672709 0.2368400 0.3418011 0.5540597 0.1083159 0.1806594 NA NA
#> 169 170 171 172 173 174 175 176
#> 0.1187105 0.7490531 0.7738375 0.7077039 0.4491471 0.5416643 0.1764396 0.2333126
#> 177 178 179 180 181 182 183 184
#> 0.3434474 0.1704628 0.7023006 NA 0.1524604 0.1153463 0.5651259 0.1351849
#> 185 186 187 188 189 190 191 192
#> 0.6161453 0.7945146 0.4382952 0.6065642 0.2328341 0.6027459 0.1139088 0.4037497
#> 193 194 195 196 197 198 199 200
#> 0.1040353 0.3422376 0.7735701 0.6661116 0.2893391 0.2011360 0.1863223 0.2107038
#> 201 202 203 204 205 206 207 208
#> 0.5327159 0.1544197 NA 0.5052145 0.1642856 0.5622725 0.3887452 0.3163883
#> 209 210 211 212 213 214 215 216
#> 0.6341619 0.5095796 0.7582940 0.2665601 NA NA 0.4774214 0.5852567
#> 217 218 219 220 221 222 223 224
#> 0.8030404 0.1067663 0.1338595 0.8855459 0.5671613 0.6707177 0.2000938 NA
#> 225 226 227 228 229 230 231 232
#> 0.4538222 0.5912550 0.3993802 0.7230688 0.3094663 NA 0.8702859 NA
#> 233 234 235 236 237 238 239 240
#> 0.2415827 0.4195113 NA 0.1483657 0.2541910 0.5957602 0.4626159 0.2496911
#> 241 242 243 244 245 246 247 248
#> 0.4020747 NA 0.7936546 0.6179515 0.6899355 0.2556513 0.4650462 0.5270157
#> 249 250 251 252 253 254 255 256
#> 0.2803714 0.2850014 0.6020472 0.3002118 0.3705819 0.3257820 0.7087374 0.5960756
#> 257 258 259 260 261 262 263 264
#> 0.3306472 0.3363861 0.7464321 0.3725508 0.8297224 0.3965473 0.3250342 NA
#> 265 266 267 268 269 270 271 272
#> 0.2345937 NA NA 0.5886632 0.6106387 0.4172303 NA 0.4137080
#> 273 274 275 276 277 278 279 280
#> 0.4312953 0.1206438 0.5164038 0.8752400 0.4176263 0.7591012 0.3016993 0.4527210
#> 281 282 283 284 285 286 287 288
#> NA 0.6501483 0.8168531 0.2393292 0.8311549 0.8851143 0.7936328 0.4317872
#> 289 290 291 292 293 294 295 296
#> 0.8222307 0.3983355 0.1356224 0.6302455 0.4602807 0.3527491 0.8562989 0.3153233
#> 297 298 299 300
#> 0.5741435 NA 0.1008201 0.2253829
# Quantile-based trimming at 5th and 95th percentiles
ps_trim(ps, method = "pctl")
#> <ps_trim; trimmed 30 of 300[300]>
#> 1 2 3 4 5 6 7
#> 0.17801124 0.49344831 0.87258189 0.13535765 0.40258554 0.47524889 0.66839132
#> 8 9 10 11 12 13 14
#> 0.35061504 NA 0.38318089 0.57380442 0.74677823 0.30385529 0.15080370
#> 15 16 17 18 19 20 21
#> 0.89972561 NA 0.71872070 0.44191837 0.75485924 0.57876473 NA
#> 22 23 24 25 26 27 28
#> 0.12443658 0.87285871 NA 0.43136394 NA 0.59392536 0.24742762
#> 29 30 31 32 33 34 35
#> 0.69381700 0.52982663 0.67786695 0.53996676 0.77078278 0.33667732 0.20379449
#> 36 37 38 39 40 41 42
#> 0.24766004 NA 0.17686094 0.25726004 0.34846140 0.30654026 NA
#> 43 44 45 46 47 48 49
#> 0.18951912 0.91394741 0.64155648 NA 0.33008953 0.39904946 0.36838768
#> 50 51 52 53 54 55 56
#> 0.12461207 0.19024995 NA 0.25641497 0.81609573 0.14940217 NA
#> 57 58 59 60 61 62 63
