Bai Adjusted-t Mean-Difference Inference for KK21 Designs

Continuous-response mean-difference inference for designs assigned by DesignSeqOneByOneKK21 (the Kapelner-Krieger 2021 sequential matching-on-the-fly design with covariate-weighted matching). The point estimate and its variance are the closed-form Bai-adjusted-t combination of the matched-pairs mean difference and the unmatched-reservoir mean difference, inverse-variance-weighted when both are usable; see InferenceBaiAdjustedTKK14 for the full formula, the pair-distance definition, and the confidence-interval/p-value construction shared with this class. The two leaves differ only in how pair distance is defined during matching: this class (KK21) uses the design's covariate-weighted squared distance \(\sum_j w_j (x_{1j} - x_{2j})^2\), where \(w_j\) are the design's covariate_weights (see DesignSeqOneByOneKK21), unlike KK14's unweighted distance. Because the estimator is closed-form, initialization does not use warm starts (there is no iterative fit to warm-start).

Details

Legacy class. Not fully tested in comprehensive_tests.R.

Super class

Inference -> InferenceBaiAdjustedTKK21

Methods

Public methods

• InferenceBaiAdjustedTKK21$set_custom_randomization_statistic_function()

• InferenceBaiAdjustedTKK21$set_custom_randomization_statistic_cpp()

• InferenceBaiAdjustedTKK21$approximate_randomization_distribution_beta_hat_T()

• InferenceBaiAdjustedTKK21$supports_rand_pval_for_incidence()

• InferenceBaiAdjustedTKK21$compute_rand_two_sided_pval()

• InferenceBaiAdjustedTKK21$clone()

+ inherited public methods from Inference

• Inference$capabilities()
• Inference$compute_asymp_confidence_interval()
• Inference$compute_asymp_two_sided_pval()
• Inference$compute_estimate()
• Inference$compute_exact_confidence_interval()
• Inference$compute_exact_two_sided_pval_for_treatment_effect()
• Inference$duplicate()
• Inference$get_analysis_data()
• Inference$get_covariates()
• Inference$get_design_object()
• Inference$get_model_formula()
• Inference$get_nonestimable_reason()
• Inference$get_nonestimable_stage()
• Inference$get_optimization_alg()
• Inference$get_response()
• Inference$get_response_type()
• Inference$get_treatment()
• Inference$initialize()
• Inference$is_nonestimable()
• Inference$set_optimization_alg()
• Inference$set_seed()
• Inference$supports()

---

InferenceBaiAdjustedTKK21$set_custom_randomization_statistic_function()

Usage

InferenceBaiAdjustedTKK21$set_custom_randomization_statistic_function(
  custom_randomization_statistic_function
)

---

InferenceBaiAdjustedTKK21$set_custom_randomization_statistic_cpp()

Usage

InferenceBaiAdjustedTKK21$set_custom_randomization_statistic_cpp(fn)

---

InferenceBaiAdjustedTKK21$approximate_randomization_distribution_beta_hat_T()

Usage

InferenceBaiAdjustedTKK21$approximate_randomization_distribution_beta_hat_T(
  r = 501,
  delta = 0,
  transform_responses = "none",
  show_progress = TRUE,
  permutations = NULL,
  debug = FALSE,
  zero_one_logit_clamp = .Machine$double.eps
)

---

InferenceBaiAdjustedTKK21$supports_rand_pval_for_incidence()

Usage

InferenceBaiAdjustedTKK21$supports_rand_pval_for_incidence()

---

InferenceBaiAdjustedTKK21$compute_rand_two_sided_pval()

Usage

InferenceBaiAdjustedTKK21$compute_rand_two_sided_pval(
  r = 501,
  delta = 0,
  transform_responses = "none",
  na.rm = TRUE,
  show_progress = TRUE,
  permutations = NULL,
  zero_one_logit_clamp = .Machine$double.eps
)

---

InferenceBaiAdjustedTKK21$clone()

The objects of this class are cloneable with this method.

Usage

InferenceBaiAdjustedTKK21$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

# \donttest{
seq_des = DesignSeqOneByOneKK14$new(n = 10, response_type = 'continuous')
for (i in 1:10) {
  seq_des$add_one_subject_to_experiment_and_assign(data.frame(x1 = rnorm(1), x2 = rnorm(1)))
}
seq_des$add_all_subject_responses(rnorm(10))
inf = InferenceBaiAdjustedTKK21$new(seq_des)
inf$compute_estimate()
#> [1] -0.7900056
# }