Package 'imt'

Description

This function takes a data frame, identifies columns that are not specified in the exclusion list, and combines them into a new column called 'group'. The original columns used for the combination are then removed. Finally, it returns a data frame with only the 'group', 'variables', 'std_diff', and 'balance' columns.

Usage

.combineColumns(df)

Arguments

Value

A data frame with the combined 'group' column and specified columns.

Description

This function takes a data frame and adds a new column named "treated" with randomly assigned TRUE/FALSE values. Randomization can be done either on the entire data frame or stratified by specified columns. The probability of being assigned to the treatment group can be specified, with a default of 0.5.

Usage

.randomize_internal(data, group_by = NULL, seed = NULL, pr_treated = 0.5)

Arguments

Value

A new data frame with the added "treated" column.

Description

Create a Baseline Balance Plot.

Usage

balancePlot(data)

Arguments

Value

ggplot2 baseline balance plot.

Examples

library(imt)
set.seed(123)  # for reproducibility
N <- 1000
fake_data <- tibble::tibble(x1 = rnorm(N), x2 = rnorm(N), t = rbinom(N, 1, 0.5))
baseline <- checkBaseline(data = fake_data, variables = c("x1", "x2"), treatment = "t")
balancePlot(data = baseline)

Description

Computes the BASIE (BAyeSian Interpretation of Estimates) posterior distribution

Computes the BASIE (BAyeSian Interpretation of Estimates) posterior distribution

Details

Implementation of BASIE

Value

A list containing the mean and standard deviation of the BASIE posterior distribution.

Active bindings

Methods

Public methods

• basie$new()

• basie$vizdraws()

• basie$probability()

• basie$plotProbabilities()

• basie$clone()

Method new()

Estimate the BASIE posterior distribution.

Usage

basie$new(priorMean, priorSD, likMean, likSD)

Arguments

Returns

An object of class Basie.

Method vizdraws()

Plot the prior and BASIE posterior distributions. See vizdraws::vizdraws() for more details.

Usage

basie$vizdraws(draws = 50000L, ...)

Arguments

Method probability()

Estimates the probability that the effect of the intervention is greater than x.

Usage

basie$probability(x)

Arguments

Method plotProbabilities()

Plot the probability that the effect of the intervention is greater than a given threshold.

Usage

basie$plotProbabilities(from = -0.08, to = 0.08, length.out = 100, ...)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

basie$clone(deep = FALSE)

Arguments

Examples

priorMean <- 0
priorSD <- 0.03
likMean <- 0.071
likSD <- 0.074

basie_estimate <- basie$new(
  priorMean = priorMean,
  priorSD = priorSD,
  likMean = likMean,
  likSD = likSD
)


txt <-
  glue::glue(
    "This example assumes that the prior distribution mean is ",
    "{scales::percent(priorMean)}, and its standard deviation ",
    "is {scales::percent(priorSD)}. ",
    "Furthermore, we imagine a study that found ",
    "a point estimate for the effect of {scales::percent(likMean)} ",
    "with a standard error of {scales::percent(likSD)}. ",
    "Finally, we assume that different decisions would be made ",
    "if lift is negative, positive but less than 5%, or greater than 5%."
  )

cat(stringr::str_wrap(txt), "\n")

basie_estimate$plotProbabilities()

Description

Bayesian Linear Model Factory

Bayesian Linear Model Factory

Active bindings

Methods

Public methods

• blm$new()

• blm$ppcDensOverlay()

• blm$tracePlot()

• blm$posteriorProb()

• blm$pointEstimate()

• blm$credibleInterval()

• blm$priorProb()

• blm$priorInterval()

• blm$vizdraws()

• blm$clone()

Method new()

Get the package version

Get the posterior draws for eta

Get the MCMC diagnostics

Get the credible interval

Create a new Bayesian Linear Model object.

Usage

blm$new(
  data,
  y,
  x,
  treatment,
  eta_mean,
  eta_sd,
  generate_fake_data = 0,
  seed = 1982,
  ...
)

Arguments

Returns

invisible

Method ppcDensOverlay()

This method compares the empirical distribution of the data 'y' to the distributions of simulated/replicated data 'yrep' from the posterior predictive distribution. This is done by creating a density overlay plot using the bayesplot::ppc_dens_overlay function.