#> 0.32473671 0.73452714 0.78590206 0.88497697 NA NA 0.22088166
#> 64 65 66 67 68 69 70
#> 0.48418025 0.60360857 0.19419787 NA 0.27900998 0.45856017 0.17830185
#> 71 72 73 74 75 76 77
#> 0.17311028 0.54393115 0.38223719 0.57963968 0.41148556 0.17594694 0.82182398
#> 78 79 80 81 82 83 84
#> 0.68775620 0.76492590 0.09594356 0.75050082 0.06658957 0.26414219 0.10059488
#> 85 86 87 88 89 90 91
#> NA 0.90461881 0.23301201 0.33651483 0.30554734 0.56324964 0.87450818
#> 92 93 94 95 96 97 98
#> 0.88631941 0.12692921 0.10234530 0.08426251 0.11660386 NA 0.43228752
#> 99 100 101 102 103 104 105
#> 0.18932487 0.24623842 0.77036380 0.51976065 0.32739386 0.20996243 0.18518227
#> 106 107 108 109 110 111 112
#> NA 0.73562548 NA 0.29548965 0.31630208 0.15300421 0.34719807
#> 113 114 115 116 117 118 119
#> 0.59212123 0.83282741 0.62270671 0.58677974 0.80657336 0.78774559 0.46630644
#> 120 121 122 123 124 125 126
#> 0.20230064 0.80878565 0.47748117 0.89038286 0.29281504 0.14999367 0.61398482
#> 127 128 129 130 131 132 133
#> 0.22901364 0.64675362 0.06419209 NA 0.66277576 0.54410827 0.71659793
#> 134 135 136 137 138 139 140
#> NA 0.80255741 0.79988215 0.75977424 0.69210372 NA 0.09079429
#> 141 142 143 144 145 146 147
#> 0.25092645 0.57110768 0.19704419 0.45903320 0.68008008 0.23294255 0.65254331
#> 148 149 150 151 152 153 154
#> 0.61803876 0.19709968 0.15848705 0.74523426 0.37316715 0.67276290 0.18894043
#> 155 156 157 158 159 160 161
#> 0.81642474 0.10432180 0.68582789 0.58954824 0.52232604 0.65581895 0.56727085
#> 162 163 164 165 166 167 168
#> 0.23684003 0.34180113 0.55405967 0.10831585 0.18065940 NA NA
#> 169 170 171 172 173 174 175
#> 0.11871049 0.74905308 0.77383752 0.70770386 0.44914706 0.54166428 0.17643958
#> 176 177 178 179 180 181 182
#> 0.23331263 0.34344738 0.17046280 0.70230062 0.07304003 0.15246043 0.11534630
#> 183 184 185 186 187 188 189
#> 0.56512587 0.13518488 0.61614535 0.79451459 0.43829517 0.60656419 0.23283410
#> 190 191 192 193 194 195 196
#> 0.60274592 0.11390882 0.40374970 0.10403526 0.34223762 0.77357008 0.66611156
#> 197 198 199 200 201 202 203
#> 0.28933909 0.20113598 0.18632231 0.21070377 0.53271587 0.15441973 NA
#> 204 205 206 207 208 209 210
#> 0.50521454 0.16428559 0.56227253 0.38874517 0.31638831 0.63416193 0.50957963
#> 211 212 213 214 215 216 217
#> 0.75829404 0.26656013 0.90146489 0.09307013 0.47742144 0.58525674 0.80304044
#> 218 219 220 221 222 223 224
#> 0.10676633 0.13385955 0.88554593 0.56716130 0.67071770 0.20009381 0.91614132
#> 225 226 227 228 229 230 231
#> 0.45382217 0.59125501 0.39938022 0.72306882 0.30946625 NA 0.87028588
#> 232 233 234 235 236 237 238
#> NA 0.24158271 0.41951127 0.06438704 0.14836574 0.25419100 0.59576019
#> 239 240 241 242 243 244 245
#> 0.46261588 0.24969111 0.40207472 NA 0.79365463 0.61795146 0.68993549
#> 246 247 248 249 250 251 252
