agentleFS
TrecekTrecek/useful-claude-skills/.claude/skills/exp-lens-validity-threats/SKILL.md

exp-lens-validity-threats

Skill1 starsChanged 8 months agoMITView on GitHub

Create a validity threat matrix identifying alternative explanations and design mitigations. Adversarial lens answering "What alternative explanations survive?

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 8 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not litter the codebase with useless comments, TODO markers, or explanatory annotations — the skill output and diagram speak for …

    line 30
  2. Very specific:

    BEFORE creating any diagram, LOAD the /mermaid skill using the Skill tool - this is MANDATORY

    line 37
  3. Distinctive:

    Look for: version, update, release, change, drift, before, after, time, date

    line 49
  4. Distinctive:

    Look for: log, metric, instrument, measurement, format, schema, API, breaking

    line 53
  5. Distinctive:

    Look for: filter, exclude, select, subsample, eligible, criteria, dropout, attrition

    line 57
  6. Distinctive:

    Look for: also, simultaneously, together, combined, alongside, additionally, migration

    line 61
  7. Distinctive:

    Look for: access, share, copy, learn, adopt, contaminate, crossover, spillover

    line 65
  8. Distinctive:

    Use the mermaid skill conventions to create a threat-flow diagram with:

    line 85
  9. Very specific:

    Write the diagram to: temp/exp-lens-validity-threats/exp_diag_validity_threats_{YYYY-MM-DD_HHMMSS}.md

    line 103
  10. Distinctive:

    Before creating the diagram, verify:

    line 201

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Validity Threats Experimental Design Lens

Philosophical Mode: Adversarial Primary Question: "What alternative explanations survive?" Focus: History Effects, Instrumentation Changes, Selection Effects, Co-interventions, Treatment Diffusion

When to Use

  • Quasi-experimental designs without full randomization
  • System experiments with environmental changes
  • Longitudinal comparisons
  • User invokes /exp-lens-validity-threats or /make-experiment-diag validity

Critical Constraints

NEVER:

  • Modify any source code files
  • Do not litter the codebase with useless comments, TODO markers, or explanatory annotations — the skill output and diagram speak for themselves

ALWAYS:

  • Apply Campbell & Stanley's full threat taxonomy — do not skip threats just because they seem unlikely
  • For every observed difference, enumerate at least 3 alternative explanations
  • Assess mitigation strength honestly — "partially mitigated" is better than false confidence
  • Distinguish threats that are ruled out by design from those that remain plausible
  • BEFORE creating any diagram, LOAD the /mermaid skill using the Skill tool - this is MANDATORY

Analysis Workflow

Step 1: Launch Parallel Exploration Subagents

Spawn Explore subagents to investigate:

Temporal Changes (History)

  • Find whether the environment changed between conditions
  • Look for: version, update, release, change, drift, before, after, time, date

Instrumentation Changes

  • Find whether measurement tools or methods changed
  • Look for: log, metric, instrument, measurement, format, schema, API, breaking

Selection & Filtering

  • Find whether groups are compared after differential filtering
  • Look for: filter, exclude, select, subsample, eligible, criteria, dropout, attrition

Co-interventions

  • Find whether multiple changes occurred simultaneously
  • Look for: also, simultaneously, together, combined, alongside, additionally, migration

Treatment Diffusion

  • Find whether control units could access treatment
  • Look for: access, share, copy, learn, adopt, contaminate, crossover, spillover

Step 2: Apply Threat Taxonomy

Apply Campbell & Stanley's threat taxonomy to the specific experiment. For each threat category:

  1. Is it plausible in this context?
  2. What design feature mitigates it?
  3. How strong is the mitigation?

Build the threat matrix.

Step 3: Analyze Alternative Explanations

CRITICAL — Analyze Alternative Explanations: For every observed difference between conditions:

  • List at least 3 alternative explanations besides the intended causal claim
  • For each: What evidence rules it out? What evidence remains consistent with it?

