Projects per year
Abstract
Probability sample surveys are considered the gold standard for populationbased inference but face many challenges due to decreasing response rates, relatively small sample sizes, and increasing costs. In contrast, the use of nonprobability sample surveys has increased significantly due to their convenience, large sample sizes, and relatively low costs, but they are susceptible to large selection biases and unknown selection mechanisms. Integrating both sample types in a way that exploits their strengths and overcomes their weaknesses is an ongoing area of methodological research. We build on previous work by proposing a method of supplementing probability samples with nonprobability samples to improve analytic inference for logistic regression coefficients and potentially reduce survey costs. Specifically, we consider a Bayesian framework, where inference is based on a probability survey with small sample size and supplementary auxiliary information from a lessexpensive (but potentially biased) nonprobability sample survey fielded in parallel is provided naturally through the prior structure. The performance of several stronglyinformative priors constructed from the nonprobability sample information is evaluated through a simulation study and realdata application. Overall, the proposed priors reduce the meansquared error (MSE) of regression coefficients or, in the worstcase, perform similarly to a weaklyinformative (baseline) prior that doesn’t utilize any nonprobability information. Potential cost savings (of up to 68%) are evident compared to a probabilityonly sampling design with the same MSE for different informative priors under different sample size and cost scenarios. The algorithm, detailed results, and interactive cost analysis are provided through a Shiny web app as guidance for survey practitioners.
Original language  English 

Journal  Journal of Survey Statistics and Methodology 
Publication status  Accepted/In press  3 Oct 2023 
Keywords
 Bayesian Inference
 Data Integration
 Online Access Panel
 Selection Bias
 Web Survey
Fingerprint
Dive into the research topics of 'Bayesian integration of probability and nonprobability samples for logistic regression'. Together they form a unique fingerprint.Projects
 1 Finished

Statistical Modelling
Onah, C. N., Shlomo, N., Chandola, T., Schoch, D., Cuitún Coronado, J., AparicioCastro, A., Thestrup, S., Olsen, W. K., Cernat, A., Shryane, N., Wisniowski, A., Smith, D., AparicioCastro, A., Shafie, T., Hannemann, T., MoralesGómez, A., Mellon, J., Troncoso Ruiz, P., Taub, J., Murphy, J., Guest, E., Kim, J., Watson, J., Pashazadeh, F., Hoór, D., Jones, P. & Gosling, Z.
1/08/19 → 30/09/22
Project: Research
Research output
 2 Article

Integrating Probability and Nonprobability Samples for Survey Inference
Wiśniowski, A., Sakshaug, J., Perez Ruiz, D. & Blom, A. G., 27 Jan 2020, In: Journal of Survey Statistics and Methodology. 8, 1, p. 120–147Research output: Contribution to journal › Article › peerreview
Open Access 
Supplementing Small Probability Samples with Nonprobability Samples: A Bayesian Approach
Sakshaug, J., Wiśniowski, A., Perez Ruiz, D. & Blom, A. G., 9 Sept 2019, (Epub ahead of print) In: Journal of Official Statistics. 35, 3, p. 653681Research output: Contribution to journal › Article › peerreview
Open Access