#> 0.25565132 0.46504618 0.52701568 0.28037136 0.28500140 0.60204716 0.30021180
#> 253 254 255 256 257 258 259
#> 0.37058188 0.32578201 0.70873738 0.59607558 0.33064721 0.33638608 0.74643214
#> 260 261 262 263 264 265 266
#> 0.37255079 0.82972242 0.39654734 0.32503416 NA 0.23459370 0.09722939
#> 267 268 269 270 271 272 273
#> 0.91692000 0.58866324 0.61063867 0.41723032 NA 0.41370803 0.43129530
#> 274 275 276 277 278 279 280
#> 0.12064385 0.51640377 0.87523995 0.41762629 0.75910122 0.30169925 0.45272098
#> 281 282 283 284 285 286 287
#> NA 0.65014831 0.81685306 0.23932916 0.83115493 0.88511432 0.79363282
#> 288 289 290 291 292 293 294
#> 0.43178717 0.82223069 0.39833546 0.13562237 0.63024555 0.46028071 0.35274910
#> 295 296 297 298 299 300
#> 0.85629892 0.31532329 0.57414353 NA 0.10082012 0.22538288
# Refit after trimming, then compute weights
trimmed <- ps_trim(ps, method = "adaptive")
refitted <- ps_refit(trimmed, fit)
wt_ate(refitted, .exposure = z)
#> ℹ Treating `.exposure` as binary
#> <psw{estimand = ate; trimmed}[300]>
#> [1] 1.240579 2.058164 1.179439 1.179070 2.479678 2.130709 1.546037 1.552398
#> [9] NA 2.593380 2.268587 1.386216 1.455212 1.200838 1.140328 NA
#> [17] 3.272545 1.782675 1.371331 2.292253 NA 1.163991 1.179032 NA
#> [25] 2.328539 NA 1.731055 3.842912 1.490847 1.926820 1.525053 1.893076
#> [33] 3.917615 2.914998 1.280093 1.352152 NA 1.238860 1.368835 1.547658
#> [41] 1.460487 NA 4.876301 NA 2.643752 NA 2.967353 2.499634
#> [49] 2.687495 1.164231 1.259096 NA 3.722181 1.266295 6.029169 NA
#> [57] 3.011351 1.409295 1.316469 1.161399 NA NA 1.307391 1.913671
#> [65] 1.704729 1.265160 NA 1.407989 1.832054 1.241014 1.233279 1.880192
#> [73] 2.599182 1.771319 1.698883 5.210593 4.889604 2.985370 1.353148 NA
#> [81] 1.379326 NA 1.381016 1.131873 NA NA 1.327354 1.521848
#> [89] 3.180870 2.219879 1.176614 1.159464 1.167411 1.134197 NA 1.153324
#> [97] NA 1.755289 1.257681 1.349711 3.911330 2.040050 2.989342 1.289839
#> [105] 1.251377 NA 3.455793 NA 3.277919 1.479963 5.902359 1.544890
#> [113] 1.736044 1.239690 2.526861 1.750976 1.281806 1.313320 1.855984 1.277750
#> [121] 4.594831 1.891632 7.509198 1.433882 1.199683 2.476368 1.320718 2.678015
#> [129] NA NA 1.558717 1.879621 1.444250 NA 4.467031 1.292873
#> [137] 1.362404 1.494456 NA NA 1.357789 1.796221 4.709750 1.833498
#> [145] 1.520225 1.327239 2.717307 1.666857 1.269646 1.211872 1.389090 1.603744
#> [153] 1.536293 4.889627 4.762739 1.136827 1.507810 1.743207 2.049824 2.740082
#> [161] 1.807638 1.333760 1.533168 2.179201 1.142162 1.244550 NA NA
#> [169] 7.402829 1.381999 1.337371 1.462140 2.244100 1.887538 1.238231 1.327856
#> [177] 2.863148 1.229363 3.113295 NA 1.203205 1.151622 2.228379 1.178829
#> [185] 2.488677 1.301856 1.772273 1.696843 1.327058 1.707044 1.149680 2.473178
