Issues

• setup_simulation_OLE() does not check T_cross against the number of outcomes

  Genentech/rdborrow#141 · Filter by author mattsecrest opened on Oct 6, 2026

• scm() has no shared core function, so its bootstrap statistic duplicates the estimator

  Genentech/rdborrow#132 · Filter by author mattsecrest opened on Oct 6, 2026

• Tests for the OLE methods use one visit split, equal arms, and alpha = 0.05 only

  Genentech/rdborrow#131 · Filter by author mattsecrest opened on Oct 6, 2026

• Tests for ec_ipw() and ec_aipw() lock snapshot values and use only balanced data

  Genentech/rdborrow#130 · Filter by author mattsecrest opened on Oct 6, 2026

• Small documentation errors and gaps (checklist)

  Genentech/rdborrow#129 · Filter by author mattsecrest opened on Oct 6, 2026

• The OLE vignette shows scm() confidence intervals from 3 bootstrap replicates and hides the warning

  Genentech/rdborrow#128 · Filter by author mattsecrest opened on Oct 6, 2026

• Documentation calls weight = NULL "data-adaptive" and "minimizing variance", but the weight does not use outcomes

  Genentech/rdborrow#127 · Filter by author mattsecrest opened on Oct 6, 2026

• Code comments cite the paper's Eq 13 and Eq 16, but the code correctly uses a different variance formula

  Genentech/rdborrow#126 · Filter by author mattsecrest opened on Oct 6, 2026

• Small input-validation gaps (checklist)

  Genentech/rdborrow#125 · Filter by author mattsecrest opened on Oct 6, 2026

• scm() lambda arguments are accepted but ignored or fail late, and the selected lambda is not reported

  Genentech/rdborrow#124 · Filter by author mattsecrest opened on Oct 6, 2026

• bootstrap_ci_type without bootstrap is ignored with no message

  Genentech/rdborrow#123 · Filter by author mattsecrest opened on Oct 6, 2026

• Methods do not reject data with an empty group

  Genentech/rdborrow#122 · Filter by author mattsecrest opened on Oct 6, 2026

• alpha accepts 0 and 1, which give infinite, missing, or zero-width intervals

  Genentech/rdborrow#121 · Filter by author mattsecrest opened on Oct 6, 2026

• The ec_aipw() sandwich variance uses a different spline basis than the point estimate

  Genentech/rdborrow#120 · Filter by author mattsecrest opened on Oct 6, 2026

• The ec_aipw() sandwich variance fails when a covariate has large values

  Genentech/rdborrow#119 · Filter by author mattsecrest opened on Oct 6, 2026

• Collinear covariates in the participation model stop the sandwich variance with a LAPACK error

  Genentech/rdborrow#118 · Filter by author mattsecrest opened on Oct 6, 2026

• The sandwich variance uses memory that increases with the square of the sample size

  Genentech/rdborrow#117 · Filter by author mattsecrest opened on Oct 6, 2026

• scm() fails with an unclear error when a covariate is a factor

  Genentech/rdborrow#116 · Filter by author mattsecrest opened on Oct 6, 2026

• scm() fails with two or fewer external controls

  Genentech/rdborrow#115 · Filter by author mattsecrest opened on Oct 6, 2026

• Methods accept missing values and give NA or incorrect results

  Genentech/rdborrow#114 · Filter by author mattsecrest opened on Oct 6, 2026

• scm() estimates depend on the units of the covariates

  Genentech/rdborrow#113 · Filter by author mattsecrest opened on Oct 6, 2026

• scm() does not check the solver status

  Genentech/rdborrow#112 · Filter by author mattsecrest opened on Oct 6, 2026

• A factor treatment column passes the 0/1 check, and trt_formula models the wrong level

  Genentech/rdborrow#111 · Filter by author mattsecrest opened on Oct 6, 2026

• Setup functions accept external controls with treatment 1

  Genentech/rdborrow#108 · Filter by author mattsecrest opened on Oct 6, 2026
  1

• Formula variables that are not in the data come from the user's R session

  Genentech/rdborrow#107 · Filter by author mattsecrest opened on Oct 6, 2026