Collaborative Research Team Project #05

Statistical Inference for Complex Surveys with Missing Observations

This project explores methods for analyzing high-dimensional data sets with missing values.

Research Category: Information Sciences
Region: National
Date: 2015-2018

Why Study Complex Surveys?

Complex surveys play an important role in providing information for science and society. For the efficient use of this information, survey data must be reliable and representative.

For the past three decades, one of the focal points in survey research has been how to handle missing data. Imputation is an increasingly used technique. However, it is notoriously difficult to impute multivariate data having an arbitrary missing pattern if the covariance structure must be preserved. Addressing this problem will be a major part of the planned research.

Areas of Exploration

Complex Survey Data

Includes collaborating with researchers at Statistics Canada and Westat on problems in large-scale, high-dimensional survey data with missing values.

Techniques for Incomplete Data

Includes extending the techniques of fractional imputation and doubly robust methods to dealing with missing values in high dimensional data.

Statistical Inference & Applications

Includes exploring the relatively new area of developing inference from incomplete functional survey data, with application to large functional data sets, such as electricity consumption data.

Solving Global Challenges

Research Team’s Goal

To carry out research and training in high-dimensional survey data analysis, with a focus on challenges in data collection resulting from non-response and missing values.

Related Events

Spring School on Statistical Inference for Survey Data with Missing Observations | June 6-9, 2017 at the Fields Institute
Learn More

Statistical Inference for Complex Surveys with Missing Observations Seminar Series | 2015-2016

Seminars include:

Surveys from large datasets of functional data and estimation of the mean and median curve with full and missing data
Small Area Quantile Estimation
Validity and Efficiency in Analyzing Ordinal Responses with Missing Observations
Inferential Issues in the Presence of Imputation for Missing Survey Data
An Application of Small Area Estimation to the Labour Force Survey Data
Some Recent Topics on Informative Sampling
Empirical Likelihood Inference with Public-Use Survey Data

People Behind the Project

Project Team

David Haziza | Université de Montréal

Collaborators

Jean-François Beaumont | Statistics Canada

Michael Brick | Westat (Rockville, USA)

Hervé Cardot | Université de Bourgogne

Camelia Goga | Université de Bourgogne

Jiahua Chen | University of British Columbia

Jae-Kwang Kim | Iowa State University

Wilson Lu | Acadia University

Changbao Wu | University of Waterloo

Relevant Publications

Chen, S and Haziza, D. (2015). Multiply robust imputation procedures for the treatment of item nonresponse in surveys. Under revision for Biometrika.
HAZIZA, D. & BEAUMONT, J.F. (2017). Construction of weights in surveys: a review. To appear in Statistical Science.
CHEN, S. & HAZIZA, D. (2017). Multiply robust imputation procedures for the treatment of item nonresponse in surveys. To appear in Biometrika.
SHE, X. and WU, C. (2015). Analysis of Ordinal Survey Responses with Missing Values. Proceedings of the Survey Methods Section of SSC, Halifax, 1-8.
SHE, X. (2016). Fully Efficient Joint Fractional Imputation for Incomplete Bivariate Ordinal Responses. Proceedings of the Survey Methods Section of SSC, St. Catherines, 1-10.
SHE, X. and WU, C. (2016). Validity and Efficiency in Analyzing Ordinal Responses with Missing Observations. Under revision for Biometrika.
SHE, X. (2016). Fully Efficient Joint Fractional Imputation for Incomplete Bivariate Ordinal Responses with Missing Observations. Under review by Statistica Sinica.
SHE, X. (2016). Fractional Imputation for Ordinal and Mixed-type Responses with Missing Observations. Doctoral Dissertation, Department of Statistics and Actuarial Science, University of Waterloo.
RAO, J.N.K. and WU, C. (2016). Empirical Likelihood Inference with Public-Use Survey Data. Under revision for Biometrika.
ZHAO, P., HAZIZA, D. & WU, C. (2016). Empirical Likelihood Inference for Complex Surveys and the Design-based Oracle Variable Selection Theory. Submitted to the Annals of Statistics.
CHEN, S. & HAZIZA, D. (2017). Multiply robust nonparametric multiple imputation for the treatment of missing data. Submitted to Biometrika.
LEFEBVRE, I. (2016) Estimation simplifiée de la variance pour des plans complexes. Master’s thesis, Department of mathematics and statistics, Université de Montréal.
CARDOT, H., DE MOLINER, A., and GOGA, C. (2017). Estimating a mean electricity consumption curve in the presence of missing data using survey sampling techniques. To be submitted to the special issue of the Canadian Journal of Statistics.
CHEN, S. & HAZIZA, D. (2017). Multiply robust imputation procedures for zero-inflated distributions in surveys. Invited paper for a special issue of Metron.
ZHAO, P., GHOSH, M., RAO, J.N.K. and WU, C. (2017). Bayesian Empirical Likelihood Inference With Complex Survey Data. Invited for a re-submission by Journal of the Royal Statistical Society B.
CARDOT, H., DE MOLINER, A., and GOGA, C. (2017). Estimating a mean electricity consumption curve in the presence of missing data using survey sampling techniques. In revision for the special issue of the Canadian Journal of Statistics.
LEFEBVRE, I. (2016) Estimation simplifiée de la variance pour des plans complexes. Master’s thesis, Department of mathematics and statistics, Université de Montréal

Contact

Statistical Inference for Complex Surveys with Missing Observations is a Collaborative Research Team project. This program tackles complex problems through a three-year research and training agenda.

CANSSI offers approximately $200,000 for this type of project, which requires a team of faculty, postdocs, and students.

Learn more