#> [193] 1.136445 1.534110 3.959981 1.551162 1.427282 1.275928 4.950895 1.291019
#> [201] 1.917087 1.206013 NA 2.013803 1.220295 2.215480 1.641148 1.480137
#> [209] 2.596564 1.997821 1.365083 1.385341 NA NA 2.121784 1.755278
#> [217] 4.476669 1.140089 1.176986 1.160579 1.807966 1.540838 1.274301 NA
#> [225] 1.817720 1.738450 1.667666 1.431467 1.466275 NA 1.182816 NA
#> [233] 3.925953 2.388469 NA 1.197366 3.751310 2.377394 1.844505 1.355652
#> [241] 2.482541 NA 1.303304 1.667082 3.003822 3.732130 1.852048 1.936381
#> [249] 3.436024 1.419123 2.410589 1.448110 2.673096 1.499338 3.173717 1.725144
#> [257] 1.509464 2.917273 1.386860 1.602298 1.244567 1.660522 1.497792 NA
#> [265] 1.329995 NA NA 2.341051 1.686081 1.714101 NA 1.704739
#> [273] 2.328877 7.296804 2.027404 1.175543 1.715160 1.363621 1.451002 1.814421
#> [281] NA 1.587922 1.265073 1.337953 1.242313 1.161201 1.303341 1.753889
#> [289] 1.256446 2.503704 1.179438 2.572282 2.194293 1.557123 1.203649 1.477989
#> [297] 1.787285 NA 1.132171 1.314740
# Trim the scores a fitted model reports, reading the exposure off the model
ps_trim(fit, method = "cr")
#> ℹ Using exposure variable "z" from the propensity score model
#> <ps_trim; trimmed 27 of 300[300]>
#> 1 2 3 4 5 6 7
#> 0.17801124 0.49344831 0.87258189 0.13535765 0.40258554 0.47524889 0.66839132
#> 8 9 10 11 12 13 14
#> 0.35061504 NA 0.38318089 0.57380442 0.74677823 0.30385529 0.15080370
#> 15 16 17 18 19 20 21
#> NA NA 0.71872070 0.44191837 0.75485924 0.57876473 NA
#> 22 23 24 25 26 27 28
#> 0.12443658 0.87285871 NA 0.43136394 NA 0.59392536 0.24742762
#> 29 30 31 32 33 34 35
#> 0.69381700 0.52982663 0.67786695 0.53996676 0.77078278 0.33667732 0.20379449
#> 36 37 38 39 40 41 42
#> 0.24766004 0.06402613 0.17686094 0.25726004 0.34846140 0.30654026 0.04714018
#> 43 44 45 46 47 48 49
#> 0.18951912 NA 0.64155648 NA 0.33008953 0.39904946 0.36838768
#> 50 51 52 53 54 55 56
#> 0.12461207 0.19024995 NA 0.25641497 0.81609573 0.14940217 0.04671897
#> 57 58 59 60 61 62 63
#> 0.32473671 0.73452714 0.78590206 0.88497697 0.05903129 NA 0.22088166
#> 64 65 66 67 68 69 70
#> 0.48418025 0.60360857 0.19419787 0.04471987 0.27900998 0.45856017 0.17830185
#> 71 72 73 74 75 76 77
#> 0.17311028 0.54393115 0.38223719 0.57963968 0.41148556 0.17594694 0.82182398
#> 78 79 80 81 82 83 84
#> 0.68775620 0.76492590 0.09594356 0.75050082 0.06658957 0.26414219 0.10059488
#> 85 86 87 88 89 90 91
#> NA NA 0.23301201 0.33651483 0.30554734 0.56324964 0.87450818
#> 92 93 94 95 96 97 98
#> 0.88631941 0.12692921 0.10234530 0.08426251 0.11660386 NA 0.43228752
#> 99 100 101 102 103 104 105
#> 0.18932487 0.24623842 0.77036380 0.51976065 0.32739386 0.20996243 0.18518227
#> 106 107 108 109 110 111 112
#> NA 0.73562548 NA 0.29548965 0.31630208 0.15300421 0.34719807
#> 113 114 115 116 117 118 119
