Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    analytics research methods and study planner

    1

    Designing an analytical study requires more than a research question.

    $2.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    analytics research methods and study planner

    analytics research methods and study planner

    Example session with this skill installed

    Plan a study to see if our new UI (Source E1) improved checkout conversion. I have a dataset of 50k users (Source E2) and need to know the sample size for a 2% detectable difference.

    • Read your context and instructions
    • Compiled the analytics research methods

    The research objective and estimand have been processed. The study design is associational based on the provided observational data.

    Study Plan — Checkout Conversion Analysis, 2024-05-20

    1. Objective, estimand and claim boundary

    The study evaluates the association between UI Version (Exposure) and Checkout Completion (Outcome) among 50,000 users over the Q1 period. The design supports an associational claim only; requested causal inferences are downgraded due to the observational nature of Source E2.

    2. Sources, access and assumptions

    SourceUsableExclusion reasonsRecords
    E1YesNone1
    E2YesNone50,000

    Assumptions: User behavior is consistent across the Q1 period (user_supplied).

    3. Measures and operationalization

    MeasureRoleTypeSource.variableStatusAvailability
    UI_TypeExposureCategoricalE2.ui_versionValidatedHigh
    Checkout_SuccessOutcomeBooleanE2.convertedAnalytical estimate — validation requiredHigh

    4. Sampling and feasibility

    ParameterValueOriginStatus
    Alpha0.05DefaultAssumption
    Power0.80DefaultAssumption
    Baseline Rate0.12E2Sourced

    Achievable precision at N=50,000 is 0.008. Minimum detectable difference (MDD) of 0.02 requires a total N of 18,450. The current dataset size is sufficient.

    5. Bias, confounding and leakage assessment

    SeverityAreaCodeFinding
    LowSelectionB01Sampling is restricted to active Q1 users only.
    NoneLeakageL01No temporal leakage detected between UI timestamp and conversion.

    6. Analysis plan and preregistration draft

    Primary method: Logistic regression. Missing-data policy: Listwise deletion for UI_Type. Data-access order: E2 accessed post-approval.

    7. Reproducibility plan

    All data transformations must be scripted. A frozen version of Source E2 is required for audit.

    8. Limitations, information gaps and items for human review

    • Information gap — organizational validation required: Confirmation of UI deployment dates in E1.
    • Management decision required: Approval for accessing sensitive user-level identifiers in E2.
    • Observation: The sample size calculation assumes a balanced split which must be verified in the raw data.

    This plan supports, and does not replace, professional judgment and the decision of the accountable people.

    Next steps

    1. Confirm the baseline conversion rate matches the latest production logs.
    2. Obtain management approval for E2 data access.
    3. Finalize the variable definitions for the 'Checkout_Success' measure.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Designing an analytical study requires more than a research question. It demands precise sample size calculations, bias screening, and traceable operationalization. Vague protocols lead to underpowered studies, data leakage, and rejected preregistrations. This skill converts raw research objectives into a transparent, deterministic study plan. It enforces a strict discipline of evidence, ensuring every measure and parameter is sourced or flagged as an assumption.

    What it does

    • Bounds claim support by comparing requested claims (causal, associational) against the actual design and estimand.
    • Calculates feasibility through exact sample size requirements and achievable precision metrics like margin of error.
    • Screens data access by checking approved purposes, population coverage, and sensitive variable requirements.
    • Evaluates risks by running automated checks for data leakage, confounding, and bias severity.
    • Drafts protocols including analysis plans, preregistration items, and reproducibility checklists.

    How it works

    1. Register sources by assigning E-numbers to all user-supplied materials to ensure traceability.
    2. Validate inputs against the schema, identifying gaps in the research question, estimand, or data sources.
    3. Execute deterministic logic via a specialized Python script to compute statistics and evaluate methodology rules without randomness.
    4. Generate the study plan structured into eight mandatory sections covering objectives through to human review items.

    Frameworks & tools

    The skill utilizes a custom deterministic Python engine for all calculations. It adheres to a strict input/output JSON schema for data integrity and supports French language requests.

    Why this beats prompting it yourself

    Standard LLMs often hallucinate effect sizes or fail to detect subtle data leakage in complex designs. This skill uses a script-driven approach to ensure every decimal is exact and every bias check follows a fixed, verifiable rule set rather than probabilistic guesswork.

    Use cases

    • Drafting a formal protocol for internal research or academic submission.
    • Determining if an existing dataset has the power to detect a specific effect.
    • Screening a proposed study design for data leakage or unsupported causal claims.
    • Preparing a reproducibility checklist and preregistration draft for a new project.

    Known limitations

    Does not perform live data collection or registration. Excludes causal inference identification strategies and A/B test execution. Decisions regarding budget, ethics, and legal compliance require human approval.

    Requires specific inputs like population and estimand. Does not handle live A/B testing or post-collection hypothesis testing. All management and ethics decisions require human approval.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 3 days ago

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Environment Variables
    Listed3 days ago

    What's inside

    Frequently Asked Questions

    Standard LLMs often struggle with precise statistics; this skill uses a deterministic Python engine to ensure all sample sizes and power calculations are mathematically exact rather than probabilistic guesses.

    You will receive the structured skill configuration, the deterministic Python logic for calculations, and a standardized JSON schema to ensure your research inputs are always valid.

    The skill evaluates your design for data leakage, confounding variables, and population coverage gaps, providing a structured risk assessment before you begin data collection.

    Yes, the skill is designed to output comprehensive study plans, reproducibility checklists, and preregistration items that align with formal academic and internal research standards.

    This skill is a design and planning tool; it does not perform live data collection, A/B test execution, or final ethical and legal compliance approvals.