Usage

blm$ppcDensOverlay(n = 50)

Arguments

Returns

ggplot2 visualization

Method tracePlot()

Plot MCMC trace for the eta and sigma parameters.

Usage

blm$tracePlot(...)

Arguments

Returns

A ggplot object.

Method posteriorProb()

Calculate posterior probability of effect exceeding a threshold

This function calculates the posterior probability of the effect being larger or smaller than a specified threshold.

Usage

blm$posteriorProb(threshold = 0, gt = TRUE)

Arguments

Details

This function uses the private$..eta_draws internal variable to obtain draws from the posterior distribution of the effect size. Based on the specified arguments, the function calculates the proportion of draws exceeding/falling below the threshold and returns a formatted statement describing the estimated probability.

Calculate point estimate of the effect

This R6 method calculates the point estimate of the effect size based on the posterior draws of the eta parameter.

Returns

A character string summarizing the estimated probability.

Method pointEstimate()

Usage

blm$pointEstimate(median = TRUE)

Arguments

Details

This method uses the private$..eta_draws internal variable which contains MCMC draws of the eta parameter representing the effect size. Based on the specified median argument, the method calculates and returns either the median or the mean of the draws. Calculates credible interval for the effect of the intervention

This R6 method calculates and returns a formatted statement summarizing the credible interval of a specified width for the effect of the intervention.

Returns

A numeric value representing the point estimate.

Method credibleInterval()

Usage

blm$credibleInterval(width = 0.75, round = 0)

Arguments

Details

This method uses the private$..eta_draws internal variable containing MCMC draws of the eta parameter representing the effect size. It calculates the credible interval, stores it internally, and returns a formatted statement summarizing the findings.

Calculate and format probability statement based on prior

This method calculates the probability that the effect is greater than or less than a given threshold based on the prior distribution of the effect. The results are formatted into a statement suitable for reporting.

Returns

A character string with the following information:

• The probability associated with the specified width

• The lower and upper bounds of the credible interval, rounded to the specified number of decimal places

Method priorProb()

Usage

blm$priorProb(threshold = 0, gt = TRUE)

Arguments

Returns

A character string containing the formatted probability statement. Calculates the prior for the effedt of the intervention based on the hyperpriors.

This method computes and formats a statement about the probability interval of the effect based on the prior distribution.

Method priorInterval()

Usage

blm$priorInterval(width = 0.75, round = 0)

Arguments

Details

This method calculates the lower and upper bounds of the interval based on the specified probability width and the prior distribution of the effect. It then formats the results into a clear and informative statement.

Note that the probability is checked to be within valid range (0-1) with consideration of machine precision using .Machine$double.eps.

Returns

A character string containing the formatted probability interval statement.

Method vizdraws()

Plots impact's prior and posterior distributions.

For more details see vizdraws::vizdraws().

Usage

blm$vizdraws(...)

Arguments

Returns

An interactive plot of the prior and posterior distributions.

Method clone()

The objects of this class are cloneable with this method.

Usage

blm$clone(deep = FALSE)

Arguments

Description

This function estimates the probability that a vector of posterior draws, represented by the parameter eta, falls within a specified range. It provides flexibility to use either prior distributions or posterior draws, and to specify one-sided or two-sided probability calculations.

Usage

calcProb(x, a = NULL, b = NULL, prior = FALSE, group_name = "group average")

Arguments

Details

This function checks the following cases:

• If both a and b are NULL, it returns an empty string.

• If b is less than or equal to a, it throws an error.

The calculated probability and range are presented in a human-readable string using the glue package for formatting.

Value

A formatted string stating the calculated probability and the specified range. The probability is the proportion of samples (either prior or posterior) that fall within the defined range.

Description

This function calculates the difference-in-differences (DID) estimate of the average treatment effect, along with its standard error. It assumes a simple DID design with two groups (control and treatment) and two time periods (pre and post).

