Identification and Estimation of Graphical Models with Nonignorable Nonresponse

We study the identification and estimation of graphical models with nonignorable nonresponse. An observable variable correlated to nonresponse is added to identify the mean of response for the unidentifiable model. An approach to estimating the marginal mean of response is proposed, based on simulation imputation methods which are introduced for a variety of models including linear, generalized linear, and monotone nonlinear models. The proposed mean estimators are N -consistent, where N is the sample size. Finite sample simulations confirm the effectiveness of the proposed method. Sensitivity analysis for the untestable assumption on our augmented model is also conducted. A real data example is employed to illustrate the use of the proposed methodology.

MATCH BIAS OR NONIGNORABLE NONRESPONSE? IMPROVED IMPUTATION AND ADMINISTRATIVE DATA IN THE CPS ASEC

We test an improved imputation technique, sequential regression multivariate imputation (SRMI), for the Current Population Survey Annual Social and Economic Supplement to address match bias. Furthermore, we augment the model with administrative tax data to test for nonignorable nonresponse. Using data from 2009, 2011, and 2013, we find that the current hot deck imputation used by the Census Bureau produces different distribution statistics, downward for poverty and inequality and upward for median income, relative to the SRMI model-based estimates. Our results suggest that these differences are a result of match bias, not nonignorable nonresponse. Nearly all poverty, median income, and inequality estimates are not significantly different when comparing imputation models with and without administrative data. However, there are clear efficiency gains from using administrative data.