Adaptive thresholding estimator for differential association structures in two independent contingency tables

Abstract

In this paper, we consider an adaptive thresholding procedure to estimate the difference of association structures in two independent two-way contingency tables of the same order. Here, we assume that the class of paired association structures have an approximately sparse difference. Under $L_1$ and $L_2$ loss functions, we establish the corresponding risk's upper bounds for our differential association adaptive thresholding estimators. Moreover, we show that these estimators perform well in a simulated setting. In this line, we carry out a simulation study and compare two well-known independent social mobility datasets.

Keywords

• Association orthogonal components
• Intrinsic association parameters
• Adaptive thresholding estimator
• sparsity

References

1. [1] A. Agresti, Categorical Data Analysis, J. Wiley, Hoboken, New Jersey, 2002.
2. [2] F. Bartolucci and A. Forcina, Extended RC association models allowing for order restrictions and marginal modeling, J. Amer. Statist. Associ., 97, 1192-1199, 2002.
3. [3] M.P. Becker and A. Agresti, Maximum likelihood estimation of the RC(M) association model, Appl. Statist., 39, 152-167, 1992.
4. [4] M.P. Becker and C. Clogg, Analysis of sets of two-way contingency tables using association models, J. Amer. Statist. Asso., 84, 142-151, 1989.
5. [5] Y.M. Bishop, S.E. Fienberg, and P.W. Holland, Discrete Multivariate Analysis: Theory and Applications, Springer, 2007.
6. [6] V.V. Buldygin and V.Y. Kozachnko, Subgaussian random variables, Ukrainian Math. J., 32, 483-489, 1980.
7. [7] T.T. Cai and A. Zhang, Inferencial for high-dimensional differential correlation matrices, J. of Multivariate Analysis, 143, 107-126, 2016.
8. [8] T.T. Cai and H.H. Zhou, Optimal rates of convergence for sparse covariance matrix estimation, Ann. Statist., 40(5), 2389-2420 2012.

9. [9] S.K. Ghoreishi and M.R. Meshkani, Bayesian analysis of association (BANOAS) in contingency tables with ordinal and interval variables, J. of Statistical Theory and Applications, 5(2), 363-372, 2006.
10. [10] S.K. Ghoreishi and M.R. Meshkani, Asymptotic Maximum Likelihood and Bayesian Analysis of Shares of Various Weighted Trends in Association Models in Contingency Tables, J. of Statistical Theory and Applications, 7(2), 229-243, 2008.
11. [11] L.O. Goodman, Simple models for the analysis of association in cross-classifications having ordered categories, J. Amer. Statist. Associ., 74, 537-552, 1979.
12. [12] L.O. Goodman, The analysis of cross-classified data having ordered and/or unordered categories: association models, correlation models and asymmetry models for contingency tables with or without missing entries, Ann. statist., 13, 10-69, 1985.
13. [13] L.O. Goodman, Measures, models, and graphical displays in the analysis of crossclassified data, J. Amer. Statist. Assoc., 86, 1085-1111, 1991.
14. [14] M. Kateri, T. Papaioannou, and R. Ahmad, New association models for the analysis of sets of two-way contingency tables, Statistica Applica, 8, 537-551, 1996.
15. [15] M. Kateri, R. Ahmad, and T. Papaioannou, New features in the class of association models, Applied Stochastic models and Data Analysis, 14, 125-136, 1998.
16. [16] A. Rothman, E. Levina, and J. Zhu, Generalized thresholding of large covariance matrices, J. Amer. Statist. Associ., 104, 177-186, 2009.