Usage

calculateDIDEffect(
  mean_pre_control,
  sd_pre_control,
  n_pre_control,
  mean_post_control,
  sd_post_control,
  n_post_control,
  mean_pre_treat,
  sd_pre_treat,
  n_pre_treat,
  mean_post_treat,
  sd_post_treat,
  n_post_treat
)

Arguments

Value

A list containing the following components:

Examples

calculateDIDEffect(
  mean_pre_control = 10, sd_pre_control = 2, n_pre_control = 50,
  mean_post_control = 12, sd_post_control = 2.5, n_post_control = 50,
  mean_pre_treat = 11, sd_pre_treat = 2.1, n_pre_treat = 50,
  mean_post_treat = 15, sd_post_treat = 2.6, n_post_treat = 50
)

Description

This function calculates effect sizes for each variable in a data frame,  comparing treatment and control groups. It handles continuous, binary,  and categorical variables using appropriate effect size measures.

Usage

calculateEffectSizes(data, treatment_column, to_check = NULL)

Arguments

Details

- For continuous variables, Hedges' g effect size is calculated.   - For binary variables, Cox's Proportional Hazards Index (Cox's C) is calculated.   - For categorical variables, the variable is converted into multiple indicator     (dummy) variables, and the average Cox's C across these indicators is reported.

Value

A data frame with two columns:   * Variable: The name of each variable in the original data frame.   * EffectSize: The calculated effect size for each variable.

Description

Check Baseline Equivalency.

Usage

checkBaseline(data, variables, treatment)

Arguments

Value

tibble with the standardized difference of the pre-intervention variables. The tibble includes variables: the variable name, std_diff: the standardized difference for that variable as a number balance: the standardized difference for that variable as a factor variable. For more details about this methodology check https://ies.ed.gov/ncee/wwc/Docs/OnlineTraining/wwc_training_m3.pdf.

Description

This function performs a series of data cleaning and preprocessing steps to ensure the data is suitable for analysis. This includes:

• Missing data handling

• Variable type checks

• Collinearity and zero-variance feature removal

Usage

cleanData(data, y, treatment, x = NULL, binary = FALSE)

Arguments

Value

A list containing the cleaned dataset and relevant metadata:

• N: The number of observations after cleaning.

• K The number of covariates after cleaning.

• X The cleaned covariate matrix.

• treat_vec: Treatment vector as integers (1 for TRUE, 0 for FALSE).

• Y: The dependent variable vector.

Description

This function takes a dataframe as input and returns a tibble summarizing the number of missing values (NA) in each column and the number of rows with at least one missing value.

Usage

countMissing(df)

Arguments

Value

A tibble with the following columns:

• NA_<column_name>: Number of NAs in each original column

• missing_rows: Number of rows with at least one NA across all columns

Description

Calculates Cox's C, a standardized effect size measure for comparing hazard rates between two groups in survival analysis.

Usage

coxsIndex(p_t, p_c, n_t, n_c)

Arguments

Details

Cox's C is a useful effect size for survival analysis when hazard ratios are not constant over time. It's calculated based on the log odds ratio of events and includes a small sample size correction. The value 1.65 is used to approximate a conversion to a Cohen's d-like scale.

Value

The calculated Cox's C effect size.

References

Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society: Series B (Methodological), 34(2), 187-202.

Description

Converts a dataframe into a named list to provide data to a Stan model

Usage

createData(
  data,
  y,
  treatment,
  x = NULL,
  eta_mean = 0,
  eta_sd = 1,
  run_estimation = 1
)

Arguments

Value

Returns
stan_data a named list providing the data for the stan model

Description

This function calculates a credible interval of the specified width from a vector of MCMC draws.

Usage

credibleInterval(draws, width)

Arguments

Details

The function calculates the lower and upper bounds of the credible interval using the quantile function based on the specified width.

Value

A named list containing three elements:

• width: The specified width of the credible interval.

• lower_bound: The lower bound of the credible interval.

• upper_bound: The upper bound of the credible interval.

Examples

# Generate example draws
draws <- rnorm(1000)

# Calculate 95% credible interval
credibleInterval(draws, width = 0.95)

Description

Fits Stan model.