#> 0.59212123 0.83282741 0.62270671 0.58677974 0.80657336 0.78774559 0.46630644
#> 120 121 122 123 124 125 126
#> 0.20230064 0.80878565 0.47748117 0.89038286 0.29281504 0.14999367 0.61398482
#> 127 128 129 130 131 132 133
#> 0.22901364 0.64675362 0.06419209 0.06266942 0.66277576 0.54410827 0.71659793
#> 134 135 136 137 138 139 140
#> 0.04369801 0.80255741 0.79988215 0.75977424 0.69210372 NA 0.09079429
#> 141 142 143 144 145 146 147
#> 0.25092645 0.57110768 0.19704419 0.45903320 0.68008008 0.23294255 0.65254331
#> 148 149 150 151 152 153 154
#> 0.61803876 0.19709968 0.15848705 0.74523426 0.37316715 0.67276290 0.18894043
#> 155 156 157 158 159 160 161
#> 0.81642474 0.10432180 0.68582789 0.58954824 0.52232604 0.65581895 0.56727085
#> 162 163 164 165 166 167 168
#> 0.23684003 0.34180113 0.55405967 0.10831585 0.18065940 NA NA
#> 169 170 171 172 173 174 175
#> 0.11871049 0.74905308 0.77383752 0.70770386 0.44914706 0.54166428 0.17643958
#> 176 177 178 179 180 181 182
#> 0.23331263 0.34344738 0.17046280 0.70230062 0.07304003 0.15246043 0.11534630
#> 183 184 185 186 187 188 189
#> 0.56512587 0.13518488 0.61614535 0.79451459 0.43829517 0.60656419 0.23283410
#> 190 191 192 193 194 195 196
#> 0.60274592 0.11390882 0.40374970 0.10403526 0.34223762 0.77357008 0.66611156
#> 197 198 199 200 201 202 203
#> 0.28933909 0.20113598 0.18632231 0.21070377 0.53271587 0.15441973 NA
#> 204 205 206 207 208 209 210
#> 0.50521454 0.16428559 0.56227253 0.38874517 0.31638831 0.63416193 0.50957963
#> 211 212 213 214 215 216 217
#> 0.75829404 0.26656013 NA 0.09307013 0.47742144 0.58525674 0.80304044
#> 218 219 220 221 222 223 224
#> 0.10676633 0.13385955 0.88554593 0.56716130 0.67071770 0.20009381 NA
#> 225 226 227 228 229 230 231
#> 0.45382217 0.59125501 0.39938022 0.72306882 0.30946625 NA 0.87028588
#> 232 233 234 235 236 237 238
#> 0.04743962 0.24158271 0.41951127 0.06438704 0.14836574 0.25419100 0.59576019
#> 239 240 241 242 243 244 245
#> 0.46261588 0.24969111 0.40207472 NA 0.79365463 0.61795146 0.68993549
#> 246 247 248 249 250 251 252
#> 0.25565132 0.46504618 0.52701568 0.28037136 0.28500140 0.60204716 0.30021180
#> 253 254 255 256 257 258 259
#> 0.37058188 0.32578201 0.70873738 0.59607558 0.33064721 0.33638608 0.74643214
#> 260 261 262 263 264 265 266
#> 0.37255079 0.82972242 0.39654734 0.32503416 0.04998908 0.23459370 0.09722939
#> 267 268 269 270 271 272 273
#> NA 0.58866324 0.61063867 0.41723032 NA 0.41370803 0.43129530
#> 274 275 276 277 278 279 280
#> 0.12064385 0.51640377 0.87523995 0.41762629 0.75910122 0.30169925 0.45272098
#> 281 282 283 284 285 286 287
#> NA 0.65014831 0.81685306 0.23932916 0.83115493 0.88511432 0.79363282
#> 288 289 290 291 292 293 294
#> 0.43178717 0.82223069 0.39833546 0.13562237 0.63024555 0.46028071 0.35274910
#> 295 296 297 298 299 300
#> 0.85629892 0.31532329 0.57414353 NA 0.10082012 0.22538288
