Private Posterior Inference Consistent with Public Information: A Case Study in Small Area Estimation from Synthetic Census Data

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 12276)


Methods for generating differentially-private (DP) synthetic data have received recent attention as large government agencies such as the U.S. Census have decided to release DP synthetic data for public usage. In the synthetic data generation process, it is common to post-process the privatized results so that the final synthetic data agrees with what the data curator considers public information. Our contributions are three fold: 1) we show empirically that using post-processing to incorporate public information in contingency tables can lead to sub-optimal inference, 2) we propose an alternative Bayesian sampling framework that directly incorporates both noise due to DP and public information constraints, leading to improved inference, and 3) we demonstrate the proposed methodology on a study of the relationship between mortality rate and race in small areas given priviatized data from the CDC and U.S. Census.



Thanks to Roberto Molinari at Penn State for helpful discussions, John Abowd and Philip Leclerc at the U.S. Census for discussions about their DP methodology, and Alexis Santos at Penn State for providing data. This work was supported in part by NSF Award No. SES-1853209 to The Pennsylvania State University.

Supplementary material


  1. 1.
    Abowd, J., Kifer, D., Moran, B., Ashmead, R., Sexton, W.: Census TopDown Algorithm: Differentially Private Data, Incremental Schemas, and Consistency with Public Knowledge (2019)Google Scholar
  2. 2.
    Abowd, J.M.: The U.S. census bureau adopts differential privacy. In: Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 2867–2867 (2018)Google Scholar
  3. 3.
    Abowd, J.M., Vilhuber, L.: How protective are synthetic data? In: Domingo-Ferrer, J., Saygın, Y. (eds.) PSD 2008. LNCS, vol. 5262, pp. 239–246. Springer, Heidelberg (2008). Scholar
  4. 4.
    Awan, J., Slavković, A.: Differentially private uniformly most powerful tests for binomial data. In: Bengio, S., Wallach, H., Larochelle, H., Grauman, K., Cesa-Bianchi, N., Garnett, R. (eds.) Advances in Neural Information Processing Systems, vol. 31, pp. 4208–4218. Curran Associates, Inc. (2018)Google Scholar
  5. 5.
  6. 6.
    Canonne, C.L., Kamath, G., McMillan, A., Smith, A., Ullman, J.: The structure of optimal private tests for simple hypotheses. In: Proceedings of the 51st Annual ACM SIGACT Symposium on Theory of Computing, pp. 310–321. ACM (2019)Google Scholar
  7. 7.
    Carroll, R.J., Ruppert, D., Stefanski, L.A., Crainiceanu, C.M.: Measurement Error in Nonlinear Models. Chapman and Hall, London (2006)CrossRefGoogle Scholar
  8. 8.
    Center for Disease Control and Prevention: CDC WONDER Online Databases.
  9. 9.
    Dwork, C., Lei, J.: Differential privacy and robust statistics. In: Proceedings of the Forty-First Annual ACM Symposium on Theory of Computing. STOC 2009, pp. 371–380. Association for Computing Machinery, New York (2009).
  10. 10.
    Dwork, C., McSherry, F., Nissim, K., Smith, A.: Calibrating noise to sensitivity in private data analysis. In: Halevi, S., Rabin, T. (eds.) TCC 2006. LNCS, vol. 3876, pp. 265–284. Springer, Heidelberg (2006). Scholar
  11. 11.
    Elfadaly, F.G., Garthwaite, P.H.: Eliciting Dirichlet and Gaussian copula prior distributions for multinomial models. Stat. Comput. 27(2), 449–467 (2016). Scholar
  12. 12.
    Fearnhead, P., Prangle, D.: Constructing summary statistics for approximate Bayesian computation: semi-automatic approximate Bayesian computation. J. R. Stat. Soc. Ser. B (Stat. Methodol.) 74(3), 419–474 (2012)Google Scholar
  13. 13.
    Fukunaga, K., Hostetler, L.D.: The estimation of the gradient of a density function, with applications in pattern recognition. IEEE Trans. Inf. Theor. 21(1), 32–40 (1975)MathSciNetCrossRefGoogle Scholar
  14. 14.