Usage

fit(stan_data, model = "blm", ...)

Arguments

Value

The complete StanFit object from the fitted Stan model.

Description

Extracts parameter from Stan model.

Usage

getStanParameter(fit, par)

Arguments

Value

A long tibble with draws for the pamete.

Description

Calculates Hedges' g, a standardized effect size for comparing means. This version includes a small-sample correction factor (omega) and uses the pooled standard deviation.

Usage

hedgesG(n_t, n_c, y_t, y_c, s_t, s_c)

Arguments

Details

Hedges' g is a variation of Cohen's d that adjusts for small-sample bias. It is calculated as the difference in means divided by the pooled standard deviation, then multiplied by a correction factor.

Value

The calculated Hedges' g effect size.

References

Hedges, L. V. (1981). Distribution theory for Glass's estimator of effect size and related estimators. Journal of Educational Statistics, 6(2), 107-128.

Description

A class for creating and managing Bayesian Hurdle Log-Normal Models

Active bindings

Methods

Public methods

• hurdleLogNormal$new()

• hurdleLogNormal$tracePlot()

• hurdleLogNormal$calcProb()

• hurdleLogNormal$pointEstimate()

• hurdleLogNormal$credibleInterval()

• hurdleLogNormal$plotPrior()

• hurdleLogNormal$posteriorPredictiveCheck()

• hurdleLogNormal$clone()

Method new()

Get the package version

Get the posterior draws for ATE

Get the posterior draws for change in zero probability

Get the MCMC diagnostics

Get the credible interval

Get the prior for ATE

Get the prior for change in zero probability

Get the list of predictions

Create a new Bayesian Hurdle Log-Normal Model object.

Usage

hurdleLogNormal$new(
  data,
  y,
  x,
  treatment,
  mean_alpha_logit = -3,
  sd_alpha_logit = 2,
  mean_beta_logit = NULL,
  sd_beta_logit = NULL,
  tau_mean_logit = 0,
  tau_sd_logit = 0.5,
  mean_tau = 0,
  sigma_tau = 0.035,
  seed = 1982,
  fit = TRUE,
  ...
)

Arguments

Returns

invisible

Method tracePlot()

Plot MCMC trace for the ATE and tau_prob_zero parameters

Usage

hurdleLogNormal$tracePlot(...)

Arguments

Returns

A ggplot object

Method calcProb()

Calculates the posterior probability of an effect being greater than, less than, or within a range defined by thresholds.

Usage

hurdleLogNormal$calcProb(effect_type, a = 0, b = NULL, prior = FALSE)

Arguments

Returns

A character string summarizing the estimated probability

Method pointEstimate()

Calculate point estimate of the effect

Usage

hurdleLogNormal$pointEstimate(effect_type, median = TRUE)

Arguments

Returns

A numeric value representing the point estimate.

Method credibleInterval()

Calculates credible interval for the effect of the intervention

Usage

hurdleLogNormal$credibleInterval(effect_type, width = 0.95, round = 2)

Arguments

Returns

A character string with the following information:

• The probability associated with the specified width

• The lower and upper bounds of the credible interval, rounded to the specified number of decimal places

Method plotPrior()

Plot prior distributions for ATE and tau_prob_zero

Usage

hurdleLogNormal$plotPrior(bins = 2000, xlim_ate = NULL, xlim_tau = NULL)

Arguments

Returns

A list containing two ggplot objects: one for ATE and one for tau_prob_zero

Method posteriorPredictiveCheck()

Performs Posterior Predictive Checks and generates a density overlay plot

Usage

hurdleLogNormal$posteriorPredictiveCheck(n = 50, xlim = NULL)

Arguments

Returns

A ggplot object representing the density overlay plot

Method clone()

The objects of this class are cloneable with this method.

