Browse categories
Explore
Fiverr Pro
English
$
USD
I will design and analyze ab tests and causal inference studies
Lead Data Scientist
I am a Lead Data Scientist with over 8 years of experience architecting production-grade machine learning systems and advanced causal inference models. I excel at framing complex problems and directin...
About this Gig
The Premium package covers one clearly defined causal question and one primary estimand using one analysis-ready dataset. The treatment may be binary, multilevel, or continuous. The package includes method selection, diagnostics, one sensitivity analysis, documented results, an executive presentation, one revision, and up to 20 hours of work. Multiple unrelated causal questions, datasets, primary outcomes, or substantially different analytical strategies require a custom offer.
FAQ
What types of projects can you help with?
I design and evaluate A/B tests, experiments, observational causal studies, and impact evaluations. I can help define the causal question, select an appropriate method, conduct the analysis, assess assumptions, and translate the findings into practical recommendations.
Which package should I select?
Choose Basic for expert review of an experiment or causal-analysis plan, Standard for designing or analyzing one A/B test, and Premium for an end-to-end causal-inference analysis. Contact me before ordering if your project involves multiple datasets, outcomes, or research questions.
Should I contact you before placing an order?
Yes, please contact me before ordering Standard or Premium. I will review your objective, dataset, expected deliverables, and timeline to confirm that the selected package fits your project.
What does analysis-ready data mean?
Analysis-ready data are structured, documented, and contain the treatment, outcome, confounder, time, and identifier variables needed for the analysis. The dataset should not have major unresolved quality, formatting, or missing-data problems.
Is data cleaning included?
Standard data validation and limited preparation are included. Extensive cleaning, database integration, record linkage, feature engineering, or combining multiple data sources requires a custom offer.
Can the treatment have multiple groups or be continuous?
Yes. The treatment can be binary, multilevel, or continuous, including A/B/n experiments, dosage levels, exposure intensity, or duration. The standard scope covers one primary causal question and one primary outcome.
Which causal-inference methods do you use?
Depending on the study design and assumptions, I may use propensity-score methods, inverse-probability weighting, generalized propensity scores, doubly robust estimation, difference-in-differences, instrumental variables, regression adjustment, or machine-learning-assisted causal estimation.
Do you guarantee a statistically significant result?
No ethical analyst can guarantee statistical significance or a preferred conclusion. I guarantee a rigorous, transparent analysis, appropriate diagnostics, clearly stated assumptions and limitations, and an objective interpretation of the evidence.
What is included in one revision?
One revision covers reasonable corrections or clarifications based on the originally approved question, dataset, and methodology. New data, outcomes, specifications, or research questions constitute additional work.
How are larger or more complex projects handled?
Projects involving multiple causal questions, datasets, primary outcomes, extensive preprocessing, alternative analytical strategies, or more than 20 hours of work will receive a custom offer with clearly defined deliverables, pricing, and milestones.