# Trim a dose model on the scale of its conditional density
dose <- 1 + 0.5 * x + rt(n, df = 3)
dose_fit <- lm(dose ~ x)
ps_trim(dose_fit, method = "density", lower = 0.05)
#> ℹ Using exposure variable "dose" from the propensity score model
#> <ps_trim; trimmed 15 of 300[300]>
#> [1] 0.42939235 1.09006968 1.94693677 0.28680810 0.92816170 1.05804921
#> [7] 1.40955162 0.83078134 2.18917407 0.89241182 1.23225111 1.57677128
#> [13] 0.73733962 0.34220638 2.06565480 -0.43428963 1.51377446 0.99904335
#> [19] 1.59575061 1.24117672 2.25412348 0.24433098 1.94803175 2.17096038
#> [25] 0.98019043 -0.52018258 1.26864305 0.61283227 1.46100564 NA
#> [31] 1.42847473 1.17197688 1.63443228 0.80362799 0.50282739 0.61338051
#> [37] -0.07695168 0.42592949 0.63573276 0.82661942 0.74290294 -0.21932966
#> [43] 0.46311104 2.13974897 1.35735721 2.19311121 0.79060333 0.92169244
#> [49] 0.86469980 0.24503825 0.46519850 2.23914942 0.63378747 1.75630633
#> [55] NA -0.22346747 0.77992308 1.54874422 1.67294893 1.99810063
#> [61] -0.11497789 2.21773756 0.54773574 1.07377241 1.28635301 0.47637020
#> [67] -0.24360284 0.68445115 1.02858683 0.43026445 0.41451402 1.17899374
#> [73] 0.89065668 1.24275403 0.94436412 0.42316534 1.77328293 1.44853829
#> [79] NA 0.11600668 1.58546323 -0.05849822 0.65142256 0.13907368
#> [85] -0.37129213 2.09001943 0.57812066 0.80330827 0.74084879 1.21334515
#> [91] 1.95459883 2.00392477 0.25429792 0.14750935 0.05332484 0.21185234
#> [97] 2.17383418 0.98184437 0.46255518 0.61002168 1.63339100 1.13633248
#> [103] 0.78523598 0.51934498 0.45059496 2.22799665 NA 2.20396205
#> [109] 0.71982466 0.76290580 NA 0.82417245 1.26535859 1.80713482
#> [115] 1.32173555 1.25566040 1.72896844 1.67777809 1.04227851 0.49877122
#> [121] 1.73522619 1.06198124 2.02192736 0.71416468 0.33942104 1.30549651
#> [127] 0.56823119 1.36731942 -0.07573634 -0.08699855 1.39846633 1.17930747
#> [133] 1.50917138 -0.25422887 1.71774635 NA 1.60750104 1.45746761
#> [139] 2.26588287 0.08927363 0.62104937 1.22740991 NA 1.02942388
#> [145] NA 0.57794987 1.37849621 1.31302677 0.48447213 0.36803416
#> [151] 1.57319103 0.87370106 1.41824676 0.46145377 1.75727019 0.15688221
#> [157] 1.44459956 1.26068221 1.14084917 1.38485812 1.22053498 0.58747867
#> [163] 0.81367151 1.19696789 0.17535374 0.43729835 2.24574142 -0.30490817
#> [169] 0.22076851 1.58207270 1.64206502 1.49010683 1.01190062 1.17498029
#> [175] 0.42465659 0.57885939 0.81688295 0.40633792 1.47869144 -0.01482674
#> [181] 0.34786509 0.20646291 1.21669791 0.28615913 1.30950600 1.69577790
#> [187] 0.99258284 1.29178761 0.57768316 1.28476948 NA 0.93028744
#> [193] 0.15553319 0.81452371 NA 1.40504011 0.70676339 0.49559335
#> [199] 0.45390678 0.52130612 1.15916550 0.35449252 2.73054913 1.11074987
#> [205] 0.38686049 1.21160039 0.90272781 0.76308099 1.34329166 1.11842326