Usage

hurdleLogNormal$clone(deep = FALSE)

Arguments

Description

A class for creating and managing Bayesian Logit Models

Active bindings

Methods

Public methods

• logit$new()

• logit$tracePlot()

• logit$calcProb()

• logit$pointEstimate()

• logit$credibleInterval()

• logit$vizdraws()

• logit$lollipop()

• logit$plotPrior()

• logit$predict()

• logit$getPred()

• logit$predSummary()

• logit$predCompare()

• logit$clone()

Method new()

Get the package version

Get the posterior draws for tau

Get the MCMC diagnostics

Get the credible interval

Get the prior for eta

Get the prior for tau

Get the prior for mean y

Get the posterior draws for eta

Get the list of predictions

Create a new Bayesian Logit Model object.

Usage

logit$new(
  data,
  y,
  x,
  treatment,
  mean_alpha,
  sd_alpha,
  mean_beta,
  sd_beta,
  tau_mean,
  tau_sd,
  seed = 1982,
  fit = TRUE,
  ...
)

Arguments

Returns

invisible

Method tracePlot()

Plot MCMC trace for the eta and sigma parameters.

Usage

logit$tracePlot(...)

Arguments

Returns

A ggplot object.

Method calcProb()

Calculates the posterior of an effect being greater than, less than, or within a range defined by thresholds.

Usage

logit$calcProb(a = 0, b = NULL, prior = FALSE)

Arguments

Returns

A character string summarizing the estimated probability

Calculate point estimate of the effect

This R6 method calculates the point estimate of the effect size based on the posterior draws of the eta parameter.

Method pointEstimate()

Usage

logit$pointEstimate(median = TRUE)

Arguments

Details

This method uses the private$..eta_draws internal variable which contains MCMC draws of the eta parameter representing the effect size. Based on the specified median argument, the method calculates and returns either the median or the mean of the draws. Calculates credible interval for the effect of the intervention

This R6 method calculates and returns a formatted statement summarizing the credible interval of a specified width for the effect of the intervention.

Returns

A numeric value representing the point estimate.

Method credibleInterval()

Usage

logit$credibleInterval(width = 0.75, round = 0)

Arguments

Details

This method uses the private$..eta_draws internal variable containing MCMC draws of the eta parameter representing the effect size. It calculates the credible interval, stores it internally, and returns a formatted statement summarizing the findings.

Returns

A character string with the following information:

• The probability associated with the specified width

• The lower and upper bounds of the credible interval, rounded to the specified number of decimal places

Method vizdraws()

Plots impact's prior and posterior distributions.

Usage

logit$vizdraws(tau = FALSE, ...)

Arguments

Returns

An interactive plot of the prior and posterior distributions.

Method lollipop()

Plots lollipop chart for the prior and posterior of the impact being greater or less than a threshold.

For more details see vizdraws::lollipops().

Usage

logit$lollipop(threshold = 0, ...)

Arguments

Returns

A lollipop chart with the prior and posterior probability of the impact being above or below a threshold.

Method plotPrior()

Plots draws from the prior distribution of the outcome, tau, and impact in percentage points.

Usage

logit$plotPrior()

Method predict()

Predict new data

Usage

logit$predict(new_data, name = NULL, M = NULL, ...)

Arguments

Returns

invisible(self)

Method getPred()

Get posterior predictive draws

Usage

logit$getPred(name = NULL, ...)

Arguments

Returns

Matrix of posterior predictive draws

Method predSummary()

Get point estimate, credible interval and prob summary of predictive draws

Usage

logit$predSummary(
  name = NULL,
  subgroup = NULL,
  median = TRUE,
  width = 0.75,
  round = 0,
  a = NULL,
  b = NULL,
  ...
)

Arguments

Returns

A character string with summary information

Method predCompare()

Compare the average of the posterior draws of two groups

Usage

logit$predCompare(
  name1,
  name2,
  subgroup1 = NULL,
  subgroup2 = NULL,
  median = TRUE,
  width = 0.75,
  round = 0,
  a = NULL,
  b = NULL,
  ...
)

Arguments

Returns

A character string with comparison summary

Method clone()

The objects of this class are cloneable with this method.

Usage

logit$clone(deep = FALSE)

Arguments

Description

This function takes prior / posterior draws of the Bayesian logit model, applies theinverse logit (logistic) transformation to obtain probabilities, and then generates random samples from binomial distributions.

