Quantile Regression Combined-Likelihood Compound Estimator for KK Designs (Proportion)

Fits the combined stacked quantile regression (matched-pair differences + reservoir) using the treatment indicator and all recorded covariates for proportion responses. Responses \(y \in (0,1)\) are transformed via \(\mathrm{logit}(y) = \log(y/(1-y))\) before regression; the estimated treatment effect \(\hat\beta_T\) is a log-odds-ratio shift at quantile tau of the logit-transformed response. Minimizes the joint check-function (pinball) loss \(\rho_\tau(u) = u(\tau - \mathbb{1}\{u<0\})\) over both data sources simultaneously in one quantreg fit, unlike the IVWC sibling, which fits matched-pair and reservoir quantile regressions separately and pools them by inverse-variance weighting. Standard errors use Powell's sandwich estimator. likelihood_tier = "none": quantile regression minimizes an asymmetric-loss objective, not a proper likelihood, so no likelihood-ratio or parametric-bootstrap methods are exposed. Requires the quantreg package.

References

Koenker, R. (2005). Quantile Regression. Cambridge University Press. doi:10.1017/CBO9780511754098

Super class

Inference -> InferencePropKKQuantileRegrOneLik

Methods

Public methods

• InferencePropKKQuantileRegrOneLik$new()

• InferencePropKKQuantileRegrOneLik$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$get_treatment()
• Inference$is_nonestimable()
• Inference$set_optimization_alg()
• Inference$set_seed()
• Inference$supports()

---

InferencePropKKQuantileRegrOneLik$new()

Initialize proportion-response KK combined-likelihood quantile-regression inference. Responses are fitted on the logit scale; the shared stacked quantile-regression fit is documented in InferenceContinKKQuantileRegrOneLik (this class's continuous-response sibling, sharing the same KKQuantileRegrOneLik component).

Usage

InferencePropKKQuantileRegrOneLik$new(
  des_obj,
  model_formula = NULL,
  tau = 0.5,
  verbose = FALSE
)

Arguments

des_obj

A DesignSeqOneByOne object whose entire n subjects are assigned and response y is recorded within.

model_formula

Optional formula for covariate adjustment. If NULL (default), the formula from the design object is used and its pre-computed design matrix is reused. If a formula is provided, a new design matrix is constructed from the design's imputed covariates.

tau

The quantile level on the logit scale, strictly between 0 and 1. Default is 0.5.

verbose

Whether to print progress messages.

---

InferencePropKKQuantileRegrOneLik$clone()

The objects of this class are cloneable with this method.

Usage

InferencePropKKQuantileRegrOneLik$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

# \donttest{
seq_des = DesignSeqOneByOneKK14$new(n = 10, response_type = 'proportion')
for (i in 1:10) {
  seq_des$add_one_subject_to_experiment_and_assign(data.frame(x1 = rnorm(1), x2 = rnorm(1)))
}
seq_des$add_all_subject_responses(runif(10))
inf = InferencePropKKQuantileRegrOneLik$new(seq_des)
inf$compute_estimate()
#> [1] 0.6704446
# }