#> [211] 1.60394523 0.65687238 2.07419130 0.10125236 1.06187604 1.25290208
#> [217] 1.71908694 0.16825996 0.28115746 2.00056180 1.22033889 1.41417151
#> [223] 0.49273803 2.15214956 1.02019530 1.26378323 0.92229838 1.52326968
#> [229] 0.74893483 2.35638147 1.93793228 -0.21640944 0.59892920 0.95888442
#> [235] NA NA 0.62864782 1.27198814 1.03575937 0.61815694
#> [241] 0.92722831 2.74124982 1.69346711 1.31286431 1.45300570 0.63202597
#> [247] 1.04005312 1.14911142 0.68742022 0.69745284 1.28348764 0.72974566
#> [253] 0.86883809 0.78201580 1.49230436 1.27256361 0.79171095 0.80305489
#> [259] 1.57596748 0.87254292 1.79740775 0.91710321 NA -0.19223190
#> [265] 0.58200010 0.12248144 2.15662179 1.25907579 1.29930313 0.95476587
#> [271] -0.68414308 0.94839325 0.98006748 0.22883164 1.13042501 1.95753594
#> [277] 0.95548128 1.60588243 0.73285216 1.01824291 2.22052580 1.37386241
#> [283] 1.75852700 0.59350771 1.80187769 1.99869379 1.69340860 0.98094845
#> [289] NA 0.92038387 0.28780107 NA 1.03163075 0.83489383
#> [295] 1.88582027 0.76091550 1.23286043 -0.30381489 0.14016646 0.55914528
ps_trim(dose_fit, method = "resid", upper = 3, .density = dens_t(4))
#> ℹ Using exposure variable "dose" from the propensity score model
#> <ps_trim; trimmed 14 of 300[300]>
#> [1] 0.42939235 1.09006968 1.94693677 0.28680810 0.92816170 1.05804921
#> [7] 1.40955162 0.83078134 2.18917407 0.89241182 1.23225111 1.57677128
#> [13] 0.73733962 0.34220638 2.06565480 -0.43428963 1.51377446 0.99904335
#> [19] 1.59575061 1.24117672 2.25412348 0.24433098 1.94803175 2.17096038
#> [25] 0.98019043 -0.52018258 1.26864305 0.61283227 1.46100564 NA
#> [31] 1.42847473 1.17197688 1.63443228 0.80362799 0.50282739 0.61338051
#> [37] -0.07695168 0.42592949 0.63573276 0.82661942 0.74290294 -0.21932966
#> [43] 0.46311104 2.13974897 1.35735721 2.19311121 0.79060333 0.92169244
#> [49] 0.86469980 0.24503825 0.46519850 2.23914942 0.63378747 1.75630633
#> [55] NA -0.22346747 0.77992308 1.54874422 1.67294893 1.99810063
#> [61] -0.11497789 2.21773756 0.54773574 1.07377241 1.28635301 0.47637020
#> [67] -0.24360284 0.68445115 1.02858683 0.43026445 0.41451402 1.17899374
#> [73] 0.89065668 1.24275403 0.94436412 0.42316534 1.77328293 1.44853829
#> [79] NA 0.11600668 1.58546323 -0.05849822 0.65142256 0.13907368
#> [85] -0.37129213 2.09001943 0.57812066 0.80330827 0.74084879 1.21334515
#> [91] 1.95459883 2.00392477 0.25429792 0.14750935 0.05332484 0.21185234
#> [97] 2.17383418 0.98184437 0.46255518 0.61002168 1.63339100 1.13633248
#> [103] 0.78523598 0.51934498 0.45059496 2.22799665 NA 2.20396205
#> [109] 0.71982466 0.76290580 NA 0.82417245 1.26535859 1.80713482
#> [115] 1.32173555 1.25566040 1.72896844 1.67777809 1.04227851 0.49877122
#> [121] 1.73522619 1.06198124 2.02192736 0.71416468 0.33942104 1.30549651
