===========================================================
.___ __ __
_________________ __ __ __| _/|__|/ |_
/ ___\_` __ \__ \ | | \/ __ | | \\_ __\
/ /_/ > | \// __ \| | / /_/ | | || |
\___ /|__| (____ /____/\____ | |__||__|
/_____/ \/ \/
grep rough audit - static analysis tool
v2.8 written by @Wireghoul
=================================[justanotherhacker.com]===
r-cran-matchit-4.0.0/vignettes/assessing-balance.Rmd-117-
r-cran-matchit-4.0.0/vignettes/assessing-balance.Rmd:118:The third table (`Percent Balance Improvement`) contains the percent balance improvement for each covariate. This is computed as $100\frac{|\theta_M| - |\theta_U|}{|\theta_U|}$, where $\theta_M$ is a given balance statistic in the matched sample and $\theta_U$ is a the same balance statistic in the unmatched sample. Values between 0 and 100 indicate that balance improved after matching as measured by the statistic; values less than 0 indicate that balance got worse after matching. When balance is good on a covariate prior to matching, it can sometimes look like balance got a lot worse after matching even though the balance statistic is quite low, so these values should not be taken too seriously and should be used primarily as heuristics. Setting `un = FALSE` or `improvement = FALSE` would have suppressed the creation of this table.
r-cran-matchit-4.0.0/vignettes/assessing-balance.Rmd-119-
##############################################
r-cran-matchit-4.0.0/vignettes/matching-methods.Rmd-87-
r-cran-matchit-4.0.0/vignettes/matching-methods.Rmd:88:In `MatchIt`, if a propensity score is specified, the default is to include the propensity score and the covariates in $\mathbf{x}$ and to optimize balance on the covariates. When `distance = "mahalanobis"` or the `mahvars` argument is specified, the propensity score is left out of $\mathbf{x}$.
r-cran-matchit-4.0.0/vignettes/matching-methods.Rmd-89-
##############################################
r-cran-matchit-4.0.0/vignettes/matching-methods.Rmd-151-
r-cran-matchit-4.0.0/vignettes/matching-methods.Rmd:152:The most common form of matching, 1:1 matching, involves pairing one control unit with each treated unit. To perform $k$:1 matching (e.g., 2:1 or 3:1), which pairs (up to) $k$ control units with each treated unit, the `ratio` argument can be specified. Performing $k$:1 matching can preserve precision by preventing too many control units from being unmatched and dropped from the matched sample, though the gain in precision by increasing $k$ diminishes rapidly after 4 [@rosenbaum2020]. Importantly, for $k>1$, the matches after the first match will generally be worse than the first match in terms of closeness to the treated unit, so increasing $k$ can also worsen balance. Austin [@austin2010a] found that 1:1 or 1:2 matching generally performed best in terms of mean squared error. In general, it makes sense to use higher values of $k$ while ensuring that balance is satisfactory. With nearest neighbor matching, variable $k$:1 matching, in which the number of controls matched to each treated unit varies, can also be used; this can have improved performance over "fixed" $k$:1 matching [@ming2000]. See `?method_nearest` for information on implementing variable $k$:1 matching.
r-cran-matchit-4.0.0/vignettes/matching-methods.Rmd-153-
##############################################
r-cran-matchit-4.0.0/R/match.data.R-12- }
r-cran-matchit-4.0.0/R/match.data.R:13: data <- eval(object$call$data, envir = env)
r-cran-matchit-4.0.0/R/match.data.R-14- if (length(data) == 0) stop("A dataset could not be found. Please supply an argument to 'data' containing the original dataset used in the matching.", call. = FALSE)
##############################################
r-cran-matchit-4.0.0/R/add_s.weights.R-12- }
r-cran-matchit-4.0.0/R/add_s.weights.R:13: data <- eval(m$call$data, envir = env)
r-cran-matchit-4.0.0/R/add_s.weights.R-14- if (length(data) == 0) stop("A dataset could not be found. Please supply an argument to 'data' containing the original dataset used in the matching.", call. = FALSE)
##############################################
r-cran-matchit-4.0.0/inst/doc/assessing-balance.Rmd-117-
r-cran-matchit-4.0.0/inst/doc/assessing-balance.Rmd:118:The third table (`Percent Balance Improvement`) contains the percent balance improvement for each covariate. This is computed as $100\frac{|\theta_M| - |\theta_U|}{|\theta_U|}$, where $\theta_M$ is a given balance statistic in the matched sample and $\theta_U$ is a the same balance statistic in the unmatched sample. Values between 0 and 100 indicate that balance improved after matching as measured by the statistic; values less than 0 indicate that balance got worse after matching. When balance is good on a covariate prior to matching, it can sometimes look like balance got a lot worse after matching even though the balance statistic is quite low, so these values should not be taken too seriously and should be used primarily as heuristics. Setting `un = FALSE` or `improvement = FALSE` would have suppressed the creation of this table.
r-cran-matchit-4.0.0/inst/doc/assessing-balance.Rmd-119-
##############################################
r-cran-matchit-4.0.0/inst/doc/matching-methods.Rmd-87-
r-cran-matchit-4.0.0/inst/doc/matching-methods.Rmd:88:In `MatchIt`, if a propensity score is specified, the default is to include the propensity score and the covariates in $\mathbf{x}$ and to optimize balance on the covariates. When `distance = "mahalanobis"` or the `mahvars` argument is specified, the propensity score is left out of $\mathbf{x}$.
r-cran-matchit-4.0.0/inst/doc/matching-methods.Rmd-89-
##############################################
r-cran-matchit-4.0.0/inst/doc/matching-methods.Rmd-151-
r-cran-matchit-4.0.0/inst/doc/matching-methods.Rmd:152:The most common form of matching, 1:1 matching, involves pairing one control unit with each treated unit. To perform $k$:1 matching (e.g., 2:1 or 3:1), which pairs (up to) $k$ control units with each treated unit, the `ratio` argument can be specified. Performing $k$:1 matching can preserve precision by preventing too many control units from being unmatched and dropped from the matched sample, though the gain in precision by increasing $k$ diminishes rapidly after 4 [@rosenbaum2020]. Importantly, for $k>1$, the matches after the first match will generally be worse than the first match in terms of closeness to the treated unit, so increasing $k$ can also worsen balance. Austin [@austin2010a] found that 1:1 or 1:2 matching generally performed best in terms of mean squared error. In general, it makes sense to use higher values of $k$ while ensuring that balance is satisfactory. With nearest neighbor matching, variable $k$:1 matching, in which the number of controls matched to each treated unit varies, can also be used; this can have improved performance over "fixed" $k$:1 matching [@ming2000]. See `?method_nearest` for information on implementing variable $k$:1 matching.
r-cran-matchit-4.0.0/inst/doc/matching-methods.Rmd-153-