Senior Statisticians – Synthetic Survey Data

One of our crown corporation clients is launching the first phase of the Synthetic Survey Data initiative to address declining business-to-business survey response rates, longer fieldwork cycles, and emerging data-quality issues.

The immediate mandate is to design, build, test, and validate a controlled statistical approach that enhances existing survey data using population and survey information. The initial phase will retain real survey responses as the evidence base and will focus on statistical augmentation rather than fully synthetic, persona-based responses. The consultants will work with the Research and Market Intelligence team, the initiative owner, the project manager, Data and Analytics, Architecture, IT, and other control functions as required. The target is to complete and validate the first statistical approach by the end of March 2027.

Level of contingent worker: Senior > 10 years
Estimated start date of the mandate: As soon as possible
Duration of the mandate: 12 months
Option to renew: Yes
Work hours: Full-time
Hybrid work: Services are to be provided 3 days a week in person and 2 days a week remotely

Responsibilities / Accountabilities

Define the statistical methodology for augmenting existing business survey data using population-level and survey datasets.
Assess candidate data sources, including historical survey data, relevant Canadian SME population data, economic indicators, and approved external reference data.
Profile data quality, coverage, representativeness, missingness, bias, and fitness for use. Document constraints and remediation needs.
Design and implement one or more transparent statistical models for imputation and synthetic record generation. Bayesian methods may be used where appropriate and explainable.
Develop the initial models and reproducible workflow in the approved Databricks environment.
Define the validation framework and acceptance criteria, including reliability, repeatability, bias, variance, subgroup accuracy, explainability, and business usefulness.
Compare model outputs with real survey results and document where the approach is reliable, where it is not, and which use cases are acceptable or unacceptable.
Pilot the approach on a selected survey use case and assess whether it can improve effective sample depth, support subgroup analysis, and reduce pressure on fieldwork.
Prepare clear technical documentation, model assumptions, data lineage, validation results, limitations, and recommendations for governance and future operation.
Work closely with Research and Market Intelligence subject-matter experts and collaborate with Data, Architecture, IT, Privacy, Security, and Risk stakeholders as required.
Provide regular progress updates, identify blockers early, and transfer knowledge to the client's staff through working sessions and documented methods.
Contribute to a final recommendation to scale, refine, pause, or limit the capability after the pilot.

Requirements

Minimum seven years of progressive experience in applied statistics, survey statistics, quantitative research, econometrics, biostatistics, or a closely related field.
Demonstrated experience designing and validating methods for survey weighting, imputation, small-area or subgroup estimation, sample augmentation, missing-data treatment, or related statistical techniques.
Experience generating synthetic or model-based survey data (e.g., Bayesian networks, sequential/CART-based synthesis, population synthesis, or statistical matching) is an asset.
Demonstrated experience working with complex survey data and population or administrative reference datasets.
Experience evaluating model bias, uncertainty, variance, generalizability, and subgroup performance.
Experience delivering statistical models in a governed enterprise, public-sector, financial-services, or similarly controlled environment is an asset.
Experience with business-to-business, small-business, economic, or market-research data is an asset.

Educational Experience

Master's degree in statistics, survey methodology, biostatistics, econometrics, mathematics, quantitative social science, or a closely related discipline.
Doctorate in a relevant quantitative discipline is an asset.
Equivalent advanced training and directly relevant experience may be considered.

Knowledge and Skills

Advanced knowledge of survey sampling, weighting, calibration, imputation, missing-data methods, uncertainty estimation, and statistical validation.
Strong knowledge of Bayesian statistics is an asset, including hierarchical modelling and clear communication of priors, assumptions, and uncertainty.
Familiarity with machine learning methods for prediction and imputation (e.g., tree-based models, classifiers) is an asset.
Strong programming skills in Python or R for statistical analysis and reproducible modelling.
Experience using Databricks or an equivalent governed analytics environment. Ability to work with notebooks, Git-based version control (e.g., GitHub), and structured data pipelines.
Working knowledge of SQL and the preparation, joining, profiling, and validation of large datasets.
Ability to design comparison tests between real and statistically augmented survey outputs, including subgroup and sensitivity analyses.
Ability to produce explainable methods, technical documentation, validation evidence, and practical guidance for non-technical research stakeholders.
Knowledge of privacy, responsible AI, research ethics, data governance, and appropriate use of synthetic data.

Soft Skills

Intellectual rigor and disciplined handling of assumptions, uncertainty, and limitations.
Clear written and verbal communication with both technical and business audiences.
Collaborative working style and ability to work with researchers, data specialists, architects, project managers, and control functions.
Results orientation, autonomy, sound judgement, and ability to progress work in an evolving environment.
Ability to challenge weak methods or unsupported conclusions constructively and propose practical alternatives.
Strong documentation and knowledge-transfer discipline.

Language

Professional proficiency in English, spoken and written, is required. Professional proficiency in French is a strong asset.