    Fung, B.C.M., Wang, K., Chen, R., Yu, P.S.: Privacy-preserving data publishing: a survey of recent developments. ACM Comput. Surv. 42(4) (2010).
  15. 15.
    Gaboardi, M., Lim, H., Rogers, R., Vadhan, S.: Differentially private chi-squared hypothesis testing: goodness of fit and independence testing. In: Balcan, M.F., Weinberger, K.Q. (eds.) Proceedings of The 33rd International Conference on Machine Learning. Proceedings of Machine Learning Research, vol. 48, pp. 2111–2120. PMLR, New York, 20–22 June 2016Google Scholar
  16. 16.
    Gong, R.: Exact Inference with Approximate Computation for Differentially Private Data via Perturbations (2019).
  17. 17.
    Inusah, S., Kozubowski, T.J.: A discrete analogue of the Laplace distribution. J. Stat. Plan. Infer. 136(3), 1090–1102 (2006)MathSciNetCrossRefGoogle Scholar
  18. 18.
    Karwa, V., Slavković, A.: Inference using noisy degrees: differentially private \(\beta \)-model and synthetic graphs. Ann. Stat. 44(1), 87–112 (2016). Scholar
  19. 19.
    Krieger, N.: Census and differential privacy: public health and health equity questions. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  20. 20.
    Leclerc, P.: The 2020 decennial census TopDownDisclosure limitation algorithm: a report on the current state of the privacy loss-accuracy trade-off. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  21. 21.
    Lekshmi, S., Sebastian, V.S.: A skewed generalized discrete Laplace distribution. Int. J. Math. Stat. Invention 2(3), 95–102 (2014)Google Scholar
  22. 22.
    Marjoram, P., Molitor, J., Plagnol, V., Tavaré, S.: Markov chain Monte Carlo without likelihoods. Proc. Natl. Acad. Sci. U. S. A. 100(26), 15324–15328 (2003)CrossRefGoogle Scholar
  23. 23.
    McClure, D., Reiter, J.P.: Differential privacy and statistical disclosure risk measures: An investigation with binary synthetic data. Trans. Data Priv. 5(3), 535–552 (2012)MathSciNetGoogle Scholar
  24. 24.
    Papadimitriou, C., Steiglitz, K.: Combinatorial Optimization: Algorithms and Complexity, vol. 32, January 1982Google Scholar
  25. 25.
    Salvo, J.J.: Establishing priorities for the privacy budget: the case for good age data. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  26. 26.
    Santos-Lozada, A.R.: Differential privacy and mortality rates in the United States. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  27. 27.
    Sojourner, A., Van Riper, D.: Child poverty by local school district and allocation of federal Title1 funding. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  28. 28.
    Spence, M., Perry, M.: Demographic findings of the 2010 census demonstration product. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  29. 29.
    Uhler, C., Slavković, A., Fienberg, S.E.: Privacy-preserving data sharing for genome-wide Association Studies. J. Priv. Confidentiality 5(1), 137–166 (2013)Google Scholar
  30. 30.
    US Census: US Census Summary File 1 Dataset.
  31. 31.
    Van Riper, D., Spielman, S.: Geographic review of differentially private demonstration data. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  32. 32.
    Vu, D., Slavkovic, A.: Differential privacy for clinical trial data: preliminary evaluations. In: Proceedings of the 2009 IEEE International Conference on Data Mining Workshops, ICDMW 2009, pp. 138–143. IEEE Computer Society, Washington, DC (2009)Google Scholar
  33. 33.
    Wang, Y., Lee, J., Kifer, D.: Revisiting Differentially Private Hypothesis Tests for Categorical Data. ArXiv e-prints, November 2015Google Scholar
  34. 34.
    Wiley, K.: Decennial census, rural housing data and differential privacy. In: Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations (2019)Google Scholar
  35. 35.
    Wilkinson, R.D.: Approximate Bayesian Computation (ABC) gives exact results under the assumption of model error. Stat. Appl. Genet. Mol. Biol. 12(2), 129–141 (2013)MathSciNetCrossRefGoogle Scholar

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.Department of StatisticsPennsylvania State UniversityUniversity ParkUSA

Personalised recommendations