Usage

logitRng(alpha, tau, beta, treat, X, N)

Arguments

Value

A vector of N samples of preditive draws

Description

Checks convergence, mixing, effective sample size, and divergent transitions

Methods

$new(fit, pars)

Runs diagnostics on the supplied stanfit object, restricted to parameters identified by the character vector pars.
Tests include:
Share of specified parameters with an Rhat less than 1.1. If any have an Rhat > 1.1, everything_looks_fine is set to FALSE.
Share of specified parameters with an n_eff at least 0.1% of the total number of posterior draws. If any have n_eff < 0.001 * N, everything_looks_fine is set to FALSE.
Share of specified parameters with an n_eff of at least 100. If any have n_eff < 100, everything_looks_fine is set to FALSE.
Number of divergent transitions during posterior sampling. If there are any whatsoever, everything_looks_fine is set to FALSE.
Share of posterior iterations where the sampler reached the maximum treedepth. If more than 25\ everything_looks_fine is set to FALSE.

Active bindings

Methods

Public methods

• mcmcChecks$new()

• mcmcChecks$clone()

Method new()

Initialize a new mcmcChecks object and run diagnostics

Usage

mcmcChecks$new(fit, pars)

Arguments

Method clone()

The objects of this class are cloneable with this method.

Usage

mcmcChecks$clone(deep = FALSE)

Arguments

Description

Create a Meta-Analysis Object Using Data From Previous Studies

Create a Meta-Analysis Object Using Data From Previous Studies

Details

A meta analysis has raw data and draws from the lift's posterior distribution. This is represented by an R6 Class.

Active bindings

Methods

Public methods

• metaAnalysis$new()

• metaAnalysis$PlotRawData()

• metaAnalysis$PlotLift()

• metaAnalysis$UpdateCI()

• metaAnalysis$probability()

• metaAnalysis$findings()

• metaAnalysis$clone()

Method new()

Create a new meta analysis object.

Usage

metaAnalysis$new(
  data,
  point_estimates,
  standard_errors,
  id,
  mean_mu = 0,
  sd_mu = 0.05,
  ci_width = 0.75,
  X = NULL,
  run_estimation = 1,
  ...
)

Arguments

Returns

A new meta_analysis object.

Method PlotRawData()

Plots the raw data.

Usage

metaAnalysis$PlotRawData()

Returns

A plot with point estimates and 95% confidence intervals.

Method PlotLift()

Plots lift's prior and posterior distributions.

For more details see vizdraws::vizdraws().

Usage

metaAnalysis$PlotLift(...)

Arguments

Returns

An interactive plot of the prior and posterior distributions.

Method UpdateCI()

Update the width of the credible interval.

Usage

metaAnalysis$UpdateCI(ci_width)

Arguments

Method probability()

Calculates that probability that lift is between a and b.

Usage

metaAnalysis$probability(a = -Inf, b = Inf, percent = TRUE)

Arguments

Returns

A string with the probability.

Method findings()

Calculates the point estimate a credible interval for the meta analysis.

Usage

metaAnalysis$findings(percent = TRUE)

Arguments

Returns

A string with the findings

Method clone()

The objects of this class are cloneable with this method.

Usage

metaAnalysis$clone(deep = FALSE)

Arguments

Description

Bayesian Negative Binomial Model Factory

Bayesian Negative Binomial Model Factory

Active bindings

Methods

Public methods

• negativeBinomial$new()

• negativeBinomial$tracePlot()

• negativeBinomial$posteriorProb()

• negativeBinomial$pointEstimate()

• negativeBinomial$credibleInterval()

• negativeBinomial$vizdraws()

• negativeBinomial$clone()

Method new()

Create a new Bayesian Negative Binomial Model object.

Usage

negativeBinomial$new(
  data,
  y,
  x,
  treatment,
  tau_mean,
  tau_sd,
  run_estimation = 1,
  seed = 1982,
  ...
)

Arguments

Returns

invisible

Method tracePlot()

Plot MCMC trace for the eta and sigma parameters.

