Estimate lee bounds under attrition

What it's for

Keeps intervention claims honest when one randomized arm yields more observed outcomes, without pretending missingness is ignorable.

What you give it

Inputs split into evidence read from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.

{
  "bootstrap_draws": 200,
  "minimum_practical_effect": 1,
  "observations": [
    {
      "id": "control-0",
      "outcome": 0,
      "selected": true,
      "treatment": 0
    },
    {
      "id": "control-1",
      "outcome": 0,
      "selected": true,
      "treatment": 0
    },
    {
      "id": "control-2",
      "outcome": 0,
      "selected": true,
      "treatment": 0
    },
    {
      "id": "control-3",
      "outcome": 0,
      "selected": true,
      "treatment": 0
    },
    {
      "id": "control-4",
      "outcome": 0,
      "selected": true,
      "treatment": 0
    },
    {
      "id": "control-5",
      "outcome": 0,
      "selected": true,
      "treatment": 0
    },
    {
      "id": "control-6",
      "outcome": 0,
      "selected": true,

What you get back

This is the actual output of running the example above — computed by the same function the platform calls, not an illustration.

{
  "assumptions": [
    "Treatment is randomized, assignment precedes selection, and stable-unit treatment value plus consistency hold for the target outcome.",
    "Selection is monotone at unit level in the declared direction; treatment never moves selection in the opposite direction for any unit.",
    "Lee trimming identifies a bounded effect for always-selected units, not the full randomized population and not an individual-level effect.",
    "Bootstrap uncertainty is conditional on the design and declared monotonicity; severe selection-order instability should trigger more data rather than a causal claim."
  ],
  "decision": "effect_robustly_positive_under_attrition",
  "effect_bounds": {
    "lower": 2,
    "minimum_practical_effect": 1,
    "naive_selected_outcome_difference": 4.6667,
    "uncertainty_outer_interval": {
      "high": 8.3541,
      "low": 2
    },
    "upper": 6,
    "width": 4
  },
  "method": "randomized_lee_monotone_selection_bounds_v1",
  "sample": {
    "bootstrap_attempts": 201,
    "bootstrap_draws": 200,
    "confidence_level": 0.95,
    "control_rows": 80,
    "control_selected": 40,
    "monotonicity_consistency_fraction": 0.995,
    "rows": 160,
    "treated_rows": 80,
    "treated_selected": 60
  },
  "selection": {
    "control_rate": 0.5,
    "monotone_direction": "treatment_weakly_increases_selection",
    "rate_difference": 0.25,
    "treated_rate": 0.75,
    "trim_fraction_in_higher_selection_arm": 0.3333
  }
}

How it works

Causal inference & experiment design — Separate what a change caused from what merely moved alongside it.

1. 1 Partially identify a randomized treatment effect under differential outcome attrition using direction-aware fractional Lee trimming and bootstrap outer bounds.
2. 2 Evaluate the method-specific diagnostics and gates returned by the function, then abstain unless the declared decision clears them under locally governed thresholds.

Before you trust it

Every tool in the catalog ships with the conditions under which its answer is meaningful — and the conditions under which it should abstain instead of guessing.

Calibrating it to your org

Calibration workflow

1. 1 Define the management decision, target outcome, aggregate unit, privacy boundary, cadence, and prediction/intervention horizon for this organization.
2. 2 Build a tenant-scoped historical cohort using only information available before each prediction or decision; preserve zero periods, censoring, assignment probabilities, and unresolved outcomes when the method requires them.
3. 3 Estimate statistical parameters on training history, but obtain costs, utilities, risk tolerance, practical-effect thresholds, capacity, and policy constraints from accountable decision owners.
4. 4 Validate on later time windows or held-out aggregate units at the deployment grain, against a simple baseline and the function-specific validation strategy.
5. 5 Deploy only if the returned decision clears evidence, overlap, calibration, robustness, and guardrail checks; warning, unsupported, schema-gap, and fallback decisions are abstentions.
6. 6 Monitor realized outcomes, data drift, coverage, and decision regret; recalibrate at a governed cadence or after a detected regime/definition change, never merely because a stakeholder dislikes the result.

Call it from your AI

You don't wire up 388 tools in your MCP client. The GitRevio MCP server exposes 18 tools, three of which let an agent search the catalog, read a tool's schema, and run it — so the assistant finds this one on its own.

gitrevio_capabilities_search
  { "q": "partially identify a randomized treatment effect" }
  → finds "estimate_lee_bounds_under_attrition"

gitrevio_capability_describe
  { "capability_id": "estimate_lee_bounds_under_attrition" }
  → returns the input schema and agent guidance shown on this page

gitrevio_capability_run
  { "capability_id": "estimate_lee_bounds_under_attrition", "arguments": { ... } }
  → returns the result shown above

Works in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.

Related tools

See every tool in Causal evidence & experiments →