Step 4: Create the Diagram

Use the mermaid skill conventions to create a threat-flow diagram with:

Direction: TB (threats flow from sources through mitigations to residual risk)

Subgraphs:

  • THREAT SOURCES
  • DESIGN MITIGATIONS
  • RESIDUAL THREATS

Node Styling:

  • gap class: Active unmitigated threats
  • detector class: Design mitigations
  • stateNode class: Threat sources
  • handler class: Partially mitigated threats
  • output class: Successfully mitigated threats

Step 5: Write Output

Write the diagram to: temp/exp-lens-validity-threats/exp_diag_validity_threats_{YYYY-MM-DD_HHMMSS}.md


Output Template

# Validity Threat Analysis: {Experiment Name}

**Lens:** Validity Threats (Adversarial)
**Question:** What alternative explanations survive?
**Date:** {YYYY-MM-DD}
**Scope:** {What was analyzed}

## Threat Matrix

| Threat | Category | Plausibility | Mitigation | Residual Risk |
|--------|----------|-------------|------------|---------------|
| {threat} | {Campbell-Stanley category} | {Low/Medium/High} | {design feature} | {Low/Medium/High} |

## Alternative Explanations

| Observed Difference | Alternative Explanation | Ruling-Out Evidence | Consistent Evidence |
|--------------------|------------------------|---------------------|---------------------|
| {difference} | {explanation} | {evidence} | {evidence} |

## Threat-Flow Diagram

```mermaid
%%{init: {'flowchart': {'nodeSpacing': 50, 'rankSpacing': 60, 'curve': 'basis'}}}%%
graph TB
    %% CLASS DEFINITIONS %%
    classDef cli fill:#1a237e,stroke:#7986cb,stroke-width:2px,color:#fff;
    classDef stateNode fill:#004d40,stroke:#4db6ac,stroke-width:2px,color:#fff;
    classDef handler fill:#e65100,stroke:#ffb74d,stroke-width:2px,color:#fff;
    classDef phase fill:#6a1b9a,stroke:#ba68c8,stroke-width:2px,color:#fff;
    classDef newComponent fill:#2e7d32,stroke:#81c784,stroke-width:2px,color:#fff;
    classDef output fill:#00695c,stroke:#4db6ac,stroke-width:2px,color:#fff;
    classDef detector fill:#b71c1c,stroke:#ef5350,stroke-width:2px,color:#fff;
    classDef gap fill:#ff6f00,stroke:#ffa726,stroke-width:2px,color:#000;
    classDef integration fill:#c62828,stroke:#ef9a9a,stroke-width:2px,color:#fff;

    subgraph Sources ["THREAT SOURCES"]
        direction TB
        HIST["History Effect<br/>━━━━━━━━━━<br/>Environmental change<br/>between conditions"]
        INST["Instrumentation<br/>━━━━━━━━━━<br/>Measurement tool<br/>change"]
        SEL["Selection Effect<br/>━━━━━━━━━━<br/>Differential filtering<br/>of units"]
    end

    subgraph Mitigations ["DESIGN MITIGATIONS"]
        direction TB
        RAND["Randomization<br/>━━━━━━━━━━<br/>Controls selection<br/>and history"]
        CTRL["Control Group<br/>━━━━━━━━━━<br/>Absorbs environmental<br/>changes"]
    end

    subgraph Residual ["RESIDUAL THREATS"]
        direction TB
        DIFF["Treatment Diffusion<br/>━━━━━━━━━━<br/>Control access<br/>to treatment"]
        COINT["Co-interventions<br/>━━━━━━━━━━<br/>Simultaneous<br/>changes"]
    end

    HIST --> CTRL
    INST --> CTRL
    SEL --> RAND
    RAND --> CTRL
    CTRL --> DIFF
    CTRL --> COINT

    %% CLASS ASSIGNMENTS %%
    class HIST,INST,SEL stateNode;
    class RAND,CTRL detector;
    class DIFF,COINT gap;

Campbell-Stanley Checklist

ThreatApplicable?Assessment
History{Yes/No/Partial}{assessment}
Maturation{Yes/No/Partial}{assessment}
Testing{Yes/No/Partial}{assessment}
Instrumentation{Yes/No/Partial}{assessment}
Statistical Regression{Yes/No/Partial}{assessment}
Selection{Yes/No/Partial}{assessment}
Experimental Mortality{Yes/No/Partial}{assessment}
Selection-Maturation Interaction{Yes/No/Partial}{assessment}
Diffusion of Treatments{Yes/No/Partial}{assessment}
Co-interventions{Yes/No/Partial}{assessment}

Key Findings

  • {Description of most critical surviving alternative explanations}

---

## Pre-Diagram Checklist

Before creating the diagram, verify:

- [ ] LOADED `/mermaid` skill using the Skill tool
- [ ] Using ONLY classDef styles from the mermaid skill (no invented colors)
- [ ] Diagram will include a color legend table

---

## Related Skills

- `/make-experiment-diag` - Parent skill for lens selection
- `/mermaid` - MUST BE LOADED before creating diagram
- `/exp-lens-causal-assumptions` - For DAG-based causal structure analysis
- `/exp-lens-severity-testing` - For statistical severity and power analysis

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.