Usage

negativeBinomial$tracePlot(...)

Arguments

Returns

A ggplot object.

Method posteriorProb()

Calculate posterior probability of effect exceeding a threshold

This function calculates the posterior probability of the effect being larger or smaller than a specified threshold.

Usage

negativeBinomial$posteriorProb(threshold = 0, gt = TRUE)

Arguments

Details

This function uses the private$..tau_draws internal variable to obtain draws from the posterior distribution of the effect size. Based on the specified arguments, the function calculates the proportion of draws exceeding/falling below the threshold and returns a formatted statement describing the estimated probability.

Calculate point estimate of the effect

This R6 method calculates the point estimate of the effect size based on the posterior draws of the tau parameter.

Returns

A character string summarizing the estimated probability.

Method pointEstimate()

Usage

negativeBinomial$pointEstimate(median = TRUE)

Arguments

Details

This method uses the private$..tau_draws internal variable which contains MCMC draws of the tau parameter representing the effect size. Based on the specified median argument, the method calculates and returns either the median or the mean of the draws. Calculates credible interval for the effect of the intervention

This R6 method calculates and returns a formatted statement summarizing the credible interval of a specified width for the effect of the intervention.

Returns

A numeric value representing the point estimate.

Method credibleInterval()

Usage

negativeBinomial$credibleInterval(width = 0.75, round = 0)

Arguments

Details

This method uses the private$..tau_draws internal variable containing MCMC draws of the tau parameter representing the effect size. It calculates the credible interval, stores it internally, and returns a formatted statement summarizing the findings.

Returns

A character string with the following information:

• The probability associated with the specified width

• The lower and upper bounds of the credible interval, rounded to the specified number of decimal places

Method vizdraws()

Plots impact's prior and posterior distributions.

For more details see vizdraws::vizdraws().

Usage

negativeBinomial$vizdraws(...)

Arguments

Returns

An interactive plot of the prior and posterior distributions.

Method clone()

The objects of this class are cloneable with this method.

Usage

negativeBinomial$clone(deep = FALSE)

Arguments

Description

This function computes a point estimate from a numeric vector, returning either the median or the mean as a percentage.

Usage

pointEstimate(x, median = TRUE)

Arguments

Details

This function provides a simple way to obtain either the median or mean of a numeric vector as a percentage. The choice between these two measures of central tendency can be controlled by the median argument.

Value

A numeric value representing the chosen point estimate (median or mean) of the input vector x, multiplied by 100 to express it as a percentage.

Description

This function repeatedly randomizes treatment assignment (using .randomizer) until baseline equivalency is achieved across specified variables, as measured by the checkBaseline function from the im package. It can optionally stratify the randomization by specified groups.

Usage

randomize(
  data,
  variables,
  standard = "Not Concerned",
  seed = NULL,
  max_attempts = 100,
  pr_treated = 0.5,
  group_by = NULL
)

Arguments

Value

A new data frame with the added "treated" column, if baseline equivalency is achieved within the specified number of attempts. Otherwise, an error is thrown.

Description

This class provides methods to randomly assign treatments to a dataset while ensuring baseline covariate balance. It can handle both simple and stratified randomization.

Active bindings

Methods

Public methods

• randomizer$new()

• randomizer$clone()

Method new()

Initialize a new Randomizer object.

Usage

randomizer$new(
  data,
  variables,
  standard = "Not Concerned",
  seed = NULL,
  max_attempts = 100,
  group_by = NULL
)

Arguments

Returns

A new randomizer object.

Method clone()

The objects of this class are cloneable with this method.

Usage

randomizer$clone(deep = FALSE)

Arguments

Description

This function checks if a given vector is a logical vector (TRUE/FALSE) and whether its length matches the number of rows in a specified matrix. It is designed to validate subgroup vectors used for subsettin data.

Usage

validate_logical_vector(subgroup, N, name = NULL)

Arguments

Details

This function performs two key validations:

1. Checks if the subgroup vector is logical.

2. Checks if the length of the subgroup vector matches the N.

Value

The original subgroup vector if it passes all validation checks.