#> [127] 0.56823119 1.36731942 -0.07573634 -0.08699855 1.39846633 1.17930747
#> [133] 1.50917138 -0.25422887 1.71774635 NA 1.60750104 1.45746761
#> [139] 2.26588287 0.08927363 0.62104937 1.22740991 NA 1.02942388
#> [145] NA 0.57794987 1.37849621 1.31302677 0.48447213 0.36803416
#> [151] 1.57319103 0.87370106 1.41824676 0.46145377 1.75727019 0.15688221
#> [157] 1.44459956 1.26068221 1.14084917 1.38485812 1.22053498 0.58747867
#> [163] 0.81367151 1.19696789 0.17535374 0.43729835 2.24574142 -0.30490817
#> [169] 0.22076851 1.58207270 1.64206502 1.49010683 1.01190062 1.17498029
#> [175] 0.42465659 0.57885939 0.81688295 0.40633792 1.47869144 -0.01482674
#> [181] 0.34786509 0.20646291 1.21669791 0.28615913 1.30950600 1.69577790
#> [187] 0.99258284 1.29178761 0.57768316 1.28476948 NA 0.93028744
#> [193] 0.15553319 0.81452371 NA 1.40504011 0.70676339 0.49559335
#> [199] 0.45390678 0.52130612 1.15916550 0.35449252 2.73054913 1.11074987
#> [205] 0.38686049 1.21160039 0.90272781 0.76308099 1.34329166 1.11842326
#> [211] 1.60394523 0.65687238 2.07419130 0.10125236 1.06187604 1.25290208
#> [217] 1.71908694 0.16825996 0.28115746 2.00056180 1.22033889 1.41417151
#> [223] 0.49273803 2.15214956 1.02019530 1.26378323 0.92229838 1.52326968
#> [229] 0.74893483 2.35638147 1.93793228 -0.21640944 0.59892920 0.95888442
#> [235] -0.07431245 NA 0.62864782 1.27198814 1.03575937 0.61815694
#> [241] 0.92722831 2.74124982 1.69346711 1.31286431 1.45300570 0.63202597
#> [247] 1.04005312 1.14911142 0.68742022 0.69745284 1.28348764 0.72974566
#> [253] 0.86883809 0.78201580 1.49230436 1.27256361 0.79171095 0.80305489
#> [259] 1.57596748 0.87254292 1.79740775 0.91710321 NA -0.19223190
#> [265] 0.58200010 0.12248144 2.15662179 1.25907579 1.29930313 0.95476587
#> [271] -0.68414308 0.94839325 0.98006748 0.22883164 1.13042501 1.95753594
#> [277] 0.95548128 1.60588243 0.73285216 1.01824291 2.22052580 1.37386241
#> [283] 1.75852700 0.59350771 1.80187769 1.99869379 1.69340860 0.98094845
#> [289] NA 0.92038387 0.28780107 NA 1.03163075 0.83489383
#> [295] 1.88582027 0.76091550 1.23286043 -0.30381489 0.14016646 0.55914528
if (rlang::is_installed("nnet")) {
trt <- factor(sample(c("a", "b", "c"), n, replace = TRUE))
multinomial_fit <- nnet::multinom(trt ~ x, trace = FALSE)
ps_trim(multinomial_fit, method = "optimal")
}
#> ℹ Using exposure variable "trt" from the propensity score model
#> <ps_trim_matrix[300 x 3]; trimmed 0 of 300; method=optimal>
#> a b c
#> 1 0.2833559 0.3409328 0.3757113
#> 2 0.2986578 0.3516829 0.3496593
#> 3 0.3183352 0.3645221 0.3171427
#> 4 0.2800465 0.3385216 0.3814319
#> 5 0.2949150 0.3491145 0.3559705
#> 6 0.2979181 0.3511784 0.3509035
#> 7 0.3060233 0.3566205 0.3373562
#> 8 0.2926613 0.3475487 0.3597900
#> 9 0.3238455 0.3679135 0.3082410
#> 10 0.2940878 0.3485415 0.3573707
#> # ... with 290 more rows