Fig. 1From: Addressing data privacy in matched studies via virtual poolingModels of horizontally and vertically partitioned data as adapted from Karr, et al. (2007). Each row is for one subject, and columns are for covariates. (a) Horizontally partitioned data. Data subjects are partitioned among database owners or nodes. (b) Vertically partitioned data. Covariates are partitioned among nodesBack to article page